Autodesk CEO, Carl Bass.
Morning, everybody. Morning. It's good to start with a little bit of a highlight reel and just try to remind people of what our customers do and what industries we serve. One of the things we're going to do today is we're going to spend a lot of time on the details of the Autodesk business. We're going to go way down in the weeds. It's going to be like the dentist. You should just raise your hand and say stop, when it's too much. There have been a lot of questions on the details of the financial transition. We're going to spend a lot of time on that.
Before we do, I thought it was important just to show the reel and just to remind people of the industries we serve and where we compete, and how powerful these industries are, how important they are, how they'll be here for generations to come. People are going to keep building the world's infrastructure. They're going to keep manufacturing. They're going to provide media and entertainment for all of us. What's most important is how we serve those customers. Let's just start with looking at the three segments that we serve. First one, as you all know, is Architecture, Engineering, and Construction. Each of these, you can see the difference between what the physical reality is and the software that's used to generate it.
In this case, we now have most of the world are building new buildings and the world's infrastructure and constructing complete digital models of the thing they're going to build before they do it. Almost all of our customers, what they make and get paid for is building physical artifacts. What we do with our software is enable them to take their ideas and turn them into the physical artifacts that they sell that drives their business. The first industry is Architecture, Engineering, and Construction. Historically, we've been very strong in architecture and engineering. Construction has become a much more important part in going forward, particularly as we start talking about market expansion. That's where a lot of market expansion is. This is a gigantic industry, and it's underserved by technology. Second industry is manufacturing. Here, we're looking at everything like automobiles.
This is a recently designed street-legal race car that was designed by a small team in the U.K. It's everything from automobiles to airliners, to consumer products, to medical devices, to industrial machinery. Again, another gigantic part of the world's economy, incredibly important, and we just need to remember the lens of what makes Autodesk solutions compelling to these customers. The third one, which always makes for a great highlight reel, is the Media & Entertainment business. All of the movies, all of the commercials, all of the games that are played use our software to make those. When you step back and try to figure out what do these three industries have in common, and what is driving customers to make purchasing decisions? What drives their thoughts about what they do next in running their business?
I would say there are three needs that go across geographies, across industries. The first one I would loosely call innovation. And sometimes I feel, remember that movie, "The Princess Bride?" In "The Princess Bride," he keeps on talking about inconceivable, and then finally says, "I don't think it means what you think it means." That's how I sometimes feel about innovation. Everyone comes in here and says they want innovation, and then you hear what they really want, and you go, "I don't think it means what you think it means." If you stand back far enough and across these industries, you say, what are they really talking about with innovation? What they want are products and services that differentiate themselves from their customers. They want differentiation. They want sustainable differentiation over time, and they would like the price advantage that comes with it.
Everyone wants the price premium that Apple gets, for example, in designing a phone that's not too dissimilar than other phones, but somehow they get $200 more per phone for it. What innovation really is, it's the antidote to commoditization, and that's what every one of our customers in every industry is fighting. The second thing all of the customers who come in here, and we do hundreds of customer briefings a year, many of them in the gallery here. The second thing they come in and talk to us about is time to market. For those who have those good ideas, have that innovation, what they want is they want to get those ideas to market faster. Take a typical automotive company. Right now, they're competing to become cars that are electric, that are autonomous, that are probably rented instead of owned.
They know where the industry's going. They can't change quickly enough. The idea of even in a pretty static market, from the time they have a new idea to getting a new car out takes three to four years is just too long. Every industry has that same problem. The third one is this reducing complexity, managing risk, managing cost, driving higher quality. In some ways, if you look at this is like Maslow's hierarchy of needs for companies. At the top is clearly innovation and to be different and to be able to charge more. Next is the ability to execute on that.
The third is really the nuts and bolts of how do we execute this and how do you, if you look at that image on the right, how do we manage this world of complexity, whether you're in manufacturing, you have this incredibly complicated supply chain around the world, or you're trying to manage a construction site like that with thousands of employees on it every day. That's the lens at which you have to look at the Autodesk business. These are the industries we serve. We've served them historically, and we will continue to. Now let's just switch over and look through another lens at what we've been doing. If you look over the last couple of years, there are two fundamental things we're doing, and we've talked about it somewhat separately, but they are really intimately tied together.
The first one is evolving the business model, and this is what we're going to talk about. We will spend a lot of time today. Andrew will go into gory detail about this. A lot of the mechanics of doing it, the economics of doing it. It's certainly what a lot of the investment community has been focused on, and rightfully so. We'll spend a lot of time talking about evolving that business model. Then the second thing is we want to talk about building this next generation of design and engineering software. This is basically enabled by the cloud. Remember, what our job is as a technology company is to take those needs of the customers I described and think of what technology is here and meet somewhere in the future, and that's what we've been busy doing.
We've been building cloud platforms because three years from now, the majority of engineering software will be run on the cloud. Five to seven years, all of the engineering software in the world will be run on the cloud. We've been building that for the future. These are the two parts that we've been walking along in tandem, in order to have a great business going forward. Excuse me. Okay, 14 months ago, we all sat in this room together. Many of the same faces I see here. That's what we talked about, was the free cash flow projection for FY 2020 was about $1 billion. During the year, we made some changes. On our February earnings call, we talked about changing that.
We talked about a number of changes we were making to get to $6 of free cash flow per share in FY 2020, and how that would be driven, in a large part, by transforming the current business. We also talked about a further-out-there goal. The reason why we brought out a goal that was much further out is we didn't want people to think of $6 in free cash flow as being some terminal value, is that it was going to hit there and there was nothing left. One of the important things I hope you take away from Investor Day this year is the possibility of market expansion and what that means. It's really important to understand the dynamics of what's going on in manufacturing construction, and how the things we're building will enable this expansion of the market.
What we think is the transformation in the current business gets us to $6-plus in free cash flow in FY 2020, and the cloud business takes us beyond to $11-plus in FY 2023. Let me spend a little bit of time right now on what we've done this year. Going back 14 months and then looking forward and describe the changes we made, and then I'll describe the stuff we're working on going forward. One of the first things that happened during this year is the long-talked-about change from selling perpetual licenses to subscription. Happened at the end of the second quarter. We no longer sell perpetual licenses. Our entire business going forward is a subscription business. It's a term-based business with some elements of consumption layered on top. One of the things I'd say about that transition is it went incredibly well.
We've got an incredible amount of compliments from both our customers and our partners on managing that transition. Second thing we did this year is we launched industry collections. Industry collections are the new analog to what we were doing with suites, but they're also a mechanism for us to consolidate our product portfolio and streamline how we take our technology to market. It's an easier way to sell, it's a compression of the product portfolio, and it gives us much greater effectiveness at selling and marketing these products. The third thing we did this year on the product side is we launched our three big platforms. These are the three big cloud platforms. One is BIM 360, one is Fusion 360, the other is Shotgun for respectively the three big industries.
Going forward, these are the products, if you will, and our technology will fold into one of these three products. These are the big platforms. Everything we do in AEC will be in BIM 360, everything in manufacturing in Fusion, and everything in M&E will be in Shotgun. Let me turn to the programmatic stuff that we did this year. First thing is we ran a number of programs to move the legacy base forward. Last year we spent a little bit of time talking about the size of the legacy base. I just wanted to show you results by comparison, is that in FY 2015 when we did a promo like this in the third quarter, that 11,000, subsequent year we did 13,000. In the first quarter, we did 28,000 of these.
Many of you remember I mentioned that we thought we could drive more success with the program at a lower discount. We lowered the discount, and we were able to drive 43,000 of these sales in the third quarter. We'll talk more about what's going on with the non-subscriber base. Andrew's going to spend a lot of time on it, but this is one of the important things we're doing. Second thing we're doing is looking at the non-paying users. We have the non-subscribers, and then we have the non-paying users, or what we call pirates. After a lot of analytical work that we've done on it, this is what we have right now. We have 1.7 million active non-paying users, and these are in accounts that are paying us now or have paid us in the past. What's the dissonance there?
Are they paying you or are they not paying you? They're partially paying us. They may have 10 seats and they're paying for five, would be somebody who falls into that category. 550,000 are the accounts that have never paid us. We think in accounts that have paid us, these are easier conversations to have with customers. They're more available market to us than I'd say in the other accounts. The other thing we did programmatically, we talked about this a lot last year, we drove our business to be much more direct in the quarter we just reported on. Direct revenues, combination of enterprise and eStore, we're at 29%. Let me just turn from some of the product stuff and the programmatic stuff to some of the financial results.
This is what the new model ARR looks like, and you can see both the subscription count and the new model ARR going according to plan. It's exceeding our plan in many ways. We will talk a little bit more about our ideas about where this heads as we go to FY 2020, as we work our way through the day. The other thing we talked about last year was the spend projection of 5%-6% increase per year. Many of you had strong thoughts on the subject, and thought that during this period of transition that we shouldn't be spending at that rate. One of the things we've done, and that led to the upping of our forecast in February, was we reduced the spend. Going forward, what we did this year is we spent less than we spent last year.
We've committed for next year not to grow expenses, and on the most recent earnings call, we added one more year to that. If you look over the period of time from FY 2016 to FY 2020, we'll be flat, slightly down on a CAGR basis. Controlling expenses through the transition. The other thing that we did is we were very opportunistic in being able to change some of our capital structure and modernize the business structure. As a result of being able to change that, we were able to now have $1.7 billion of cash that was formerly offshore, it's offshore, but in U.S. subsidiaries, and the money is totally already tax affected. It's available for us to use starting a couple of weeks ago.
We've been able to move $1.7 billion back and have it available to us, and already the tax effect has been applied to it. Scott's going to spend a bunch more on this and talk about how we plan to use it, majority of which will be used for share buyback. That should be some new news for some of you. Okay, let's just talk about what we're going to do, and then we'll have people go into a lot of detail about many of these things. What are we going to do this year? First one is continue to drive product subscriptions. This is the most important thing we possibly do is continue to drive the product subscriptions. Second thing is continue to extend the cloud products.
One of the ideas that we decided to do this year, because there's been a little lack of understanding of what our forecasts are for FY 2020 and what the financial model looks like, we thought we'd be very transparent and show you what the model looks like, what we're expecting from each of the various categories and the price bands that we expect to get for each of those product categories. What you'll see is what we expect from the cloud. While we're in the early years of the cloud business, it's really important to get a strong, large foundation right now. Third thing is we will continue to build upon those legacy and anti-piracy programs. That is one of the source of new customers we've been disclosing on the calls. We'll disclose more today. Really three sources of new customers.
It's the legacy customers, it's the people who've been pirating, and it's customers who are new to Autodesk and coming from share shift. Fourth one is we are going to encourage the move from maintenance to subscriptions, and we'll talk about how we're going to do this programmatically over the next few years and what this means for the economics of the business. Steve and Andrew will both talk about higher price realization through optimization in price as well as changes in channel strategy and just the general go-to-market strategy, so you'll hear about both those things. You've heard about the commitment to maintain the flat spend for the next two years. We will continue to do more work on our capital structure, and Scott will detail a little bit more about what we're able to do to enjoy some of the benefits we have today on an ongoing basis.
That brings us back to where do we think we are? If you look at the blue snake, as we think about, that's what we showed last year at Investor Day. What we think, we've done a number of things to change the shape of that. We've been able to get incrementally higher, been able to do it sooner. The trough isn't quite as deep. We have a lot less risk in the plan, and we have much more confidence in our ability to execute given what we've done over the last 14 months. Going forward, this is where we think we get to. Still in excess of $6 in free cash flow per share in FY 2020 and excess of $11 in FY 2023, but with much greater visibility and much greater confidence.
Let me now turn this over to Andrew and let him walk you through the details. I think we refer to it as Ops University. We'll have Andrew walk you through that.
Thank you, Carl. Morning, everybody. Jay, it's always good to see you in the front row. Makes me feel good and welcome. Welcome, fans and naysayers. Favorite fan. Kash, are you out there, my favorite naysayer? No, he's not out. Welcome. It's good to see you, too. I hope what I'm going to do, or what I'm intending to do through this entire presentation, is give you the same level of confidence in this business model transformation that I have. To do that, I'm going to have to go into some details, probably some more details than we've gone into in the past. What I'd like to encourage you is get your iPads up, your laptops up, pencils up, spiral notebooks, whatever you use, class is in session. All right, let's pay attention. The first thing, we're reaffirming exactly what we said the previous year.
We're going after the exact same targets, the same subscriber CAGR, the same ARR CAGR, and yes, the same ARPS CAGR that drives in the middle. That's the first thing that I want you to hear that gives you a lot of confidence. The next thing I want to do is dig a little deeper into how we're going to get there. Let's start that journey with this slide right here. This shows you a few things that we haven't really talked about in detail before. It shows the actual journey from where we started in FY 2016 to where we're going to end up in FY 2020. It breaks out what we think the contribution from cloud's going to be, what we think the contribution from enterprise is going to be, and you probably noticed that I blended product subscription and maintenance together in this journey.
The specific reason I did that, and you'll hear this as we progress through the presentation, is that we have a high level of confidence that by FY 2020, that entire maintenance space is going to be bled down over to subscription, and I'll talk a little bit about that. There's a couple of things I want you to pay attention to here. One, obviously, cloud plays a strong role in the subscriber growth over the term of this business model transformation. It plays a smaller role in the ARR growth. The product and maintenance subs and the EBA subs, the enterprise subs, play a much more significant role. The other piece I want you to pay attention to is obviously we've given you ranges of where we think the terminal ARPS are for this. Obviously, we've given you ranges. Precision is impossible in this kind of modeling.
However, we have a lot of confidence in these ranges. You can see where these ranges are falling out. Cloud's high volume, it touches a lot of people in construction supply chains, in the design supply chains, in manufacturing. It's got the lowest ARPS. You can also see where product subscription ARPS are going to end up. I hope this gives you kind of a sense for how things are going to lay out over the next couple of years. The other piece of this, I get a lot of questions about all of these things. First off, how are you going to hit these ARR targets with the prices you're driving out there, with the ARPS you're currently at? I get a lot of questions about how are you going to hit these subscriber goals? Where are these subscribers going to come from?
How are you going to get these subscribers? A lot of questions about just the overall outcome. What I'm going to do is I'm going to touch on every single one of those areas. I'm going to start with talking about the ARPS evolution and helping you understand where we started, where we're going, and how it's evolving, and what factors are impacting it right now. I'm going to go into more detail than I've gone into in the past about where the subscribers are coming from, and also talk to you about some of the things we've learned over the last year, which I think is very important.
When I was standing up here last year, I said, "We're going to learn a lot as we move into this transition, and it's going to inform us, and it's going to help us understand exactly what we have to do to continue to drive the targets." We've learned a lot. I'm going to expose a lot of that to you. The last thing I want to pay attention to is what I want to call the non-volume ARR drivers, things like pricing optimization and other things that are associated with that. We're going to go into all three of these categories. I think these address all of the areas I get frequent questions about. If they don't, we can talk more later in the Q&A and afterwards. Let's start with the ARPS piece.
Before we get into the details of that, it's probably helpful to remind you how we calculate recurring revenue. Scott's going to go into it again. I'm going to go into it here, but it's important to understand the general equation, and I'll explain why in a minute. Here's how we compute ARR. It's very simply the revenue we recognize, the recurring revenue we recognize in a quarter times four. Now, the important word there is recognize. I'll give you an example so we're all on the same page about how this works. You can pay attention to the inclusions and exclusions. They're just puts and takes, but the important thing is the formula. Because it's driven by recognized recurring revenue in a quarter, let me help you understand how some anomalies make it a lagging indicator in terms of absolute magnitude.
The trends are always correct, the trends are irrefutable, but the magnitude lags the actual magnitude of what's happening in the business. Simple example. We all know that if I sell you a $1,200 subscription for a year, the real ARR is, what, $1,200, right? The way we compute this, it depends on when we sell it to you in the quarter. You sell it in the beginning of the quarter, you recognize all of it. You sell it in the middle of the quarter, you recognize about a third of it. You sell it towards the end of the quarter, you recognize a fraction of it. As a result, what we know terminally in the next quarter will be $3,600 in ARR only shows up as $1,600 in ARR in the quarter that we recognize it.
Of course, the ARPS, which we know is going to be 1,200 in the next quarter, only shows up as 538. I just want you to understand this. I want you to understand the lagging nature of the ARR and ARPS calculation, so you understand the trends are accurate. The magnitudes actually lag where the business is. Okay? Check. We're all on the same page there. Now let's look at the ARPS trend. A few things I want you to take away from this. First, this is exactly what we expected. This is exactly why when we talked to you last year, we said, "Look, the important thing is the subscriber growth and the accumulated ARR that comes out of it. Don't pay attention to the ARPS." Now, I realize that's harder for some of you to do. You want to pay attention to the ARPS.
The first thing I want to say is this is exactly what we expected. The second thing is that half of this trend, is driven by product subscription. The other half is driven by other two factors. The most important one is the enterprise factor. I'm going to tell you about that as well, is that you should be delighted about the trend with enterprise, and I hope you understand why you want to be delighted about that. I'm not going to talk a lot about cloud because let's face it, I think you all understand cloud's a lower price point product. It sells at higher volumes. There's really no need to go into it. It's kind of obvious how it affects these various trends.
What I want to do is spend a little bit more time on what trend did this with product subs, what's trending this with enterprise, and how it's going to evolve over time and why. Again, this is exactly what we expected. First off, let's talk about enterprise. Let me give you a little reminder about how our enterprise agreements work, and Steve's going to give you even more data so you'll understand exactly the mechanics. We go out there, we sell a three-year deal to the customer, and the customer buys a certain amount of consumption in that deal. Our goal, their goal, is to use as much of that consumption as quickly as possible, get them more access to more of the portfolio as quickly as we can, get them using our offerings deeper and deeper. Remember, these customers have access to everything we own.
What this means is that the adoption precedes the renewal, which is the increase in ARR. Very simply, what's happening with the enterprise agreements is there's a 65% average growth in the subscribers, the people using it, but only a 60% growth in the ARR. All of this improves with the renewal event. You want the customers to use more in their existing three-year band, because if they consume their consumption faster, when they renew, they're renewing at a higher consumption level. It's very simple. Adoption drives the subscription growth, renewal drives the ARR growth. You'll get a lot more visibility in the mechanics of this when Steve presents, but this is why the trend is what it is, and this is a good thing. We want the customers using more during the term of the contract so that they renew at higher rates. This trend will continue.
As they renew, ARPS will go up, but as they start to consume within the renewal, it'll cycle down a little bit, then up and down and up and down. This trend will not stop for the whole life of this entire journey. There you go. That's enterprise. Now let's dig a little deeper into product subscription. There's a couple of things in this trend I want to point out, because I want to factor out some things right up front. Two of those big dips right there are directly related to the highly successful non-subscriber promos we had. That dip right there is the Q1 one. That dip right there is the one that we just had in the previous quarter. And notice the dips are smaller from one to the next. That's the effect of moving from a 70% discount to a 50% discount.
We all know about the factors that are affected by this promotion. What I want to do is factor those out for a minute and start looking at what's really happening, because promos are a transient thing. As I talk about where we're going with the promos heading into next year's, the effects of these things are going to fall out pretty rapidly over time. What's really happening? To really understand what's happening, you have to look at like for like kind of comparison. You have to look at a single product family in a single country through a single channel and get a sense for what's happening with new seats. This is what's happening with new seats, and we're seeing this all over the world. There's only a few emerging countries in Europe where we don't see this new seat trend.
The new seat trend is a predictor of the future. The trend you're seeing today is an artifact of the mix shift we've been going through as we transition, and I'm going to help you understand that mix shift. What you see is, generally speaking, the new seat trends are defying the overall product subscription trends. Point one, this is incredibly important because this is the future. This is what predicts where we're going to end up. What's going on? Why is the overall product subscription trend declining? What's driving that? Again, fairly simple factors. There's a set of opposing things that drive the ARPS up or down in a transitionary state, and here they are.
This is the list, I'm going to explore each one of them for you so that you can understand how the transition has affected the ARPS trend-wise and when it's going to settle out. If we sell more in mature countries, we're making more than in emerging countries, 2X as a matter of fact. If we sell more monthly initially than we do annual or multi-year, it's another 2X effect. Selling on the e-store versus indirect has similar patterns of 2X effects on what we realize. What does that mean? These are the things that kind of drive the mix and have an effect. Let's look at them one by one so that you can understand exactly what they've been doing in the past, but what they won't do in the future. Here's the past.
This is the trend from when we started the major parts of this journey. You can see when we began selling subscriptions, it was almost entirely in mature markets. That's the highest ARPS realization we generally have is in the beginning with the mature markets. We were selling mostly LT. You can see as things ramped up and we started rolling out subscriptions, the emerging markets started to have a bigger and bigger effect on the mix. As a result, the general trend from the day we started was down. You also see we have a terminal state in FY 2020. This terminal state isn't a guess. This is based on our historical steady state of the business, where the money's been coming from. We're very close to that steady state between emerging and mature.
You can see as we journeyed here, that is a big driver of depressing ARPS, and we're about to converge into a steady state. One, that trend is washing itself out and it's almost done. Here's the other trend. When we first started selling subscription, we were selling primarily on the e-store. Our channel partners weren't selling much, and we were selling a lot of monthly, a lot of tire kickers, lots of people exploring and testing this thing. Monthly, if somebody pays monthly on a regular basis, it's about 50% higher than an annual contract. We also sell multi-year at discounts today, and that's obviously a different dynamic. You can see initially we were selling tons of monthly, relatively speaking, and over time the monthly has decreased and the annual and multi-year has increased. Another simple factor driving the ARPS down in the short term.
Again, the steady state looks something like this, and we have high confidence in the steady state. This factor is already settling out as well. That's looking backwards. The last factor, which is equally important, is how we started selling on the e-store initially. Remember, the channel wasn't selling a lot of this when we started this journey. We were selling a lot of it on the e-store. As you move from today closer to our steady from Q1 of 2016 to closer to our steady state, you can see that's going to depress the ARPS in the short term. All of these things are driving product subscription ARPS down in the short term, and they're all washing out of the system. All of these things are moving out of the system and moving forward.
This gives you a sense for why the trend for product subscription is what it was. It's real, but it's transient. Why is it going to change? A few things are going to change. I just said these mixed effects are going to diminish. We were highly sensitive to them in the early phases because of where we started, but we're going to be significantly less sensitive to them moving forward. The other really important thing is that we're going to be shifting to higher prices. The promos are going to be much less of an impact moving forward than they were in the past. Remember, this past year, we had 70% and 50% discounts. The biggest discounts you'll see next year are 40% and 30%. After that, you're probably not going to see much discounts at all.
The effects of these promotions and their depressive effect is washing right out of the system. The other thing you're going to hear about from me, and in a bit of detail, is maintenance migration. This is going to have a nice positive effect on lifting the whole entire ARPS up as we move forward. The next thing is the move to collection for higher-priced offerings. That's only just begun. There's a huge set of forces that are driving us up to the higher-priced offerings. The last piece is price realization. Steve's going to talk a lot about that. That's a fix of how much we're going to be selling direct, how much we're going to be dealing with the channel, what we're going to be doing on the e-stores and in the digital direct ecosystem. That's going to drive these things up as well.
All of these things are happening. I want to go back to the beginning of what I said. The trend is real, but we expected it. It is exactly what we expect. Here's what we expect moving forward. This is exactly what we expected heading into this transition, and this is how we expect it to move out. In the second half of FY 2018, we expect the new model ARPS to start drifting up and start heading up again as we move out into the future. We got what we expected in the beginning. We fully expect to get what we were going to expect in the latter half of this. It's totally what we were expecting to see, and we understand what's driving it. Let's just kind of summarize this, make sure we're all on the same page.
Like I said earlier, the trends are real, but they're transitory. If you look at things like for like by individual country, by product, by channel, you see the new seat product subscriptions are heading exactly in the direction you would expect them to head. Several of the mixed factors were affecting the trend. We've gone through them. I can talk to any of you in any more detail about the factors if you want to, but those are all leveling out, and the trend's going to reverse in the second half of FY 2018. We knew it was going to happen this way. We know what's going to happen in the future. That's why we wanted you to pay much more attention on the subscriber growth targets than on the ARPS target in the middle. I hope that helps understand this a bit.
I know it's a lot of information, but it requires information for you to actually get a sense for what's actually happening behind the scenes. All right. Let's move away from ARPS to the next aspect of these, driving volume. Every year we've talked to you about the four places we get volume from. One is the install base, and I'll briefly touch on this because this is just the normal course of business. The other three are converting the non-subscribers, the non-paying customers, otherwise known as pirates. I want to call them all pirates because not everybody knows they're a pirate. Then we also want to talk about the net new customer acquisition. I'm going to give you some visibility into how net new is trending and why it's going to continue to trend the direction it's trending right now.
Let's talk about the install base real quickly. Two factors in the install base that drive things moving forward. One, very simply on the design side, customers buy more seats. They buy them every year, and they're going to be buying more collections. I'm going to show you as we move forward through this, especially as we talk about maintenance, how we're going to encourage more of that install base to move up to collections and move forward the way to subscription. Amar is going to talk a lot about the make side and how BIM 360 is accelerating in construction and how Fusion is accelerating in the manufacturing side and in the lifecycle management side.
We're seeing a lot of increases in adoption there, and I'll actually be able to show you some data that'll help you understand that we're seeing more adoption of the cloud in our current accounts. This is basically the same dynamic we've been talking about year-over-year, and this is what drives installed base growth, and it's going to continue to go, and it's actually accelerating. Now let's talk about the non-subscribers. This is the third year we've been talking about the non-subscriber base. The first year we talked about it, we talked about a non-subscriber base of 2.9 million. Second year we talked about it, we talked about a non-subscriber base of 2.8 million. This year, we're talking about a non-subscriber base of 2.6 million. Remember, this is five years back. This is the five-year back base.
What's different this year than previous years is we have a lot more understanding about this base. We have a lot more analytical insight into what's happening with this base, how active they are, what products they're using. I can now give you specifics about what we're doing, and some of those specifics actually give us closed-loop ability to understand how well our programs are working, how well our targeting is working, and what things we can change to do better. First off, what did we learn? We learned that 51% of that base is active. That's about 1.3 million seats. 51% is active. They are actively using the software on a regular basis. We've also learned that our programs are actively converting them. Notice these are monthly snapshots, June, July, August. We snapshot this on a regular basis. Why is this important?
Not so much that we are able to see that we're having an impact. It gives us that closed loop, that ability to understand you do this, you get this kind of result. The promos tell us one thing. This tells us a lot more. The other piece is we're now sure what releases they're on, how far back they are, and most of them are back-end loaded like we previously thought. You can see that most of the people sit back-end loaded in three to five plus years. This is real data. This is real activity levels. We know what products they're using. Half of them are on LT, 26% of them are on our AutoCAD family, another 22% on the suites. We also have a lot more knowledge on where to focus the programs. What countries, what regions are going to provide the biggest return?
How do we go after this base in a way that gets them engaged with us? As a result of all this knowledge that we've gained over the last year, we have a much better handle on the programs we rolled out, how effective they are, what new programs are going to work, and how we continue to turn the crank in reaching these customers. I told you last year, I said, "We're going to get better at this. We're going to start to understand what's happening, what's not happening, how we can reach these customers." We are. We're getting a lot better. We've got low-touch programs. We've got high-touch programs. We're integrating things with our sales hubs and our digital e-commerce system. We've got lots of ways that we can reach these customers and bring them back. You saw some of the numbers earlier.
I'm going to go into those numbers in a minute, but you can see that we're getting more effective, and we're going to get more effective moving forward. What's interesting is we learned something else. Carl showed you this chart earlier, and there's a couple of things in this chart. One, you can see we're actually getting a lot more effective at reaching non-subscribers than we've ever been in the past. Totally expected. We expected we were going to be able to do that. By the way, remember, the discount went from 70% to 50% here, and we did even better. What drives the customer is not so much the discount. It is the impending knowledge that the discount gets smaller. They already know that this is in the cycle. You're going to see the same kind of virtuous cycle here.
As the clock runs out on discounts, customers tend to move forward. Here's the other thing we learned. A whole bunch of these seats were coming from greater than five years back, and that was a bit of a surprise. We expected a different mix. We expected more to be coming from five years than more than five years back. I've told you what the activity level is for the five-year back level base. If you go more than five years past, there's another 4.6 million valid licenses out there. We have no idea how active they are. We don't have analytics on the activity there. If I sat and just said 20%-- by the way, my 50% last year was fairly accurate. If I just said 20% of them are still active, you've got another 0.9 million out there to reach in the base.
That gives us a total of about 2.2 million active estimated non-subscribers. Base is active. We know how to reach them. Our programs are working. You should have a lot of confidence that we know how to bring these people back into the system. The other thing I want you to have confidence in is why these customers are going to move forward. That's changing as well. The value proposition is evolving significantly. You're going to hear me repeat this slide over and over again, but I want you to understand there's a customer reason they're going to move. One, all of these subscriptions are going to have support from Autodesk integrated in with them in some way moving forward. That alone justifies a lot of the value or the price gap that they're going to be seeing moving to these offerings.
They're also going to be seeing rapidly increasing improvements in user license and update management. These improvements save the customer money. They help the customer manage their relationship with Autodesk more efficiently. They save the customer money. They're actually real value delivered directly to the user. The other value that the customers are really starting to embrace now is, "Hmm, I bought AutoCAD. It's the wrong product for me. I really needed Revit." They don't have a sunk cost. They can immediately switch to Revit. We're starting to see a lot more realization in the base around that basic kind of value proposition. Of course, the integration of the cloud. When you talk about these non-subscribers, the other real benefit is being part of an up-to-date ecosystem. The ecosystem of our products is going to start changing much more rapidly.
It is economically disadvantaged to not be up to date with the current ecosystem. The customers that get on board are going to be up to date with a rapidly changing ecosystem and better participants in the value chain of the market spaces. A lot more data, a lot more insights into the group, a lot more programs, a lot more focus. They're working. We're going to continue to spin the wheel on this and make them better. Now let's talk about the non-paying users, the pirates. When I was up here last year, and Carl kind of alluded to this, I said our piracy rate was maybe like the software average, 43%. Again, we were able to start applying some analytics to this, find out specifics about the pool of active invalid licenses out there, and we found something interesting.
Our piracy rate's more like 70%. This is a big number. Now I'm going to help you try to understand the number a little bit, then I'm going to help you understand how we're targeting it. If you look at how this plays out, we have 12 million non-paying users worldwide, and there's how they break down by charter, 12 million. Three million in the Americas, three million in Europe, six million in APAC. I know you look at this and you go, "Okay, all right. How many of those are in places where you'll just never get them to buy? Or where the rule of law doesn't apply or you have difficulty?" Fair question. 12 million goes down to four million in mature markets. That's more than we can execute on in five years. Four million, you can see exactly how they break down.
One million, 1.5 million, another 1.5 million in APAC. Four million in the mature markets. As Carl alluded to, and I want to give you a little bit more color on, over a million of them are sitting in accounts that are paying us today or have paid us in the past. Just to make sure you understood that point, if you're a subscriber today, you're paying us money. You have five paying seats, and you just so happen to have 10 pirated seats as well. That's a paying customer. If you bought a seat from us in the past, you're now a non-subscriber, but you happen to have two non-subscriber seats and five pirated seats, you've paid us money in the past. We also know about another 150 accounts that have never paid us before.
There's a whole bunch of $2.8 million that are currently unknown, but we'll actually get insights on them over time as well. The big point I want you to take away from this slide, $1.2 million, again, that's more than we can execute on in 3 years. Most of them are sitting in places where we already have some kind of relationship or have had a relationship. This information was pretty powerful, and we started using it and turning it into specific programs. Again, we use the analytics to decide which programs do we want to drive high touch, which programs do we want to do low touch?
These programs, some of them are augmentations of existing ones, like for instance, license compliance programs, reseller engagement programs, some of them are net new programs we've never done before, things we've never actually tried to use to reach this audience before. We're using the analytics to help us do it. To give you some examples, here's a couple of examples. If somebody goes and tries to search on pirate AutoCAD or hacked AutoCAD, they're going to end up back at us now. We're also introducing, as we head into next year, notifications inside the software that say, "By the way, you're using a pirated piece of software." This is going to help us reach the unknowns. The knowns we're going to reach through high touch programs and other types of programs.
These types of tools help us reach the unknowns, they're net new. We've never done anything like this before. They're going to move for the same reasons that everybody else moves, the same reasons I gave previously, but the added reason for the pirate is very simply the lower upfront cost of subscription. This is why we're able to bring them in. The lower upfront cost just makes it attractive for them to get on board. They're going to want the up-to-date ecosystem too, and all the things associated with this, but this is the real driver for a pirate. That gives you a sense for some of the details we have around the piracy base and how we're going to reach out to them.
Like I said, a lot of these programs we've never done before, they're all ramping up as we move forward into the future. The next piece I want to talk about is acquiring new customers. To do that, I want to give you a sense for how we're doing in terms of net new, where the net news are showing up, how some of these trends might have changed from things we said previously. If you look at the total subscriptions, 36% of our total subscriptions are coming from net new right now. These are people that never showed up in our database. I know one of the questions you might have is, are these overlapping with pirates?
As I show you more of the data, I'll give you the answer up front, and then I'll show you some data that supports it. The answer is no, and the reason is very simple. The least pirated product we have out there is LT. It's down in the 1% level of piracy. The most net new product we have is LT. They're not the same customers. They're not the same people. There's very little overlap. The other thing I want you to pay attention to is the total cloud subscriptions. In the past, we've shown you a much larger number than 24%, so hold on for a minute because I'm going to give you a different view. It's 24% coming in from net new. Why is that? We're seeing a lot more of adoption of cloud in our install base, which is a good thing.
We want to see a lot of adoption in the cloud in our install base, and that's exactly what we're seeing. If you look at this from a slightly different view, an account view and not a subscription view, you get an even better story. Of all the subscriptions, 44% are net new. Of cloud, clear half of these customers have never been in our database before. Half. That's the kind of results you want to see if you're expanding the market and expanding the TAM of the company. You're going to hear a lot more about that, but I want you to understand the underlying dynamics here, so that you can understand how we're going to continue to drive this net new number up. First off, what are they buying?
Obviously, it's mostly AutoCAD and LT right now, which actually gives us a great tool for continuing to drive this. The AutoCAD brand's an asset to us. We want them to buy collections, but they all know about AutoCAD. What we do programmatically, especially in AEC, is we go out there and we hit them with AutoCAD. AutoCAD brand, AutoCAD LT. Bring them into subscription. Because they didn't buy anything, we go and then upsell them afterwards. We upsell them to Collections, we upsell them to Revit. We get them in with the AutoCAD brand, we move them up. This is a repeatable, highly efficient machine for us, leveraging that brand and bringing them in. We just keep turning the crank, and then we turn another crank to upsell them over time. We're early in the upsell.
The other machinery we've built that's starting to work very effectively, and Amar is going to give you a lot of color on this, is the Fusion machinery. We're out there reaching customers completely outside of our install base in design with Fusion, and we're bringing them in. The machinery we've built to do that is starting to become just as effective as the machinery we've built for AutoCAD and LT in the AEC base. You'll hear a lot more about that. The other piece of machinery we're building is the Make targets. Using Fusion Lifecycle, Fusion Connect, some of the other Fusion manufacturing pieces to reach down into the rest of the process, as well as with BIM 360. You're going to hear a lot more about that.
The point I want to make here is not only are we bringing in a lot of people from net new, we have the machinery to maintain that run rate of net new moving forward, which is critically important to our growth. That just means we have clear line of sight to the volume drivers. Clear line of sight in a couple of ways. One, we're attaching more cloud to our install base. Exactly what we want to see. It makes each account bigger, and that's what we want, make each account bigger. Analytics is letting us see exactly who's active in the non-subscriber base, exactly what's active in the non-paying base. It's giving us the information we need to target more effectively and also understand how effective the programs we deploy out there are.
The net new subscriber volume is going to continue to be a key contributor. That gives you a little bit more depth and insight into some of the volume drivers and what's going on there and how we're going to maintain the volume. The last piece I want to talk about are the non-volume ARR drivers. These come in three basic categories. Maintenance migration, the AutoCAD family pricing optimization, which I got a lot of questions about, and I'll talk a little bit about and show you some information, and the move from standalone to collections. I'll say quite unabashedly, the most important one that we're going to talk about has the least number of slides, but you're going to get a lot more information about it in the next couple of months as we roll out some of the details with the customers.
The maintenance migration is a critical factor in moving our customers forward. Here's how we're going to do this. We're going to learn from things we've done in the past, things that strike a right balance between customer friendly, customer pressure, and channel readiness to drive this. Let's take a look at what we're going to do. What's the pressure in the system first? You're going to hear shortly from us a set of price increases that we'll be telling our customers about around the maintenance base. I'm not giving you specifics because we're going to roll them out in a couple of months, but I want you to understand the general idea of what we're doing. The customers are going to know there's a series of two price increases that are going to be out there on the maintenance base moving forward.
By the way, these kind of timings where you have an event at the end of a fiscal year moving forward offer us this kind of opportunity for programmatic execution. A programmatic execution we know very well how to run and how to play. What kind of programmatic execution are we going to introduce into this gap, the pressure gap? We're going to introduce a very customer-friendly loyalty price that encourages the customer to move early as rapidly as possible at a price less than the price increase they see in the future, and they will get that advantage price for some period of time. That's the carrot and the stick that helps drive the system. This is a simple program that the partners know how to run, the sales force knows how to run very effectively, and it's going to work very well.
We move into the next year, we introduce another loyalty rate. It's higher than the one that was previously. The ones that moved earlier got a more advantaged price locked in than the ones that moved later, but it's out there. This is going to have another big increase. If they wait, there's no migration discount. They end up with a migration price that's equivalent to the increased maintenance price. These are programs we know how to run. They're identical to the programs we ran when we moved our customers from upgrades to maintenance. They're identical to the programs we ran when we discontinued upgrades and we moved our customers to subscription, and we know how to run them effectively, repeatedly, and successfully. Again, this blends all the right elements.
There's a customer-friendly element, there's a pressure driver in there's a program that the machinery knows how to execute on. Again, why are they going to move? Same reasons, the loyalty pricing is going to be a huge driver. More details will come out of this in the next couple of months, this is going to be one of the most important programs we're going to be running over the next few years. I think it should give you a lot of confidence about how we're going to be driving the ARR in the right direction as we continue to drive the volume in the right direction as well. Let's go into the other piece of the puzzle, the AutoCAD price optimization. This one generated a lot of questions.
The results speak for themselves, I want you to understand exactly why we started down the journey we started down. It actually had a very firm rationale, it's not over yet. It has a second step. The first step was this. Back in the day when we had perpetuals, we had a very well-tuned price gap between LT and AutoCAD and between AutoCAD and the verticals. This price gap had evolved over a lot of time, a lot of iteration, a lot of understanding of dynamics. We knew what price points got the most LT buyers in and also got the most AutoCAD buyers in. Something happened when we rolled out subscriptions. We tried something different. Basically what we did when we rolled out subscription is we dramatically increased the price gap there, out of the range that it had historically been.
Guess what? We saw some interesting dynamics. AutoCAD volume went down, LT volume went up, vertical volume went down as well. Basically what we said is, all right, we need to restore the price dynamics back to the historical averages, drive the volume to AutoCAD that's supposed to be at AutoCAD, drive more volume to verticals, get ready for an even greater simplification of the portfolio. That's exactly what we did. We drove it back to the historical norm. That's all this was. That's all this pricing optimization does. What was the goal? Drive the right mix, drive the right adjustment in channel margins, that you get a result that's beneficial. That's exactly what happened. The mix shifted away from LT.
We saw a quarter-over-quarter increase, because you want to measure this over quarter-over-quarter to understand exactly what the impact of the price increase is. We saw a quarter-over-quarter ARR increase of 8%. By the way, that's against seasonality, just so you know. Seasonality for Q2 to Q3 is always down. We saw an ARPS increase of 23%, and we saw massive volume increases in AutoCAD year-over-year, which is exactly what we wanted to see. It's exactly the result we expected from the optimization. I got lots of concern about this, lots of emails, lots of phone calls, people worried about, "What are you doing here?" Between the price change, the mix optimization, and some of the upfront margin adjustments, we got the results we were expecting to get. We're not done optimizing the AutoCAD family.
I want to talk about one more optimization you're going to see next year that simplifies the portfolio, allows us to execute more rapidly, and actually drives more people up to some higher value offerings. Let's talk about that. Again, here's where we're ending this year, back where we were historically. A couple of little tweaks we're going to be making next year, that actually small tweaks on a large base have interesting implications, but here's what it's going to look like. We're going to increase the price of AutoCAD LT about 5%. AutoCAD is going to drift up exactly the same amount. What we're going to do at the same time is we're going to level the prices of the verticals to the same price.
We're going to have a tier of prices that people can go up to that's higher value, and we're going to have a general 5% shift up in the total base. This is going to have another dramatic effect on mix as more people secure the verticals at the AutoCAD price. Also, it's going to have a dramatic effect on ARR as people continue to renew LT at a slightly higher price. We see no volume implications associated with this. Volume's going to go up. We see no negative volume drivers here. This is the last little price optimization we're going to do around AutoCAD, and I wanted you to understand that. Okay. The last piece here is standalone to collections. I want to give you a sense for where we're at with standalone to collections. We're doing well, but we just started this journey.
Here's where we're at so far today since we rolled it out. We've sold 19,000 collections. There's a couple of important things I want you to pay attention to here that are important moving forward. This is the number of people that have actually switched from, say, an AutoCAD subscription to a collection. It's a small number right now, but it's accelerating and growing. This is the number of new seats, but this number is particularly interesting. This number is, Jay, you took your old AutoCAD licenses. I'm going for that legacy offer, that non-subscriber offer. Instead of going like for like and getting an AutoCAD, I'm going to go for a collection. We just moved these customers, 9,500 of them, to a collection using the non-subscriber promo to get them there.
They get an advantage price for a while, but then they're back up to the full collection price, and they were previously an AutoCAD legacy customer or potentially an AutoCAD maintenance customer that fell off. Things like this are knobs we're going to turn to move these customers. There's another important knob we're going to turn, and this is connected to the maintenance migration I talked about earlier. Just digest this slide. I'll explain to you exactly what this means. Remember I talked earlier about these loyalty pricings and this whole maintenance migration program? It's having a big impact on what an AutoCAD maintenance customer sees in terms of wanting to move to collections. Before, what they saw was a 5X change to move from their AutoCAD maintenance to a collection, say an AEC collection.
As you can imagine, not a lot of people were taking us up on this offer. Okay? Not a surprise. However, with the new maintenance migration program, an AutoCAD customer opting to take one of these loyalty discounts and move over from maintenance subscription has a choice in front of them now. That choice looks much more achievable than the choice previously. It's only a 2X delta between that and the collection. This is another programmatic lever that we can turn in terms of incenting our customers and our channel partners to move not only to maintenance, but move up to a collection maintenance seat, or net collection subscription as well. All right. You're going to see us move that motion just as aggressively as we move the other motions. What that tells you is the non-volume ARR drivers are delivering incremental growth.
They're delivering incremental growth in a couple of ways. The increased focus on maintenance subscription is going to deliver a lot of incremental ARR growth, and the customers are going to engage with this program. Pricing optimization for AutoCAD family has yielded incremental ARR growth. It's going to continue to yield incremental ARR growth, and the move to collections is going to accelerate dramatically as we move into FY 2018 and beyond. All of these are things we have line of sight to, and we understand exactly how to execute on. All right. That was a lot. You guys have been asking a lot of questions, and you've been asking for a lot of detail. I gave you a sense for the ARPS evolution. That was backward-looking. I think the net takeaway I want you to take away from that is the forward-looking view.
We talked about subscriber volume. We talked about the non-volume ARR drivers. Just the net-net. The ARPS trend was real, but it's transitory. Got it. You can ask more questions, but that's the truth, and that's the net-net of all this. We have a clear line of sight to the volume drivers. Analytics has helped us get an even clearer line of sight. You saw some of the data. You understand it. That's the real dynamics of the base right now. The non-ARR volume drivers are delivering a lot of incremental growth. That's why there's so much confidence and reaffirmation of the plan of record. I hope the additional detail has driven some clarification. I'm sure it's also created some questions, but this is what we felt we needed to do to help you understand where we've been and where we're going.
What I want to do is introduce my colleague, Mr. Hanspal, to come and talk to you about how we're expanding the market deeper into design and make. Amar?
Thank you, Andrew. Good morning, everyone.
Andrew did a great job of outlining how we're building a better company by working through the business model and exercising all those mechanics and levers of volume and price optimization. What I'd like to do is cover how we're building a bigger company by driving innovation in the cloud. Over the past few investor day sessions, we've talked about the cloud and how we see it as a new platform. It's not just an add-on, but it's a fundamental new platform, the future platform. I'd actually even say it's the today platform for design and engineering. When we look at the cloud and building technology for the cloud, we're not just thinking of doing a like for like. For example, we have Revit that runs on Windows. We're just going to take Revit and make it run in a browser.
We look at the cloud as a new platform to let us fundamentally reimagine what we can do for the industries that Carl talked about. By approaching this as a redefinition problem, we see two clear opportunities that emerge by building this next generation of design and engineering software for the cloud. One opportunity, which we will call Manufacturing 2.0, is really a competitive share shift and new customers who prefer a cloud-based way of doing design and engineering and as they retool for the digital realities that they all face. The second opportunity is in the world of AEC, where really the C in AEC, the construction companies, adopt technology for the first time in a big way to address their age-old chronic problems of collaboration and visibility into their process.
I'd like to cover these two opportunities with you, and I'd like to begin with manufacturing, because admittedly, I think this is the place you guys have the most skepticism. Let me just remind you what we mean by the word manufacturing. Manufacturing is the industry that builds physical goods, discrete manufacturing, ranging from aerospace to automotive to industrial machinery, building products, consumer products. It's a busy industry. It's a large industry, $13 trillion of GDP, 30,000 products added every month, lots and lots of activity happening in this space. Historically, manufacturing companies are big believers in technology, big believers in automation. As a result of that, there's always been a healthy software market for CAD, CAM, CAE, and PLM software. It's about $13 billion.
In the past few years, it's been dominated by four or five players, mostly large enterprise-oriented companies building complex software. Our role at Autodesk has really been to act as the democratizer of the technology for this industry, driving value, high levels of price and performance. I actually want to take this opportunity to remind you all that one not very well-known fact about Autodesk is that we have the second-largest install base in manufacturing. You all think of us as the AEC company. We have the second-largest install base in manufacturing. We historically challenged and driven high levels of price performance through our software stack. At a surface level, this group of players may look relatively stable. The industry, the CAD/CAM software industry grows in single digits for manufacturing software.
Underneath the surface, there's actually a massive shift taking place in terms of preference as companies retool for the digital future. As Carl said earlier, manufacturing companies are always looking to drive competitive advantage through product quality, through faster time to market, through all the things that deliver differentiation in their products. Recently, the big investment that manufacturing companies have made more aggressively is to go all digital in all aspects of their company, not just the factory, but new methods of fabrication, connect all of the players involved in the process, and drive entirely new capabilities, such as radical supply collaboration and customer collaboration, mass customization, new methods of fabrication such as additive manufacturing or new materials such as composites.
There's a whole bunch of investment happening to digitize and connect all of the pieces that drive product development and product manufacturing to drive that competitive advantage. The challenge for manufacturing companies that are trying to be an all-digital company is that in their product development process, the tools that they use are really antiquated. Most of the tools that they use, the CAD/CAM/CAE tools, were built in the late '80s and early '90s. The challenge they have as they try to connect and digitize or digitize and connect everything, is that these tools are hard to learn, hard to deploy, they're very specialized, they don't lend themselves to collaboration, and they certainly don't lend themselves to integration. If you're a company that's trying to drive fast time to market or customer collaboration or supplier collaboration, these are huge barriers.
I mean, the analog I would use is if you're an e-commerce company, this would be like trying to drive a digital online business with IT tools that were built before Y2K. More aptly, in the world of manufacturing, it's like trying to design the Model S with tools that were used to design the Model T back in the '80s. Company after company is looking to retool or figure out the next generation of tools that can drive the evolution of their product development and manufacturing process. Many look to the cloud as the right platform upon which to reimagine their product development and manufacturing process. Why? The cloud provides inherent capabilities that address many of those needs. The cloud makes collaboration brain-dead simple. It makes deployment and ease of use a natural outcome of using cloud-based tools.
It changes the economics of things like simulation that are becoming a requirement for better engineering if you're trying to do all those kinds of advanced processes. It also lets you connect all the way down to your means of manufacturing, which is a huge deal when you start thinking of how additive manufacturing really needs to work, because this whole idea of designing for manufacturability is moving all the way to the front of the process. These are the inherent advantages the cloud provides, and we're seeing many customers recognize that the cloud is the right next platform for product design and engineering and manufacturing for their industries. We're building Fusion 360 to address these needs. Fusion, first of all, of course, it's cloud-based, but it is a no-compromises, complete solution.
One of the challenges people have when they imagine cloud solutions, they think it is too simple to do the kind of complex work manufacturing companies do. Fusion has a no-compromises set of capabilities, and I'll show you in a minute that it can be used to design everything from consumer products all the way to automotive, a full automotive car or automotive design in Fusion. Fusion has capabilities that let you start at conceptual sketching, go all the way to production and beyond. Fusion, just to show it to you in action, is a platform that is capable of holding an entire product definition. It delivers something that looks so close to reality that you might think that you're looking at a photograph, when you're actually looking at a piece of software.
It's connected, anyone, anywhere can use it to view, to edit, to mark up, to participate in that collaboration in near real time. As you can see, this is a design of a real supercar by a company called BAC Mono, that's using advanced styling and industrial design, advanced mechanical design, electrical design, doing CAM, additive manufacturing, data collaboration, data management, all in a single platform, all in Fusion 360. If you start asking, "What is Fusion? Is Fusion a CAD product? Is it a CAM product? Is it a simulation product? Is it a PLM product?" The answer is it's not one of those things, it is all of those things. It is a complete solution that replaces all of those fragmented, siloed, specialized tools that you saw.
At the heart of what it does is it delivers this integrated experience, which has never been possible before, but is possible now because the cloud provides the platform, not just for collaboration, but for integration. We're seeing a lot of success with Fusion. To begin with, we're seeing Fusion adopted by large enterprise companies. Many enterprise companies have teams within their organizations that are specifically tasked to look at new ways to design going forward. We're certainly seeing those teams turn to Fusion. I'm going to highlight one of those, which is Boston Scientific. If you're familiar with Boston Scientific, they make a lot of medical devices, and the team inside Boston Scientific was chartered to try and figure out a better way to address the challenges they've seen over the years.
One of the challenges they've seen over the years is that their medical devices, in terms of bringing them to market, require lots of collaboration with surgeons, with hospitals, and they go through many iterations and many revisions to get these devices to market. Their existing tool set was just too hard for them to work this way, and it just took way too long for them to get those products to market. They switched to Fusion, what they've seen is a 90% gain in their product development cycle. 90% is a huge win for somebody like Boston Scientific. We're certainly seeing large companies, the who's who of consumer products and medical devices switch, and packaging, industrial machinery, automotive. We're seeing the who's who look at Fusion and start to adopt it or switch to it.
The one interesting thing about all these logos up on screen is to sort of underline the point Andrew made and the point I was trying to start this conversation with, is that Fusion is fundamentally expanding our opportunity by generating a share shift. Guess what? A majority of these customers are SOLIDWORKS customers that are adopting or switching to Fusion because they believe in the cloud, because they believe in an integrated process, because they believe in this ease of use, and this next-generation approach to product design and engineering. It's not just SOLIDWORKS customers. In the very chairs that you're sitting in, over the last six months, we've hosted the 12 largest automotive companies around the world, over 250 executives. This is their senior teams coming to Autodesk to listen to what we're doing in the next generation of product design and manufacturing software.
In fact, just last week, two of Germany's finest were sitting just where you are. We're seeing this idea of the next-generation platform really resonate with the industry. Established companies are turning to Fusion. We certainly are seeing the next-generation users and the next-generation companies adopt Fusion. One thing that may not be visible from the outside is that the number of hardware startups has simply exploded. There's been an estimate made by O'Reilly & Associates, that we're seeing about 150,000 new companies start every year. Okay? We have not seen this kind of explosion in terms of product engineering companies in a long time. All these companies are looking for software tools for the first time, and they're turning to Fusion. We are seeing huge take-up in startups with Fusion.
The other group that we're seeing a huge take-up with Fusion is students who are joining the workforce who are bringing new approaches and new tools. Because guess what? People who are joining companies today don't want to use tools that were built before they were born. Okay? This is the generation that has brought Google Docs and Slack to the workplace. In the world of manufacturing, they're starting to bring Fusion into the workplace. The net-net combination of the enterprises adopting, the startups adopting, and the students joining the workforce is we're seeing an accelerating rate of adoption of Fusion. That is our monthly active use count. If you want to understand how fast it's growing, our active users are doubling every six to eight months.
Actually, I should say eight to six months because it started at nine, headed to eight, and is now heading to six. The rate of adoption of Fusion is accelerating, and we believe we are on the cusp, and we actually have already started disrupting the status quo in manufacturing with this next-generation platform for the cloud. I just want to take this moment to remind you that the history of Autodesk and the power of our earnings has always been based on our ability to create categories through disruption. We created this category of CAD on a PC way back in the '80s, and it generated a great run for us. Building information modeling was another category we created, as was mid-range modeling, which was what Inventor was a part of.
Fusion is that same category creation in terms of the next-generation cloud-based product design, engineering, and manufacturing. We're already seeing huge levels of adoption of Fusion, customers moving there. We really believe this is the next wave of adoption of a platform for product design and engineering, and we're in pole position to win this particular opportunity. That was one of the opportunities that the cloud enables that we are taking advantage of. The second one is in construction. While manufacturing was really a competitive share shift and a product preference story, construction is all about new users, new customers that are adopting technology for the first time. Just to remind you what we mean by the construction industry or really the AEC industry, this is the group that builds everything from stadiums to skyscrapers, to oil and gas platforms, to infrastructure like roads and bridges.
It's a very large industry. It's about 5% of the world's GDP, about $10 trillion of economic activity every year. Some estimates tell us that in the next 30 years, we're going to have to build 1,000 buildings a day just to keep up with population growth and the level of urbanization that's going to take place around the world. The challenge with this industry has been it's historically been a lower spender in terms of IT technology, really because of the bespoke and disaggregated nature of the industry. They spend about $10 billion, mainly on design and data management software, driven mostly by architects, engineers, and some general contractors. They have been the investors in technology. Because of their lower level of investment in automation, you can see that the industry has suffered from chronic cost and schedule overruns.
As the cloud and mobile technologies have appeared, that has given them a platform to start automating many of the things behind these chronic cost and schedule overruns. The Boston Consulting Group estimates that if construction companies digitize to the level that they can, it would result in an annual savings for the industry of $1.2 trillion, or about 12% of that economic activity. We'd love for all of that money to come to software, but if you assume a certain percentage of that comes to the category of software we develop, we believe our addressable market expands by $6 billion over the next five years easily. We'll see how fast the industry automates and grows its IT spend. The whole move from analog to digital is underway.
As you can see, the picture on the left is really the current state of technology, if you will, out in the job site. What the industry's really trying to use as the cornerstone of moving to digital is building information modeling. I know you all know what BIM is, but just to remind you, BIM is the digital representation of the physical and functional characteristics of what is being built. BIM is to the AEC industry or to projects, what CRM is to go-to-market. It's I think a fundamental building block of what you try to do when you start a project. It's used to understand the what is being built and the how it's being built. At a high level and at a detail level, driving everything from retrofit to design, to engineering, to fabrication.
It's being adopted by many disciplines, all of the disciplines in the AEC industry. Around the world, BIM continues to be adopted because of its huge return of investment. McGraw Hill estimates that it's at least 25%, could be higher in many cases, depending on the project and the level of sophistication. The story I really want to emphasize today is that BIM's moving to the next level on the cloud and mobile platform. We, I prefer to use the shorthand of connected BIM, which is taking BIM all the way out into the field and connecting the last mile to the first mile. The cloud mobile is a perfect platform for the AEC industry, which is distributed, disaggregated, and so much of the work happens in the field.
The cloud gives all of those participants a platform that gives them collaboration, coordination, and visibility, which is a massive win for all of those participants. We've been building a platform called BIM 360 to do this connected BIM experience, to connect the first mile to the last mile. BIM 360, at its heart, has a common data environment, which holds together all of the models, all of the documents, all of the drawings, all of the photos, all of the RFI submittals, all of the work that needs to be remembered and shared for all of the participants on the project and make it available to people anywhere at any time.
In addition to that data environment, it provides a series of applications that can be used by contractors and subcontractors to do things like detect issues before they happen on the job site, to do inspections, to do equipment handover, to do all of the kind of scheduling and planning that they need to be successful. It does it all on a tablet or a smartphone or a web browser. It's really easy to use and deploy. It's been a huge success since we brought it out three or four years ago with the large construction companies, the who's who, again, of the industry. I want to actually highlight the company all the way to the right, Austin Commercial, because it's not one of the largest general contractors. It's more of a mid-tier construction company.
Why I want to highlight is the way they adopted BIM 360 was they started with a pilot on one project in one office. They drove one process, which was quality. As they saw the return on investment, they've upped their commitment to go to all projects, all offices, all trades, and many processes that are being run through BIM 360. BIM 360 is being used by at least 100 of the ENR 400 and many other of the ENR 400 are in pilot or early adoption phases with BIM 360. We're seeing a lot of take-up. The take-up has not just been in the large enterprises. In fact, we've seen BIM 360 move to mid-market. Our channel has engaged in driving BIM 360, we're getting to a larger and larger pool of companies.
As you know. The blessing and curse of the AEC industry has been its disaggregation. Well, the blessing of that is there's a network effect that the AEC industry has, is once the adoption begins of a particular technology, it kind of rifles through the whole supply chain, the value chain. We're starting to see those kinds of effects show up in BIM 360. It remains one of the fastest-growing products in Autodesk history. Here is another place we're confident that given our reputation for the industry and brand awareness, the adoption of BIM, and our bet on the cloud, we're going to monetize the opportunity in construction and make it a substantial business going forward. I just want to summarize these two things that the cloud has enabled in terms of expanding our opportunity.
There's two clear opportunities that the cloud brings while we continue to optimize our existing business, as Andrew described. We have an opportunity in manufacturing through competitive share shift and new companies that are preferring a cloud-based platform. We have a brand new opportunity in construction with companies that are using the cloud to automate their processes for the first time. Our investment in cloud, all the product innovation work we're doing, all the IP that we're reusing to drive these opportunities forward really sets us up to win. As Andrew pointed out at the beginning of his talk, it may not be the bigger part of the bars as you look out to FY 2020, but as Carl pointed out, as you start going beyond FY 2020, the cloud continues to be the biggest growing component of our business.
We're building momentum and we feel really confident about how these opportunities are playing out in front of us. I think now we go to a break. I hope you've enjoyed this first session and look forward to talking to you more about these. We come back in 20 minutes and start with Steve. Thank you.
Ladies and gentlemen, please welcome Senior Vice President, Worldwide Sales and Services, Steve Blum.
Welcome back. I know there's still a lot of folks in the back. I hope you can hear me at least, or you can give Andrew a break and come and take your seat again. I hope you did have a good break. I know there were a lot of follow-up questions after the first set of sessions this morning. I'm very pleased to be here to give you an update on our go-to-market. Our go-to-market strategy is all about playing to our strengths. It's about taking the things that have been working well and using those as a foundation, and then layering on more ways to connect directly with customers so that we can drive more demand, and ultimately meet and exceed our subscription and ARR goals.
I'm really pleased with the progress we've made in preparing ourselves and now executing in a subscription and consumption world. Today, I want to give you updates on three key areas that I know are of interest of yours. They're things I've talked about in the past, and I want to make sure that we at least let you know where we are heading with each of them. I'm going to give you an update on how we're doing with our named accounts, especially moving to our enterprise business agreements, as was referenced earlier by both Carl and Andrew. I'm going to give you an update on how we're continuing to build out our digital direct capabilities.
Then I'm going to give you an update on our channel strategy and give you some insights and some numbers about how we're working with our channel partners that maybe you haven't had in the past. All right. Let's start with named accounts. As we've discussed before in last year's session, in fact, the last couple of sessions, we're focusing on moving our named accounts to enterprise business agreements. Now, EBA is a simple term. I just want to make sure that we're all on the same page relative to what is an enterprise business agreement. It's a consumption-based model based upon tokens. You'll hear us refer to it Token Flex or T-Flex. That's the internal name that we use overall. It gives enterprises access to the entire Autodesk portfolio, and it's a consumption-based model.
The more users that sign up, they can access the entire portfolio. What's critical about the success, though, is that we also dedicate customer success managers to each of our named accounts in an enterprise business agreement. Those customer success managers build joint customer success plans together with the customers. They identify the key initiatives, the big projects, the adoption plans to drive Effective use of the portfolio across the entire organization. Then what we do is based upon those plans, we'll dedicate resources, consulting resources or support resources, to help our customers execute on those plans. We jointly own those plans together. Now, one of the greatest benefits to both the customer and Autodesk is that we're developing strategic relationships with our customers. We're moving away from transactional relationships, and we're focusing on their key priorities, their key business initiatives.
We hold numerous joint planning sessions together with our customers. In fact, we hold specific future of making things workshops with our named accounts, where we identify the disruptions that are impacting their businesses the greatest, and then we build plans with them to address those disruptions to gain a competitive advantage in their marketplace. Perhaps to go after some new opportunities that are critical to their long-term success. They get a lot of value out of the engagement and the change in the strategic nature of our conversations. Of course, we're also getting some great benefits as well when we move customers to our EBAs.
What I'm showing you here is that we took a look at all of our customers and the named accounts that have moved to our T-Flex consumption model, our EBA, and we took a look at the number of subscriptions they had prior to making the move to the EBA, and then the number of subscriptions they have in the EBA. What we found is for every subscription they had before moving, they now have 2.8 times the number of subscriptions in the enterprise business agreement. We see the enterprise business agreement as a key way of driving continuous growth in subscriptions from our named accounts. When we build those joint plans, those customer success plans, we can identify specific needs or opportunities or areas of focus the customers want, and then we get great results out of that, both for the customer and for Autodesk.
This is a very specific example. I'm giving you an actual customer example, one of our named accounts that was implementing BIM 360, but they wanted help in driving expansion and use of BIM 360 across multiple projects around the world. What I'm showing you is we started this project at this baseline, and then we actually dedicated a BIM 360 adoption specialist to the account who worked on execution of this plan together with the customer. Over a 17-month window, we turned the subscription base into 3.7 times the size now actually being implemented on all of their new projects. This is the type of scale we can get with our new offerings, our cloud-based offerings, by having dedicated specialists in a CSM role working together with our named accounts. We also get a great lift in ARR.
What I'm showing you here is we took a look at all of our customers, all of our named accounts that moved to an enterprise business agreement for the first time between the start of FY 2016 and through the end of last quarter. What we did is we took a look. Most of these contracts are three-year contracts. We took a look at the total spend we had gotten from those customers for the three prior years before signing up for the enterprise business agreement. Keep in mind, that's three years back, so it includes perpetual license sales, maintenance, anything that they've been purchasing from us. We compared that total value to the total contract value that was committed in the enterprise business agreements, and the total value was 29% higher in the enterprise business agreement.
Now ARR growth was larger than that because everything in the enterprise business agreement column is contributing to ARR growth, but there were elements, large elements in the prior spend that were non-ARR specific, such as perpetual license sales. We're also seeing enterprise business agreements giving us great lift in overall ARR. I've given you this update. I wanted to give you the latest view of how are we doing as far as converting our named accounts to enterprise business agreements. We're about a quarter of the way through the process, projection-wise, as we end this fiscal year. We have still our Q4 to complete. While we've made great progress, I'm very pleased with it, the opportunity is still very significant to move all of our named accounts over to enterprise business agreements, and that's my personal goal for Autodesk.
Let me talk about digital direct. Andrew mentioned it a little bit. We've been talking a bit about it. First, what is digital direct? Digital direct is all about how we can connect directly to end customers leveraging technology. For us, it's a combination of technology, our autodesk.com site, as well as the means of reaching more people through our hub-based sales and the seamless interconnections that we have between the eStore and the hub from the eyes of a customer who's looking to engage with Autodesk for many different reasons. We have many ways of connecting digitally and directly with end customers, whether it's on autodesk.com or through its in-product capabilities, or by chatting online from autodesk.com with someone in the hub, or even opting to make a phone call and reaching someone in the hub.
We've been building out our capabilities to address larger capacity coverage. From an eStore or from an autodesk.com perspective, we've built out more robust capabilities. Not only can we take transactions online, but we can initiate trials. There's webinars online. There's online chat, as I mentioned. There's a knowledge network that provides easy access to support. In our hub, we talk about inside sales hubs, but the term sales is really not representing that the hub is a multidisciplinary team. It's a co-location of sales resources, technical resources, Customer success resources, support resources, even marketing resources, all co-located in one particular location. This gives us a lot more coverage and a lot more scale with greater efficiencies in communicating directly with customers.
We're also able to take on more of the responsibility of that interaction, and we're getting better results because of us taking some of these things on ourselves. Here's a particular example of a pilot that we ran last quarter. In Q3, we made a change in our process. During the quarter, if a customer who was on autodesk.com went to the 800 number and called for any information, as opposed to going to a telemarketing firm, which is what we had been doing in the past, it now actually went to someone in our inside hub. That person, a sales development rep, quickly determined whether the call was a sales-related call or a support-related call. If it was support-related, we routed the person to the appropriate support specialist that's in the hub.
If it was a sales opportunity, the SDR rep continued to engage in the process and moved the transaction opportunity as close, if not to, the endpoint of closure. As a result of doing this, we closed three times as many transactions. For every one transaction we got in our old process, we were getting three times the number of transactions by actually owning this part of the process ourselves in the hub. What's really cool is that we accelerated the time to closure. 55% of all the transactions we closed happened on the same day the initial call had occurred, where before it was taking multiple days, and may have even gotten lost in the process. A great benefit to us, but also a much better customer experience.
As far as digital direct from a transaction perspective, we are seeing accelerated growth in the number of transactions that we're having on autodesk.com. What I'm showing you here, just to be very clear, is a four-year view of the actual results from Q1 through Q3 of each of those years. This large bar that has significant growth over the last bar is what our actual results were on our direct transactions on our eStores through the first three quarters of this year. We're getting significant growth and acceleration in growth of transactions online, which is contributing to the amount of business we're doing direct, as you saw in our results from Q3. That's really exciting, but what is even more exciting to me is that our eStore has become the single best channel for new customer acquisition, for new logos.
What I'm showing is actual data from the third quarter, from the quarter we just completed. We took a look at all of the transactions that were done through all channels with new customers. Net new, as Andrew was saying, but they're new logos. Brand-new customers, haven't purchased from Autodesk before. Over half of those customers were acquired through our eStore. The eStore is giving us big pickup in subscription growth and ARR growth, but it also has become our single best channel for new customer acquisition. Let me give you an update now on our channel strategy and give you some specific information I think will be enlightening to you overall.
We talk about a lot of things that we do with our partners, but our partners actually provide tremendous value, and we haven't given you an articulation of what value are we getting for having channel partners in the first place. I'm going to give you some specifics. Each and every day, Autodesk has 11,500 feet on the street through our partners that are representing, that are selling, that are supporting our solutions to end customers. We get those 11,500 people through 1,600 partners around the world. We haven't given the feet-on-the-street number before. We have talked about the number of partners in the past, and if you have been keeping track at home, that was a much larger number. For the last few years, I've talked about the fact that there's partner consolidation happening. It's a good thing.
We're promoting that to happen, the number of partners globally is down, but the number of feet on the street is up, and that's a really important thing to take away. While we're going through fewer partners around the world, we have more feet on the street representing us each and every day. That's important because they are providing us scale and leverage in 170 countries around the world, taking transactions in local currencies and providing the Autodesk face to customers around the world. They're also driving value for us, not just in the representation overall. They drive demand, as an example. In fact, the single best thing that our partners do from a demand generation point of view is hosting events. They host about 10,000 of these events around the world each and every year.
We talked about the changes in MDF and how they're hitting our ARR. Some of you had questions about what is MDF. MDF is market development funds. A big chunk of those market development funds are being used to hold these events that are driving great new demand generation for all of us. On top of that, our partners are fielding approximately 2 million customer support and technical calls on Autodesk's behalf. I don't have to actually have people answering those phones. Our partners are the ones addressing all those issues on behalf of the customers. There's another type of partner I wanted to hit on, because mostly what I've been talking about are our VARs who work with our VADs. We also have application development partners through our Autodesk Developer Network.
We have over 20,000 programmers and members adding more value on top of our platform through our ADN program. We have a growing list of partners that are developing value on top of our cloud offerings. We're up to 3,800 partners who are actually adding value on top of our cloud offerings. From a partner incentive perspective, I know some of you talk with our partners, you heard about changes we're continuously making overall. What I wanted to give you is some insights into where we're incenting our partners. The focus is basically on driving subscription growth and ARR. Should be no surprise, those are our most important metrics.
As an example, we put more incentive on closing a new subscription as opposed to closing a renewal, since it takes more energy and more work to close a new customer than to renew an existing customer. We also do have incentives on net new subscriptions. When we use the term net new, that's total number of subscriptions. It's the sum of all the new subscriptions closed and then all the renewals that happened as well. This is all about not just focusing on new, but minimizing churn and driving renewal rates up, and we have incentives on that. We also have incentives on closing large deals through a deal registration program because we want to reward the partners that are out there hunting and developing new opportunities each and every day. We do still have a tiering.
We have platinum and gold partners, and that's based upon the volume of business that partners are closing, the number of subscriptions that they're closing in a particular year, NPS or customer success scores. There's a series of things that we use for tiering our partners, and incentives are aligned based upon that partner tiering. We also have specialized partners that have dedicated specialist sales folks or technical folks in particular areas that we need them to invest in, and we incent them for that as well. The consolidation in the channel will continue. One of the ways that we drive that is by continuing to raise the VAR minimum requirements, raising the minimum purchase requirements, raising the number of dedicated sales and technical people, number of subscriptions required to close. We're going to continue to move that model up.
One of the other things that we're counting on our partners to do, it's important to them, it's important to us, and it's certainly important to our customers, is we're focusing them on delivering services, especially to the SMBs in our territories. We used to actually deliver services direct to SMBs, we being Autodesk, and I stopped doing that. We divested from delivery of services to the SMBs in our territories. We're focusing our resources from my global services organization in the named account space, like I just showed you. We're counting on our partners to build larger consulting practices because our customers need the help in implementing the solution and driving adoption overall. We're investing in partner enablement tools and people and processes to teach our partners how to take the services that we've developed and apply them more broadly.
We're focusing those predominantly in the area of services around our cloud offerings, BIM 360 and Fusion 360, as Amar talked about earlier, or in driving customer adoption and success in general overall. Our partners are also really excited about adding more value on top of our platform, and they're focusing on layering development through Forge. Even in the named account space, some of our named accounts are very large global entities, and they have locations all around the world. Sometimes in places I don't have local resources, but partners may. In those instances, if there's service requirements, we even team up with the partners to help drive that local delivery through the partner.
We've created a customer engagement model that's suited to meet the needs and the preferences of all of our customers, whether they're named accounts, our largest customers that are looking for strategic relationships and engaging with us in a consumption model, whether in the SMB space or the VSB space, if they want to connect with us digitally, or if they want to work through any of our partners in 170 countries around the world. We are truly meeting the needs of our customers and providing them choice to match up their preferences in an engagement model directly with us. There's one more thing I wanted to give you an update on because I know it's something very important to all of you, which is what's the projection for the direct versus indirect business? I gave you an update on this last year.
Two things to take away, or a few things to take away from this chart. The pie represents the size of the net billing. I'm showing you FY 2016 on the left and FY 2020 on the right. I've chosen FY 2016 as opposed to FY 2020 because billings are down in FY 2017 as we're working through the transition. It's a better metric to compare to. You can see the billings, of course, are much larger in FY 2020. Scott's going to give you more details on that next overall. You're going to notice the amount indirect is growing at an accelerated rate. It's growing at a very rapid pace overall and becoming a much bigger part of the overall pie in FY 2020 from where it was in FY 2016.
You've seen us progressing along that path here as we've reported out on the results even here in FY 2017. There is an important takeaway, though, and it's something I want to make sure I reinforce. The actual size of the indirect is also larger. It's not growing at the same rate by any means, but it is larger in FY 2020 than it is in FY 2016. The absolute dollars in FY 2020 are larger than they are in FY 2016. Our partners continue to be a very critical asset to our success. What we'll see, though, in FY 2020 is the number of partners will be lower than what I just shared with you they are today.
If we continue down the same path along, the number of feet on the street, though, representing us will be larger, which is what it's going to take to close that much more business. The pie is going to be split amongst fewer partners who are making more money on what they're currently selling, and that's going to enable them to invest in the skills and the capabilities and the capacity to drive that type of growth. To summarize, we're very well positioned to meet our goals for FY 2020. We have strategic direct relationships with our named accounts. We continue to build out digital direct capabilities that are creating new opportunities, giving us new scale and coverage, and providing preferences to customers in how they interact directly with Autodesk.
We continue to have a very strong partner network that's giving us local relationships and representation in 170 countries around the world and giving us scale and reach that we couldn't do on our own. As you can see, we're providing an integrated, customer-friendly go-to-market experience that meets the needs of all of our customers around the world. I have a phrase internally, which I haven't used with all of you. I always end my comments internally with saying, "I expect to win." I absolutely expect to win through FY 2020 by executing on our go-to-market strategies. On that, I thank you, and it is my pleasure to introduce to you Scott Herren.
Expect to win.
Expect to win.
Love hearing that. It's Steve's tagline in every email that we get from him. I'm going to take you home, get through a handful of slides, more of an update from the time we sat in this room a little more than a year ago today. We'll bring back up Carl for some closing comments and then go right into Q&A. As I'm sure with the number of data points that we've given you today, there may be a couple of questions in your mind. Housekeeping first. We have talked earlier today with some forward-looking statements. I'll continue to give you a few forward-looking statements. Those are made subject to the provisions of the Safe Harbor Act, and the outlook has not changed from the guidance we gave you a week ago today. Including, by the way, the net sub adds increase.
We brought the bottom end of that range up. Net sub adds still in the 515-525 range. Start with the punchline. The transition is on track. You've seen from Andrew a lot of the knobs and dials we've been turning. I'm also going to talk about ARPS because I understand the concern. I understand the math, the data points we're giving you and the math that's being done, what that trend looks like. I'll peel that back again because as I've answered a lot of these questions over the last several months, it's clear to me that a lot of the questions really tie to ARR versus ARPS. I'll peel that back again. Amar talked about the future opportunity that we have in moving to the cloud and some of the success we're having in moving to the cloud.
Steve, about where we're going from a channel strategy standpoint, how we continue to increase our direct sales. Of course, the eStore piece of that has a much higher price realization for us. It doesn't have a channel discount on top. The net is the transition is on track to the goals that we laid out to a much more highly recurring and predictable revenue stream and cash flow stream out in time. Versus a year ago when we stood up here, we have now transitioned our way through the end-of-sale of perpetual licenses in two phases. First on individual products and now on suites. Those are some pretty difficult-to-predict events, both the quarter that they happened and the following quarter. I think you've seen us execute quite nicely through both of those.
You've also seen us increase our focus on spending, and that was really based on a lot of feedback we got after this meeting in September of last year that says, "Hey, as you're going through this trough, I understand the need to invest, but you also need to be prudent in the way that you manage spend through this trough." We committed to be flat this year, which we've now we're actually going to be down about two percentage points in total spend. We've committed to be flat again in fiscal '18, and recently on our call a week ago, on our earnings call, have now extended that to being flat again in fiscal '19. I'll talk a little bit about how we're doing that. That has increased our confidence by building a buffer between our internal plans and some of the external commitments that we've made.
The fiscal '20 goals, 24,320 that we talked about a year ago, the $1.4 billion of free cash flow, which if you do the math, turns into a little bit above $6 of free cash flow per share. We feel quite confident in those goals. In fact, as we've executed our way through the end of sale and twisted some of the dials that we've twisted to improve our business, our internal plan has buffer to each of those. We're not taking up what those commitments are, but feel a lot more confident in our ability to achieve those. Then lastly, Carl touched on this a little bit in his opening commentary. I'll dive a little bit deeper into what we've done to get access to some of our foreign cash.
As you know, at the end of Q3, we had $2.4 billion of cash on the balance sheet, of which $2.2 billion was offshore. That was broken out in the 10-Q and in our prepared remarks. We've made several changes. There's more to do on this front, but have access to that cash. You've seen us already increase the rate and pace of our buyback, and you'll see that continue to increase going forward. Those are the key takeaways for my piece. Let me dive in a bit. As I said, continue to have a significant amount of confidence. In fact, increased confidence, both based on the way we've executed and based on some of the changes that we've made over the last 12 months in getting to those goals. One more. Okay. Andrew showed this slide earlier, I think it's important.
I'm going to step through the example on how we calculate ARR. The first point is what's in and what's out. What's in are the four recurring revenue streams that we've talked about, maintenance, and then the 3 types of recurring revenue that we categorize, that we sum for new model. That's product subs, cloud, and enterprise. We'll never get to 100% ARR. I think if you saw Andrew's slide early on, he talked about what fiscal '20 ARR looks like. That's not total revenue in fiscal '20. There will continue to be some element of non-recurring, some element of consulting services. There's a handful of legacy products that aren't recurring revenue. I'll show you a slide on this. The percent that's recurring will continue to increase, but it's not going to get to 100%. It's going to settle in somewhere between 90%-100%. All right?
The way we calculate ARR is we sum the actual recurring revenue for the quarter, then multiply that by four. The benefit of that calculation is you can tie our ARR directly to the P&L, directly to the reported results. There's no estimation going on there. There's no, "I think my ARPU is X, and I'll multiply by an exit rate." The downside is it creates the confusion that I'm about to walk you through. Timing makes a difference. When a sale takes place in a quarter obviously makes a difference on how much of that revenue gets recorded during the quarter. Since the way we do ARR is the actual reported revenue times four, a sale on the last day of the quarter generates a lot less reported revenue, and therefore, a lot less ARR than a sale on the first day of the quarter.
You've seen that effect in our ARR, and subsequently in the ARPS already. Two other things that have been a fairly significant headwind. One is FX, and we break this out for you again in the prepared remarks so you can see exactly what the impact of foreign exchange rates has been on our ARR. MDF for the first time last quarter. MDF is not new. We've had that MDF program to do what Steve just talked about. That's what our VARs use to do a lot of the demand gen to actually get into the field and sell our promotions and to sell our products. We've had the MDF program forever. What was the change last quarter is typically that MDF, which is not an expense.
MDF is a contra-revenue, so it's subtracted from the revenue line as opposed to being dropped into sales and marketing spend or something like that. Typically, that had been allocated mostly to license, because that's most of what our partners sold. As we phased out licenses and everything our partners sell are subscriptions, all of that $10 million, $10 million to $11 million a quarter of MDF got allocated to recurring revenue, to product subs, then we annualized it. So a $10 million haircut to the recurring revenue in the quarter times four said MDF had an effect of about a $40 million headwind to our reported ARR during the quarter. Let me walk through this example again just to make sure that everyone's got it. Andrew showed this in a summary level. You sell a $1,200 annual subscription on the very first day of the quarter.
Think of that as $100 a month in recurring revenue. $1,200 over 12 months. That's $300 reported during the quarter because you have three months, then we annualize it. You sell that exact same subscription on the beginning of the third month. So day 61, beginning of the third month of the quarter. What that means is you only get to recognize one month of recurring revenue during that quarter. So $100 gets recognized that quarter. We annualize that. That sale only contributed $400 of ARR during the quarter. Exact same sale as the one we did on the first day of the quarter. Then finally, to stretch this example to the extreme, you sell that same subscription on the very last day of the quarter.
You get, in effect, one day, we get in effect, one day of recurring revenue from that sale, which contributes, when you annualize it, about $13 of recurring revenue. What it means is, when you sum those up, 1,200 for the sale that happened on day one, 400 for the sale that happened on day 61, 13 for the exact same sale that happened on the last day of the quarter. The recurring revenue for the quarter is 1,613. We know on a normalized basis, and in fact, the next quarter, this will happen, those sales will contribute an entire quarter worth of recurring revenue. In the quarter that the sales happen, we have $1,613 of ARR from those three sales. In the next quarter and the quarter after that, those same three sales accrete 3,600 of recurring revenue. It's not an exit rate calculation.
In fact, when you then divide by the number of subs, in each case, it was three subscriptions. We just sold them at different points during the quarter. In the first case, during the quarter those sales took place, those three subs would calculate out to an ARPS of $538. In the next quarter, because they're in the deferred revenue, you get a full quarter's worth of revenue, those same three sales accrete the $1,200 of ARPS. The key point is timing makes a big difference. When those sales happen, and we're never going to get to a quarter where we sell everything we're going to sell for the quarter on day one. We're always going to have this. The effect dampens. By the way, you see this in maintenance. The effect dampens when the population size grows.
As the number of sales in the base that are accreting that 3,600 of ARR gets much bigger in relationship to the new sales in a quarter, this effect will dampen a bit, but it'll always be a lagging indicator. I think it's just important to understand that it's easy to think of ARR as, well, that's the exit rate. You made those three sales, ARR must have been $3,600. That's not the way we calculate ARR, and I think it's an important point for everyone to remember. Let's shift gears. How has the transition played out? Carl showed a version of this slide. There's a couple of things I would point out. This is the sum of new model ARR is the yellow line, and the blue bars are the new model subscriptions out through time.
One is on the yellow line, you see it actually it flex down a little bit from Q2 to Q3, and that's again, the effect of market development funds, the MDF. It was about a $16 million headwind during the quarter. We announced ARR in Q3, new model ARR of $414 million, which was a year-on-year growth of 88%. MDF was about a six-point headwind to that number, and currency was about a three-point headwind to that number. To the ARR growth rate. Think of that as 88 plus three for currency plus another six for MDF, if you want to normalize what that ARR growth rate looks like. The second point is, and I don't think it's surprising, but I'll point it out. Beginning in Q1, you notice the slope of both lines pick up. That's, of course, because we stopped selling perpetual licenses.
We stopped on individual products at the end of Q4 and on suites at the end of Q2. Not surprisingly, you see the new model ARR and the new model subs both inflect up at that point. Maintenance is going the opposite direction. You need to build this expectation in, going forward. I don't think it's new news. We've talked about this before. Our maintenance renewal rates are actually quite strong. They're stronger year-on-year because the incentive to stay current on maintenance, if you're bought into a perpetual license and you want to sweat that license for some period of time, you need to stay current on maintenance. If you fall off, you can't ever get back on. Renewal rates are stronger, churn rates are down, but we're not selling any new maintenance agreements. Any churn, one out of a million, it's not quite that good.
Any churn is going to drive down the maintenance subscription base and drive down the maintenance subscription ARR. For the quarter we just reported, subs came down, maintenance subs came down 34,000. We said, remember, buried inside there was a 13,000 inorganic add to maintenance from our Solid Angle acquisition. If you pull the inorganic piece out, what was the organic change in maintenance subs last quarter? It was a decline of 47,000 subs. Again, even with a very strong renewal rate, very low churn rate, you need to build your expectation for Q4 that the churn rate will continue to be very low. The renewal rate will continue to be very high, but the number of maintenance subscriptions that come due for renewal in Q4 is bigger than it is in any other quarter in the year, right?
Because historically, that's the quarter where we've sold the most perpetual licenses, and the maintenance agreement starts when you buy the perpetual license. The renewal opportunity in Q4 is huge. Even with a very strong renewal rate and a low churn rate, expect to see maintenance subs decline more than the 47,000 that we talked about in Q3, in Q4. Of course, that expectation is built into the guidance that we've provided you. All right. Let's look ahead. That's how we've done so far, and that's a little bit more of trying to articulate the way we calculate ARR and the effect that it has on ARPS. As we look ahead, if I can get the slide to go. Here we go. Showed you a version of this slide last year. We've made several changes in the last year. We've made refinements. We have made price optimizations.
Andrew talked about some of the things we've done to start to generate maintenance to subscription moving more quickly. A lot of changes under the covers, obviously, spend, and our position on spend growth has changed from this meeting last year. Two things have not changed. One is the endpoint. The endpoint is still the 24,320 target we talked about and $6 of free cash flow per share. Second thing that hasn't changed is the crossover point. By the end of next year, which is our FY 2018, by the end of next year, new model ARR and new model subs will be bigger than maintenance subs and maintenance ARR. All right. Talked about recurring revenue and that it's going to settle in somewhere between 90% and 100%. I just want to put this slide out again.
This is not a change from a year ago, I think it's important as you build your models. I think it's easy to get caught up in the P times Q on subscription and ARR and forget that there is a non-recurring component that you need to add back in. Expect this to settle in somewhere between 90% and 100%. If you look at the quarter that we just announced, we said it was 76% recurring revenue, which was a significant step up from Q2. It'll step up again in Q4, in the quarter that we're in the middle of right now, and start to approach that 90% range. All right. Expect it to settle in in that somewhere between 90% and 100% long term. Deferred revenue.
Deferred revenue continues to grow strongly, you can see what's happened and what the slope of that line looks like. This is simply an artifact of moving to subscription sales. All right. To having a lot more of the billings drop into deferred than drop immediately into revenue. Perpetual license sale would never go into deferred. Obviously, 100% subscription drives growth in deferred. We've seen it grow in the mid-20% range. We expect to continue to see it grow in the mid-20% range. Let me point out one thing. Andrew and Steve both spoke about the importance of enterprise business agreements. Right. Going after three-year agreements with our biggest and most sophisticated customers, and the benefit that we get, both in terms of the ARR that comes off of that and the increased adoption that happens inside those accounts.
Traditionally, those have been sold where it's a three-year agreement, even though it's a subscription style, they pay us all three years up front. That has been a very difficult set of conversations. I can't tell you how many conversations I've personally had with CFOs on the opposite side of the table saying, "I do my deal with Salesforce. I sign a three-year agreement. I don't have to pay for three years up front." What's built into the model is we will change from, we've already made some of this change internationally, it's really more of a change mostly in domestic, we'll change from selling our enterprise business agreements on a three-year paid-up-front basis to an annual billed basis, which will have the one-time effect of pulling down deferred revenue growth next year, while we build up a big unbilled deferred revenue balance.
If you do the quick math, fiscal 2018, we expect to end the year in the $300 million-$350 million of unbilled deferred revenue. We have a small amount today. It's not a material amount. It's less than $100 million. By the end of fiscal 2018, that number grows. Unbilled deferred revenue will grow to the $300 million-$350 million range. If you drop that back on, you'll see the total deferred balance, including the unbilled, will continue to grow in that mid-20% range that we've seen historically. Of course, once you make that reset once, it comes back to the mid-20% range going forward. I just don't want that to be a surprise as we get to next year and we start reporting deferred revenue, and the question is, what in the hell is happening?
Why isn't deferred revenue growing as quickly as it has done historically? It's not because there's a change in the underlying dynamics of the business. It's because there's a change in the way we are going to go out and sell those enterprise business agreements going forward.
There's very little this year.
There's not a material amount. Steve, can we have a Q&A at the very end? Just to answer your question quickly, there's very little at this point. It's significantly less than $100 million right now. All right. That's what's happening in deferred. Spend management, there's been a lot of questions on this. We've committed now to stay flat in spend for three years, right? FY 2017, FY 2018, and FY 2019. One of the questions is, how can you do that? I think some of the naysayers will say, "You can't keep spend flat and grow revenue the way you're talking about growing revenue." Here's the way we've gone after it. We are prioritizing continuing to invest in the areas that you've seen today.
In Amar's team, it's building out the cloud products, and he walked through both the opportunity and some of the success we're having with those cloud products today. We've continued to prioritize investing there. In Steve's space, we've continued to prioritize investing in the hubs and building out an inside sales process that we haven't really had historically. We've continued to invest in some of the internal business systems that we need to invest in to manage in the new world, and we've continued to invest in some of the digital direct marketing side of things. What that's meant is we've had to reduce spending elsewhere. The restructuring that we announced in February of this fiscal year, of this calendar year as well, was really about resetting our cost structure and resetting our cost base. We've gone through that. We've also looked at a couple of areas.
We've been relentlessly focused on driving efficiency in the engineering team and doing it by simplifying the product offerings and reducing the number of sites. Part of that restructuring you probably saw included a reduction in the number of sites that we have. We've also taken a very sharp eye to more mature and more speculative products and reduced our investment there, and that's what's allowed us to continue to invest in the areas we have to invest in that are more important to making the transition. You've seen us reduce M&A spend. You can expect to continue to see that be at a somewhat nominal level. It won't go to zero. There's always a certain amount that's opportunistic and a certain amount that's kind of tuck in, build versus buy. You've seen that our appetite for M&A reduced substantially over the last four quarters.
As I said, committing now as of last Tuesday to be flat, not just into next year, which is our fiscal 2018, but also into fiscal 2019. What that's done is had an effect, this is a great slide. Standing in this room my first Investor Day a year ago, kind of the same setup, although it's not like there were people standing against the walls. It was like a cast of thousands in here. I was the last one to present then as well. I showed a slide that this is the op margin, right? The gray line is the op margin from a year ago. I could barely talk over the clatter of keyboards when this slide hit the wall because it was just like, "Oh my God, did you see that?" This is what it looked like a year ago.
With the changes that we've talked about, with the optimizations that we've done, that we've made, continue to drive the top line to where it was. Of course, we have reduced what our appetite for spending is through this model. This is what that same curve looks like today. This is operating margin. The yellow line is what our current expectations are. You can see not as low in fiscal 2017, then continuing to increase into the low 30% range by the time we get to fiscal 2020. One of the questions I get a lot is, okay, well, what does that look like? If you're 33% operating margin, how much are you spending on sales and marketing versus R&D versus G&A and gross margin?
Just to give you a sense of that, we think gross margin continues to run in the 89%-90% range, R&D in the 18%-19%, these are percents of revenue, percent range. Sales and marketing, 30%-32%, which is competitive certainly against all the benchmarks of company our size and companies that sell the way we sell. G&A in the round numbers 8% range. That's how you get to the 33% operating margin. Last topic, cash management and capital allocation. Carl alluded to this earlier. We've made some changes. Well, let me start with cash flows. Cash flows from a year ago, before we made the spend commitments that we've made that led us to about $1 billion in fiscal 2020. Here's what this slide looks like today.
Right, $1.4 billion of free cash flow by the time we get to FY 2020, which is the $6 and $11. $6 of free cash flow per share, $6-plus of free cash flow per share in FY 2020, $11 by the time we get to FY 2023. No change from the commitments out in FY 2020, as I've said. As we've made a lot of these optimizations and some of the other changes that we've made internally, it's increased our confidence in achieving this by having the effect of we've built more buffer in our internal plans versus the external commitments that we've made. I do need to add one footnote to this slide, and many of you may know this. The SEC has come out with a pronouncement recently that said free cash flow is a non-GAAP measure. I think we all knew that.
It's a measure of liquidity. They specifically prohibit, not just us, any company, from providing liquidity measures on a per share basis. I'm looking at Jay, he's nodding his head. What we'll do going forward, there's no change in the business, there's no change in our commitment on free cash flow. What we'll do is give you the two components of it so that you can do the math to get the free cash flow per share to comply with the latest guidelines from the SEC. All right? No change in the commitment, no change in where we're headed, just a change in the way we are going to need to talk about the numbers going forward. Cash management and capital allocation. We've talked for several quarters about the work we've been doing on our operating structure.
This is modernizing an operating structure that is, in effect, the operating structure of a 30-year-old company. A 30-year-old company that's grown over that timeframe, its geographic footprint, and grown via acquisition. If you think of the tax org chart, the tax entities and how those are reporting, it was not optimum. It certainly was not optimum in terms of our ability to pool and, across the international entities, be able to pool and share cash. We've been working on that. There's actually several things we've been working on the tax front. The first is just to get our entity org structure in shape. By that I mean one where we have the ability to much more effectively share cash across international entities.
As we've done that, what has made this a particularly good time to make that change is, of course, we've got a tax loss in FY 2017. We've got accumulated deferred tax assets on the balance sheet that you probably saw about a year ago, we put in a valuation allowance again. We've got several hundred million of accumulated deferred tax assets on the balance sheet. Making this change, moving the cash into a branch of Autodesk U.S., it's fully tax affected. There's minimal tax friction because we can use our current year tax loss and some of our accumulated deferred tax assets, many of which were expiring, to offset that. What it's meant is we now have access in the U.S. to about $1.7 billion of the $2.2 billion of foreign cash.
Remember I talked about at the end of Q3, we had $2.4 billion of total cash on the balance sheet, $2.2 billion of that was offshore. We now have access to about $1.7 billion of that's in a foreign sub, but that's a branch of the U.S. entity and is fully U.S. tax effective. What are we going to do with that? As Carl talked about, the majority of that will be used to fund share buybacks. You've already seen us be more aggressive on our share buybacks. Year to date through the end of Q3, we've bought back about seven million shares. We continue to be active on the share buyback front. The majority of that $1.7 billion that we now have access to in the U.S. will go into share buybacks over the next several quarters. Some of it will be needed for liquidity.
If you remember the blue line and the yellow line on free cash flow, fiscal 2018 is the bottom of that curve. Some of the $1.7 billion will be needed just to meet our liquidity needs. I wouldn't rule out opportunistic, small tuck-in, kind of M&A that'll utilize some cash. That gives you a sense of what we've done, the amount of cash that we now have access to, and how we plan to use it. There's more we can continue to do on this front. We've hit the pause button on some of that as we try to understand exactly what the outcome of the election's going to be, and certainly some of the things that have been kicked around in terms of U.S. tax repatriation. There's more we can do, but right now we've paused that activity.
One of the other questions, last point that I'll make before I sum up again, is, hey, bridge me from your earnings per share to your free cash flow commitments in fiscal 2020. Having a hard time having my model. I'm doing the math and I'm coming up with X, and I'm not getting up to the $6 of free cash flow per share. This is not on a per share basis, this is in aggregate dollars. You can see net income of around $800 million. Add to it non-cash. I see people take a picture of the slide. You'll get the slide. Maybe I'll do this and I'll stand out of the way so you can take a picture of the slide. Add to it the non-cash elements that are deducted from our net income.
That runs in the, I'm going to have to go to my notes now, around $200 million range. Add the change in deferred revenue from fiscal 2019 to fiscal 2020, which accretes cash. That runs in the $600 million range. Subtract the growth in accounts receivable. Again, year-on-year growth in accounts receivable. Revenue's growing, accounts receivable grows, it's about $175 million. Our normal CapEx. We're not expecting a change in our annual capital expenditures. They tend to run in the $70 million-$80 million a year range. If you work your way through that bridge, that gets you to the $1.4 billion of free cash flow that we've talked about in fiscal 2020. Again, you'll have a chance to download the slides. All right. Last thing before we come back to the same key takeaways slide. It's a complex story.
We've been working hard to increase your visibility as we go through this transition. If you look at what we've already done, providing a breakout between maintenance and new models, we've built that in. We've also helped you start to bridge the way we currently report revenue in two line items to the three ARRs that we report. You see that in the prepared remarks. Many people have said, "I don't understand subscription revenue and why it's not growing more quickly." You go to the prepared remarks. We're giving you the bridge so you can see that. Looking at next year, instead of providing the bridge, what we will do is change our revenue reporting on the three line items. That you can tie directly to the three line items that we talk about. One for maintenance, that you'll be able to tie directly to maintenance ARR.
One for, it'll be probably called subscription, but think of it as new model, and then one for license and other. That's where we're headed. I think it'll simplify a lot of your own models and your ability to understand where we're headed. The second is, I think part of the concern on ARPS, and again, I get it, if you're doing the math on ARR divided by subscriptions and you see this downward trend, and we haven't told you what our expectations are on ARR, you got to do a P times Q to get the ARR. When we give you the Q, we give you what our subscription forecast is, but we haven't been giving you the outcome. With a declining trend, and all I've got is the Q, it concerns me that you can't get to your ARR.
We'll start to give you better insight under our expectations, not just on subscriptions, but also on ARR next year. Should take some of the angst out of the ARPS calculation. As we've done today, periodically we'll give you insight into the different types of new model subscriptions. Transition is on track. Lots of changes. The model is a living model. Each quarter we look at all the assumptions that are inside there. We look at the changes that we have made. We look at the changes that we want to make. A couple of things have not changed. The endpoint's not changed, and the crossover point has not changed. We've been diligent on spend control. We've committed to stay flat now on total spend this year, next year in fiscal 2019.
Talked about the free cash flow goals, this may be the last time you see in print the $6, which doesn't mean that we're changing our commitment on that, it's just reflecting the change in stance from the SEC. We've taken several steps to increase our access to cash, and that increased access to cash will go into an increased share buyback program. That's my piece. I think now we move to Q&A, let me just invite Carl back in because he's got a one-slide summary just to pull back together the various parts that we've done to de-risk the model going forward. Off to you first. Okay. If you want to do the de-risk slide. Yeah.
Let's just jump in.
Okay.
I think we've beat people over the head.
Jump into Q&A.
Yeah.
All right.
Let's start answering people's questions. Let's get everybody else in here and we'll start with questions. Who knows how to do the ARPS calculation? Is that what you're raising your hand? You know how to do it?
Roadies.
What's that?
Roadies for the investors.
Yeah, exactly.
Thank you.
Okay.
Steve Ashley, Robert W. Baird. I'd just like to go back to the slide on the maintenance loyalty discounting and the plans there. On the slide, it kind of inferred that we could see a maintenance price increase on the maybe first quarter of next fiscal year, FY 2019. Is that right? Was it a slide that should be interpreted literally or was it hypothetical?
That slide was literally correct.
That was correct. What, if anything, do customers get who agree to the loyalty discount? If they raise their hand and they increase price, is there anything they get in return for doing that other than locking in the lower price for a longer period of time?
They get to lock in the lower price for a period of time, they also get to move over to all the benefits of the subscription program versus the maintenance program. Those are real and tangible. The first one I want to make sure that you remember is the fact that they get more direct support from Autodesk. That's a real tangible asset and it's worth money. The other thing that you'll see is a pretty significant acceleration of our roadmap around the tools that allow them to manage their assets with Autodesk, their licenses, their users, their software updates. People, especially in mid-size and larger mid-size businesses, they spend a lot of money just managing the assets they get from Autodesk.
We are going to be able to remove real cost structure out of that long term, that's another benefit they're going to get in addition to the cloud application. There's real tangible benefits they get to move, yes, if they move early, they also get to lock in a better price for a period of time.
Gaurav Khotani from Soroban Capital . Thank you so much for all the detail you provided in your presentation. I just had one clarification question and one question about the channel. On the overall point of clarification, what I took away was you feel increasingly confident, I think that Scott said, have a bigger buffer to your $6 of free cash flow number. It also seemed as I understood your tax structure, you also have basically another $1.7 billion of costless capital that you're going to deploy against share repurchases. When I add the implicit $6 with greater confidence buffer, plus roughly $1 in incremental share repurchases, which is the $1.7 billion divided by your market cap without any cost capital. Again, did I get to a substantially higher number than the $6 you've guided to?
Is that a reasonable way of looking at it, understanding you do not want to change that number, but just want to build in a base of cushion?
Yeah, you're thinking about it right, Gaurav. The only refinement I'd make to what you said is some of the $1.7 billion, the majority will go into share buybacks, but not every penny of that.
Right.
Some of it we need for liquidity needs as we go through this dip in circulation.
Right. Maybe talking about one other component of the transition that you guys talked about on the call and you talked about today, which was the conversion of the maintenance subs to a subscription. Wouldn't that also have an incremental cost benefit as you close down a system that is a bit more antiquated? Can you help us put some parentheses around what that could mean?
Definitely retiring, going to one program definitely helps us retire systems. There is an overhang because there'll be a few long-term maintenance contracts out there and things, but it will allow us to consolidate some of the back office efforts, more focused on a subscription effort. It's hard to quantify right now, but of course, there's a cost of maintaining the two models.
Finally, on the channel side, obviously there's a positive mix benefit as you transition and grow direct faster, but also as you increase incentives for net new customers as opposed to just renewing customers at a baseline level. Should that also increase your net revenue realization as the same customer runs through the channel?
Yes, that's built into our model overall, but you're thinking about it the right way.
This is Keith Weiss from Morgan Stanley. Thank you for all the additional detail today and for the Q&A. Very helpful. A couple of more nuanced questions. When we're thinking about cost savings on a going-forward basis, and Scott, you kind of pointed to this in your presentation, it seems like there's significant costs that do have to come onto the income statement. You're going to be doing, it looked like around 40% of your business direct. That's going to be a billion-dollar-plus direct business. We take best-in-class sales organizations. That's like $250 million-$300 million in direct selling costs that have to come incrementally into the equation. You're going to be giving your customers more support. Scott, can you help us a little bit more on what's the flip side of the equation? Where are those costs going to come out of the equation on a go-forward basis?
We saw the restructuring that you guys did last year. Where does it come out on a go-forward basis that's going to offset all of that sort of incremental investment that's needed to support this growth?
Yeah. Keith, let me start by kind of correcting what I think your base thesis is. We are going to grow direct sales. You've seen it grow, right? It's grown with 29% last quarter. If you go back to the quarter before that, it was 25%. Don't equate that all into hiring more account managers, hiring more high-end sales. In fact, as Steve showed you, a lot of that sales coming through the eStore, which is a lot lower cost than the math you just did in terms of adding to the sales and marketing structure. As we continue to grow the direct sales effort, it's not we have to scale up bodies to grow that. I've seen kind of your thesis on it's hard to keep spending flat and grow revenue at the same rate.
While I don't disagree with that on the surface, if we were trying to grow revenue in the future the exact same way we'd grown it in the past, I'd agree with you. If you look at the way our revenue grows in the future, of course, it's much more a function of this shift to a subscription model with very low churn rates as we go into that subscription model. Everything we sell this year with very little friction becomes base sales for next year. You add new sales onto that, which drives growth. Same thing happens the following year. A lot of the growth is not we have to scale up the same way we've always scaled up. It's the growth that comes as a result of having made the shift over to a subscription model.
I think it's a little bit different than, hey, this is steady state. They're doing what they've always done, yet they somehow think they can grow revenue without growing spend.
What are the parts of the cost that come out of the equation as you make that shift?
Yeah, it's the things we've talked about, right? I'll let, actually, because you guys have done all the heavy lifting on this, I'll let you weigh in, but it's optimizations and simplification that have happened in the R&D team. It's Steve, in his case, rotating spend away from channel management as that becomes a smaller piece and onto hub sales and onto some of the more low cost but direct touch sales methods. It's some of the digital direct things that Andrew's done. You guys weigh in. As well as mature products and more speculative products.
I think you said it perfectly.
Let me sneak one more in. Just on one of those last slides that you showed, the bridge between net income and free cash flow. It seemed like deferred revenue was a really big component of that, about $600 million, just trying to eyeball the chart, which would seem to imply an acceleration in deferred revenue growth in the out years. Is there some mechanism that should accelerate deferred revenue growth as we go further out?
Yeah, no, I think you eyeballed it right. It's in that $600 million range, but it'll continue to be in the kind of the mid-20% range that we've seen over the last few years as we've made the shift to subscription. That growth, which again is driven by renewals that drop into deferred instead of jumping right into reported revenues, and then the continued increase in subscription models. While the numbers get big, don't think of it as there's a programmatic shift that's happening out there that's somehow going to dump more deferred revenue on the balance sheet.
Thank you very much.
Thanks, Keith. Jay Vleeschhouwer . Two quick questions. One, you've spoken of engineering software in the cloud as inevitable and how you're investing towards that, particularly with respect to your infrastructure, plus working with Amazon for infrastructure. Could you address a little bit more what you fundamentally will do in terms of changing the underlying architecture or deliverables? This may be more technical than you really want to address today, but it did come up a lot at AU a couple of weeks ago, and it seems to me that you're going to have to fundamentally change, as you laid out, what your underlying architecture and the nature of your deliverables are, to meet this inevitability of engineering software in the cloud. Particularly if you go to a converged licensing model where everything becomes just one single service or license.
That's question number one, if you could address that, please.
Sure. Jay, we've been working on this for quite a few years now. We're not using the same code that we've used on the desktop to underpin our cloud efforts. We've been rewriting, re-architecting it for scalability, multi-tenancy, for security. One of the big changes we have between the old world and the new one is we're sharing a lot more code across all of the applications, right? I think in the desktop world, Oops. Don't kill myself with this. We would have 30% of the code be similar between applications. In the cloud world, we have almost 70% of the code that's similar between our applications. We really have architected next-generation scalable solutions that are very large shared code base. One of the reasons now, I think you were at AU, you saw us talk about BIM connecting to fabrication through the cloud.
A lot of it is leveraging some of the IP that we've built for Fusion. That kind of capability now comes to us as we build our next-generation software. We've taken advantage of everything that's available from Amazon, from the other guys, like open source. Lots of leverage that's available in cloud architecture. We've been building towards that.
Yeah, let me just add. We back up to where I started this morning and said, "Here's how the market expands, here's what's going on." This is the view of the industry through the customer's eyes. There are two things that the new technology platform affords that the old one doesn't. One is this instant access to kind of infinitely scalable amount of computing. Rather than buy for your peak demand, instead, you can farm off whatever you need done, whether it's a simulation job or a visualization job. On-demand, you get as much computing power as you want. The second thing that's important is the biggest problems our customers face is the one of communication and collaboration. It's the sharing of information. I would say the collection of single-product solutions that we have on the desktop has been effective, but very difficult for sharing amongst teams.
All of our customers work in teams, some of them quite large and global, it was an unnatural architecture to be behind one company's firewall for an inherently collaborative process. If you look going forward, just as we all enjoy in our personal lives, the ability to have something in the cloud that we can all easily share, it's the exact same thing for whether it's a manufacturer and their supply chain, whether it's the hundreds of companies working on a construction site. It's this natural kind of isomorphism between the IT architecture and the economic structure of an industry that makes sense in the cloud.
All right. For Steve, question is, you've had some good success over the last number of years in not only growing the number of large accounts, but substantially increasing the number of very large transactions, GE and others. Sort of to go through the whole list. The question there is, what has changed that has allowed you to go from the days when a few hundred thousand dollars was reason for high-fiving in the halls at Autodesk to seven-figure, even eight-figure deals at the company? On the other side, you've also talked about the inevitability of channel consolidation. We've seen it now, we've seen it before. How do you manage the risks of that, to the extent there are risks of disruption as they go through that? We saw it 15 years ago, for example. You got out of it, but you know what I mean.
How do you manage that potential consolidation risk?
Sure. On the first point, I would tie it to process and people. We have reshaped our engagement process with our largest customers overall. We have trained our folks on it. We have a repeatable process that everyone follows through. We've defined who's responsible for what, how do we actually engage with our customers. I'll tell you, the actual reshaping of the company's focus and the future of making things has actually been the source of changing the conversation in general overall. It's a combination of those things that's really positioned us to engage with these large companies very differently than before. You're right. My first quarter here, 14 years ago, we couldn't find 10 deals over $100,000 in size. Now I only track million-dollar-plus deals because we have a large pipeline.
We have great people that we invest in to develop those relationships and to understand how to engage effectively in a world of consumption in order to be able to drive that business, as well as investing even in customer success. Reshaping our services focus on customer success is what's enabling us to drive expansion and enlarge renewals once we get the first EBA. I talked about the benefits we get by closing the first EBA compared to engaging with these customers in the older business models or the non-EBA world. What I didn't talk about, but I'll share with you, is we renew at much larger values than the first contract we closed, and that engagement process has changed dramatically as well.
Yeah. One of the things I would add
Is I think Steve and his team have done a fantastic job. The delta between where these deals start and where they are now is enormous. There's another part of it that says, we've transformed the company into one in which the customers, at times of change, at times of consternation about their future, seek us out as a partner. They aspire to work with us. I think if you would go back to the beginning, we were more of a commodity supplier to them, and many of the thoughts were, how cheaply can I get the thing I need from these folks? I talked today to you, I know you were there, Jay, about I've seen more auto execs this year, in one year than I've seen in the previous 10. Time of crisis in the industry, they're wrestling with a number of dramatic changes.
They see us as a company that's innovating. One of the things that I think is really important and is often missed is positioning yourself as a company that other companies aspire to work with, learn from, have information. When we do some of the things like we've done on generative design or additive manufacturing, or this idea of combining both the design with the making, that's something where customers now reach out and want a very different relationship. In general, I'd say these relationships are spearheaded by very different parts of the organization. These are vice presidents of engineering or the CEO who's driving the agenda rather than an IT director.
Jay, that was what I meant when I said the future of making things. We're engaging very differently with these customers, and they're not just coming and visiting, spending time. We actually host a lot of customers in this room, like Amar mentioned earlier, and it's not just one executive. We'll actually see automotive companies bring their entire leadership team, and we're actually mapping out initiatives that they're working on, the areas that they're most concerned with. They're looking to us as a strategic relationship partner to help them actually deal with those disruptions in a way where they gain an advantage in their market. The conversations have changed dramatically with those customers. Oh, yeah. Jay asked about partner consolidation. We've actually gone through a lot of this already.
If you recall back, we were talking probably about 2,500 partners not that long ago. Our partners are really good business people, and I'm actually really pleased in how we're engaging with our top partners around the world. They see opportunity. Our best partners see opportunity to continue to grow, and they actually see the opportunity of consolidation as a benefit to them, because if they can expand and have fewer partners they're butting heads with so that they can make strategic investments in their business, life is good for everybody overall. We work very closely with them on that. We recognize that not all partners are going to make it through this transition, and that's okay. What I'm pleased with is that our best partners aren't buying the really weak partners. The better partners are combining with each other to become strong or even stronger.
Many of the ones that are really weak and weren't set up to make this transition aren't going to make the transition, and that's just fine overall. We track it very closely, Jay. We work with our closest partners around the world with partner advisory board sessions and things to make sure that they're aligned with where we're heading and vice versa. That's how we manage through the process. I'm really comfortable with how it's gone, and I'm very confident that we'll continue to work with our top partners to ensure that we have the coverage we need for FY 2020.
Hi. Jay was asking the front row question. This is the back row question. Kash Rangan from BMO. One on pricing. I think I asked you, Carl, over break, do you offer these discounts? I'm curious why you offer a three-year discount, because the longer customers are lulled into a 70% discount, whatnot, they wake up three years later, and then they're stuck with a 300% price increase. You're selling into lower margin businesses. These are not exactly financial services and spend tons of money on IT. Just curious, what drove the decision to do three-year terms on these discounts versus what could be a one-year thing, like an Adobe did a one-year, very quick resurgence back to normal pricing. Also, to follow up to the channel question, curious if the channel consolidation is even more intense?
Clearly what you've seen in the last 10, 15 years is not indicative of what's going to happen in the next five years, because the pace at which you're moving business direct versus indirect is a lot more exaggerated. How are you managing the next three to four years in a different way than what has happened the last four or five years so you don't lose any lockstep? Thank you.
I'll take the price one. Kash, on the price one, it's actually pretty simple what led us to the kind of promotion strategy we have. The other element that you're not taking into account is that we ask the customer to surrender their perpetual right. What we found is that if you balance the term and the offer appropriately, it removes what the perceived loss associated with turning in the perpetual right. Really what you're seeing with the multi-year term is an optimization to basically barter out the perpetual right. Adobe didn't have this challenge to the degree that we would have it.
On the partner consolidation. Actually, you'd be surprised to see how we've mapped out the progression of the consolidation, and the pace of change has actually been very dramatic over the last couple of years. We've just been able to manage it really well so that impacts have not been negative. They've actually been positive. I would expect to see the same rate of change over the next several years through FY 2020 that we've seen over the last couple of years. It actually will get us to a target zone where we'll have a smaller base of partners around the world who actually will provide greater coverage, more feet on the street, and more scale than what we currently have.
If you compare to what we did prior to a couple of years ago, the pace of change is very different now from that. If you take a look at how we're mapping things out compared to how it's been working over the last couple of years, we're actually executing on a plan, and the pace should be fairly straightforward and similar for the next few years.
Sterling Auty with JP Morgan. One question, one follow-up. On the piracy or the non-pay, you gave us great detail in terms of how you're going to locate those non-payers, the ones that you don't know. A simple question I get over and over again is, how are you going to enforce the anti-piracy in the go-forward model? I think some of it seems straightforward, but just help investors out in terms of technologically, what are you doing to not have pay for five but use 10 or be an outright pirate?
Right now, when you go to a subscription model, obviously you're subscribing. We're able to track the device you're using and the ID. Ultimately, if we really want to enforce piracy really deliberately in the subscription model, if someone is taking a single subscription, yet it's lighting up across multiple devices at the same time, we know there's a pirate there. We haven't really seen a need to actually go do anything like that in the subscription model right now. If you see us moving forward, we'll easily be able to understand if somebody's lighting up a seat across multiple devices simultaneously and be able to comment on that. That's not a problem right now, but it's easy to see how we do it in the future.
Okay, great. Then one follow-up for Scott. You talked about the incremental or the increased pace of share repurchase. How should we think about the pace of the number of shares declining? What's embedded in that question is, what should we be thinking about as the dilution from stock-based compensation as the offset to the repurchases that you're doing?
There's actually two elements to new shares that come in each year. One is the stock-based comp that you talked about, the equity plans that go fairly broad inside the company. The second is our employee stock purchase plan, and we have an extremely high participation rate on our employee stock purchase plan. The sum of those two result in about between five and six million shares each year. If you look back over the last couple of years, in addition to those new shares that came in from those two programs, we had a long tail of options overhang. We'd issued options with a 10-year tail for a number of years. As the stock price started to perform, a lot of those options also got loaded in.
Over the last couple of years, we were spending, round numbers, $400 million a year simply to offset the dilution from those three effects. The option overhang is negligible at this point. You can see it in our 10-Q. Think of the new shares that come in somewhere between five and six million shares a year. The amount that we talked about, obviously, we'll continue to have the ongoing offset dilution, and then we'll layer on additional share buyback.
Hi, it's Gregg Moskowitz from Cowen. Andrew, I wanted to go back to a couple of your earlier slides. It looked like you were forecasting something like a 70% increase in product subscription ARPS between now and fiscal 2020, going from roughly $430 to $700-$750. I know there are a few components at play, how much of this is from the lag effect on ARPS, and how much is from pricing changes?
It's going to mostly be from the latter. The retiring of the promotions or the ramping down of promotions is going to have a big impact on that. The shift to the higher ARPS offerings is going to have a bigger impact. It's mostly that than the former, the lagging indicator from the ARR customer.
Maintenance migration will play a-
Maintenance migrations are going to play a big factor there as well.
Yeah. Okay, just a question.
Remember, I blended those two numbers in that slide. I blended the product subscriptions and the maintenance. Basically what you're seeing is the aggregate change.
Yeah. Obviously once we get to FY20, the maintenance piece is going to be quite a bit smaller than it is.
Exactly.
It's mostly a product sub.
Exactly
delta there. Question for Carl. I'm curious, if we remove a one-off dynamic such as Japan, for instance, what's the biggest pushback that you're getting or hearing from customers as it relates to the new model? Thanks.
We're not seeing a big pushback from customers or from partners. Matter of fact, I've been surprised when I've read people's notes about this, because it doesn't correspond with our experience. I think some of it may be some of the grouchiness in some of the smaller partners.
Right.
From a customer point of view, one of the things that's happened that's to the positive is when we first started this transition, we were kind of leading the charge. Now we are not alone in this at all. I would say IT departments have come to accept this as the way they're going to purchase software. We're just not the high pole in the tent anymore. I think we will see with the current model, someone comes in and says, "We offer you a lower entry price that after three and a half years will cost you more," there's no problem with that. The place where I think we have to be careful in how we roll this out is the one Andrew addressed. Asking people to give up what they perceive to be their ownership rights and moving to a subscription model.
Remember we said, these are our best customers. They've bought a perpetual license. They may have paid for five or 10 years. I think there's a little sensitivity in how you handle them, because they think they're giving up something. I'd be surprised if in the three to five-year timeframe, just in the software industry in general, you'll see anything that much resembles perpetual licenses and maintenance. I think customers are getting actually quite comfortable with it.
Very.
Yep.
Even those that actually, a couple of years ago when they first heard about it, were saying that they didn't like it, are now actually moving to-
Quickly
in fact, to the new model.
Hey, thanks. Richard Davis, Canaccord. Is there a pull it and take at all with regard to your cost structure depending on how your customers use your cloud back end? Because you talk about the AWS back end, so if I was a really heavy user as opposed-
Yeah
to using my on-premise processing power and storage, et cetera. How does that play into the margin structure? Because a lot of-
Yeah
maintenance margins are 80-something%. Most subscription guys are 70-something%.
Yeah.
You have a little headwind there.
It absolutely does.
Yeah.
One of the ways that we've addressed, and the way that customers pay for it, is think of the ongoing kind of steady state stuff as being in the cost of subscriptions.
Yeah.
Most of the heavy lifting is on a consumption basis. While we won't charge that way, it's tied very directly to how much computing resource somebody uses to do the task. In the extreme case, where we now have people rendering entire movies, taking hundreds of thousands of CPU hours, it's all on a consumption basis. That kind of addresses the flexible part versus the fixed part.
Anil Doradla from William Blair. Amar, you talked about the BIM 360 and the Fusion 360 products. Obviously, these are a massive new TAM expansion.
Yep.
You guys have been working on it for a couple of years. Can you talk about the readiness of these products in terms of going after the whole market? I would assume BIM 360 is well ahead of Fusion 360. I mean, there's some new technologies. Are you able to address 1/3 of the market, 1/4 of the market? Any kind of color on that?
I don't know if I'd address it as a percentage. I would say, if I look at Fusion, it's capable of doing a lot. I would say, if you look at what happens in a medical device process or consumer products or automotive, Fusion's capable of doing 90%-100% of what's needed. I think the main difference between a desktop product and something like Fusion today is something like Inventor will do assemblies that are hundreds of thousands of piece parts, right? The cloud architecture is on its way to getting there. I think it's not so much of a capability differentiation as it is mostly capacity at this point. Just like any cloud-based technology, it's just getting better. I would say Fusion's very capable of doing most of the demanding tasks in the world of manufacturing.
BIM 360 is right now focused on construction process. We haven't added the design component into BIM 360. At AU, we previewed design and engineering and analysis and fabrication coming to BIM 360. BIM 360 is much more tightly focused right now on the process part than on the design part. That's the way I would think about it.
As a follow-up, over the next couple of years, could we expect more internal development or external M&A from a feature point of view?
A little bit of both. I actually think that we have a very aggressive development schedule, lots of things going on. I think the majority of what we will do is internal development. I think we have a lot of the pieces under in-flight right now that will be landing, whether it's more manufacturing capabilities, sheet metal, those kinds of things. I think a lot of it is internal development. Obviously, we will continue to look opportunistically for something that is unique and interesting that we might add to what we're doing. I think most of what is in front of us is internal execution.
Hey, Saket Kalia from Barclays. One question, one follow-up for Andrew, if I may. First, just as it stands right now with the new program to convert maintenance to subscription, where should we see maintenance subs end up roughly by the end of 2020?
I think you're going to see it basically fall down to next to nothing by FY 2020.
Got it. The follow-up, you talked about the opportunity to convert a bunch of active non-paying users. Last year, we talked about that infamous 2.8 million converting 30% of them. How do you think about the conversion rate on maybe this slightly larger pool and this pool that we've got a little bit more analytical data on? Thanks.
I think history's the best guide here. There's 2 levels of how this works. The customers that take advantage of the programs. We've just done 70,000-plus this year on the programs. There's a bunch that come in just as the natural course of doing business outside the programs. We don't have clear line of sight to those yet, but I think the performance we're seeing this year, around 70,000-plus, let's just say another 20,000 or 30,000 that come in from the natural course. I think that's the kind of performance you're going to see us driving to moving forward, maybe a little bit better. You're going to see us start driving those kind of rates. I wouldn't anticipate some dramatic acceleration until a certain period of aging happens.
Remember, we talked about this notion of some of these people are just going to age out. As you see us get further and further out, one, two years, you're going to start to see more people just age out. I would expect past performance to predict future performance fairly reliably here.
Thanks, gentlemen. Steve Koenig here from Wedbush. Thanks for the Investor Day and the detailed presentations. Maybe one kind of a housekeeping sort of question and then one follow-up. On the free cash flow guidance, Scott, can you help us compare the $1.4 billion to what was implied by the prior guidance in terms of absolute dollars of free cash flow?
Yeah. The $1.4 billion, do the math, right? The basic shares outstanding, if you hold aside the fully diluted, around 225 million shares.
The $1.4 billion is entirely consistent with the $6 that we talked about previously.
Okay. On the maintenance price increase, the chart was conceptually helpful. Whatever details you could provide us would be helpful in addition about that loyalty pricing and where maintenance will end up. Will it end up after the two price increases being as expensive as a subscription? The loyalty pricing that's offered, would that be roughly equivalent to the subscription price or would it be some ways still more expensive but less than the maintenance price? Any sort of directional indicators you can give here might help us out.
I'll give you some directional indicator, before I do that, let me tell you what we optimize for. This is important. We're solving an equation for maximizing ARR. This is our goal. If I simplify it to the most base simplification, and those of you who do the spreadsheets know it's not that simple. If it's P times Q minus churn, the formula has to be optimized so that you're setting prices so that you minimize churn. The way we're going to roll this out is to maximize ARR, not try to provide maximum leverage on the maintenance program or something else. It's going to be looking at this balance of P times Q minus the churn. I just want to make sure you understand how we're arriving at some of the things we're doing. What I showed you conceptually on the graph is accurate.
There was a gap between the subscription and the terminal maintenance price. That will exist. The price that ultimately rolls out to customers in FY 2020 that are on maintenance will be below the subscription level. Everything I showed you on there is conceptually correct. What's missing is the absolute numbers. The absolute numbers you'll start to get in the next couple of months, if you look at the slide, everything there is conceptually accurate to the direction we're going. Except for the scale, the gap difference, because I know some of you are now going to start taking rulers. That part of it is not accurate.
Okay.
Thanks. It's Brent Thill with UBS. Carl, if you look at your app peers, Adobe, Intuit, they had breakout plays going through these transitions. Adobe with the Marketing Cloud, a $1 billion business growing 30%, Intuit went global, from being a U.S. company. What's your breakout play? Is it construction still? Is there something else that's-
I think the two biggest opportunities for us continue to be construction and broad-based manufacturing. We competed in AEC, but as we talked about, it's mostly been in A&E, the construction is by far the largest part of that multitrillion-dollar industry, and growing the fastest as they continue to invest in IT. The second one is while we played a good role in manufacturing, being able to do the new kinds of things in this transition, that is the breakout. I would look at construction and manufacturing as it, I wouldn't look for a big, new kind of diversionary opportunity.
Steve, can I just clarify on the 25% that are EBAs? Where do you think that is long term?
Ideally, I'd like to get all of our named accounts, that's more aspirational because there's reasons why customers move and when they move overall. I'll just leave it as a large majority of our named accounts should be on enterprise business agreements by the time we get to FY 2020. One of the things we do, we're using a lot of data to determine who should be in our named account program. A lot more analytical in our approach overall. One of the key elements of defining whether an account should even be in our named account program is likelihood of benefiting from going into an enterprise business agreement. I'd like to get them all there. Saying 100% is a hard number to hit. Let's say a large majority of them.
Hi, it's Gal Munda from Berenberg. I have a few questions. The first one, just in terms of the churn. How are you seeing maintenance churn at the moment, compared to previous years? When people churn, how much of them go into subscription, and how many of them just stop paying and maybe become quiet for a while?
On the maintenance side, yeah, we have seen it actually, the churn rate go down and the renewal rate go up. Part of the reason for that is, historically, if you were on a maintenance plan and you, for whatever reason, decided to let it lapse, you could get back in. You could get back in by buying a new perpetual license and then adding maintenance to it. That's no longer available. We don't have any kind of a, "I've got a perpetual license, can I somehow get current on maintenance?" If you fall off of maintenance, you're off forever. That's had the effect, as you'd expect, for people that took advantage of that last buy to buy that last perpetual license. They're going to stay current. They want to sweat that license, right? They bought it knowing they're going to need it for several years.
The attach rates went up toward the end of sale of perpetual license, and we see the renewal rates on maintenance continue to stay high.
Now that you have a few data points about subscriptions already, how do you see subscription churn or renewal rates trending versus maintenance?
I'll start, and then you guys feel free to weigh in. There's a big difference between the renewal rate for subscription and maintenance, simply in that you could drop off of the maintenance program, your license still works. If you drop off the subscription program, it no longer works. As you'd expect, we would expect to see those renewal rates, particularly outside of monthly. Monthly will have the highest come in for a project base, drop back out. That's not unexpected on monthly, but as Andrew showed you, monthly's become a pretty small part of the base. For the remainder, we expect to see it be at or above what we would see in maintenance renewals.
Exactly. Monthly has high turn, annual has fairly low turn.
Can I just have a follow-up on how you compensate your channel and your internal sales force on the new booking side and on renewals, and how that might trend in the future?
I could take a long time to talk about all the different compensation plans. Let me give you a high-level general feel, because it varies by role. That's why I say that. From an internal point of view, a majority of our sales reps are focused on new and not renew, because the activity and the actions we want them to go drive is finding new customers and closing demand. We do have some folks that are focused on the renewal side as well, so we have bifurcation of roles overall. You should think about for territory sales folks and inside sales folks that are outside of the named account program, that the majority of their focus is on driving growth in ARR. That's the best simplified way of thinking about it overall.
Named accounts is driving overall ARR growth, and because they're responsible for everything, they've got all the relationships. They actually focus on moving the renewals of the older business models to the newer business models and the EBAs. From a channel point of view, I gave you a guideline in my deck of how we're incenting the partners on the different components of the business overall. As opposed to repeating them again, if we have any particular areas of question, I'm happy to answer later on. There's a difference between new versus renew. We do have them doing both of them. We pay more for new than we do renew because the sales motions are more costly and more focus there.
Thank you.
Hi, guys.
How you doing?
Hi, Ken Wong from Citi. I think the thing I want to focus on was on cloud subscription ARPS. Today it's at $290-ish. It looks like you guys have it trending to $200, $250. I can't help but notice you guys are talking about greater functionality, and it's going to align with the desktop products. Why wouldn't the ARPS for cloud be higher?
I think, again, we'll get to having to break this out because we're going to have real differentials within that cloud bucket.
Yeah.
When you look at broad-based cloud products like BIM 360 and spreading it to thousands of people within a construction company, those are going to be at the lower price points. When you look at something like Fusion 360, where Amar was showing it was being used for the design of an automobile, it's going to be at or higher than our other product subscriptions. Again, it's really a mix issue, but we're estimating on the manufacturing side and on the construction side, driving much greater volume. It really comes from that. Otherwise, the functionality is what should drive the price points. In cases where we're providing more functionality than people get today, they should and will be paying more for it.
Got it. We should think about the products that will line up with Revit and full AutoCAD-
Exactly.
-at some point, they end up at our buying point.
They're more expensive because what they have. This is probably worth spending a second on. If you look at the comparison between Inventor and Fusion or Fusion and a competitor like SOLIDWORKS, one of you actually said to me at the break when they've asked Fusion customers, the description they had is Fusion does 80% of what SOLIDWORKS does, but SOLIDWORKS only does 50% of what Fusion does. I think both those things are true, even though at first you have to spend some time trying to parse how that could be. What we decided to do with the Fusion product is address a larger problem set. In the narrow space of design, it does less, and that's how you get to 80. In terms of being able to do simulation and CAM and collaboration, traditional desktop products don't do that.
In all those cases where we're going after an industry and providing way more functionality, it should be considerably more expensive than, let's say, what someone paid for a product design suite or for Inventor standalone.
Got it. As we weigh things like dialing back the promos, you guys are adding in new maintenance pricing. How should we think about the seasonality of subs? Is that going to be less volatile than we've seen in the past, or equally volatile?
Oh, you're the soothsayer.
Well, it's absolutely-
What are you thinking about for seasonality?
It's absolutely going to be less seasonal, especially as we accumulate more and more into the base. You're right. As we dial down, the promos have driven some kind of artificial seasonality, which are purely connected solely to the quarters we chose to do those promos in. As those go, the discounts decrease, you're going to see less of an effect there. Obviously, as you accumulate more and more ARR, you're less sensitive to some of the seasonal surges with new seat purchases. Yeah, Ken, the only qualifier I'd add to that is our EBAs, and you know the way we count subs-
Yeah, all the EBAs-
-we sell a lot of them in Q4. In some cases, if it's a new EBA, we're selling a new account that has a lot of subs on maintenance, and they don't renew that. They buy an EBA instead, right? The maintenance count goes down, but it takes us 90 days to count the monthly. We count the monthly active users for 90 days before we add it to the subs count. If you look at our trends, EBAs in particular, enterprise adds, have been high in Q1 as we catch up the monthly active user measurement of the big load of EBAs we sell in Q4. That's the only qualifier I'd add to what-
I would agree. Totally agree.
Hey, thank you so much for taking my questions. Rishi Jaluria at JMP. Two quick ones on cash flow. Apologies if I'm making you repeat yourself. First, in terms of the $1.4 billion free cash flow target, what sort of headwinds should we expect from the shift to 3-year billings up front to 1-year billings? One follow-up after that.
What I mentioned when I was showing you the deferred revenue slide is I wanted to point out that it's not going to grow as strongly in FY 2018. We think the unbilled deferred, so the amount that you'd add to that line, is in the $300 million-$350 million range by the end of FY 2018.
Okay, thanks. Just how should we be thinking about the delta between operating margins and cash flow margins trending over the next couple of years and maybe even just out to 2023 with your super long-term target? Thank you.
Yeah. I think the easiest way to think about that is to go look at the bridge slide that I showed you, where you start with net income, which is a pretty close derivative. We're not assuming big changes in tax rate. Pretty close derivative to op margin. Look at the various changes that bridge that net income to the free cash flow number, and you can get a good sense of it.
Great. Thank you. Heather Bellini with Goldman Sachs. Thanks for all the great detail today. Carl, I had a question for you. If we go back years ago to past analyst days, one of the things you used to talk about was a growth rate, long-term growth target for Autodesk of 8% to 12%, so call it 10%. I'm wondering if we could think a little bit past the transition. Should we be thinking of a similar level for Autodesk? Because of the 10 expansion opportunities and the fact that you're looking like you're having a lot of success finding new subscribers to some of these cloud offerings, could it be better than that? Then I just had a quick follow-up for Scott.
Okay. Well, first of all, you missed my compliment the other day on your model, Heather.
Sorry, I was on a flight. Shatil should take the credit, though.
We knew that anyhow.
I know.
Shatil did a great job on that.
Yeah. I would put it as the same nominal organic growth rate that's in there. There's probably a slight bias higher, particularly in the two places where, like in construction. In any secular increasing market, I think adoption will be fast and you'll be going up that ramp. There'll be an aspect to it, but it'll blend into a bigger business. I think on the margin, it's higher than that growth rate, but moderated by the fact that the base will be fairly large at that point.
Okay, great. My follow-up for Scott is, just in the bridge you gave, which was very helpful, again, thank you. When you think about that change in deferred revenue that you showed, that roughly $600 million, what percentage or how big is long-term deferred of that $600 million? I know that's a tough question, but that's one we get asked is, look, is a lot of this just going to be long-term deferred revenue? Is 40% of your DR balance going to be long term and therefore, maybe people don't want to give you as much credit for it?
Sure. I don't actually have that cell in the spreadsheet memorized, I apologize. If you think of the trends that would drive that, one of the key things that'll work against a faster growth in long term versus short term is what we talked about on EBAs, moving those to annual billings. Right? Before, we'd sell a three-year EBA, and we'd put it all, anything beyond two years would go in long term.
I mean, mathematically, it should be small. Right. That's what I was trying to get to. Thank you.
Hi, Ken Poladian from Evercore ISI. First question, you've talked a lot about collaboration. When I think about how architects work, a lot of times they'll build a building, right? They'll design, build a building. That will be finished, they'll go back, start from scratch. Rebuild another one. As you enable more collaboration, is it possible that you could actually facilitate a move towards more either open sourced or pre-built models that reduces demand for engineers then ultimately CAD seats?
I thought you were going somewhere else with that question. I would say that definitely as there's more data available and as the cloud lets you recognize patterns, the ability for architects to reuse information or to even apply machine learning to do better designs and do more generative type approaches becomes larger. I think that's what you're going to see happen. I don't think that reduces the number of jobs. I think it increases the ability for them to do more of what today is being done through discovering errors on the job site. I don't think it eliminates the job. I think it just makes what they do more valuable.
Yeah, I think a good example to look back at is when we introduced Revit.
Yeah.
Everybody thought, okay, you have hundreds of architects who are just slaving over drawings, making sure that there's a totally coordinated drawing set. They're basically doing an update in an Excel spreadsheet, but over huge drawing sets. We worried about it somewhat. It's turned out to be a totally unnecessary worry. What it enabled people to do is concentrate on more important parts of the job that they had to do. Now everybody just assumes there's a base of functionality, that they can make changes at the last minute, and clients tend to ask for more changes at the last minute, and they get updated construction sets. I think the same thing's true here with some of the newer technologies that Amar was talking about.
Okay, for the second, Scott, could you just walk us through how you've changed your process for building guidance now that you have perpetuals out of the picture?
I don't think the process has changed as much as the inputs to that have changed, right? If you think of 100% subscription model, we're not 100%, I told you, but it's going to be in that 90%-100% range that's recurring. Obviously, you start the quarter with a lot of the revenue sitting in the balance sheet. Right? We have a lesser amount that we have to forecast because we've got a substantial sum in hand. It just goes back to the way we've always done forecasting, which is what is the sales activity, what's the timing of that sales? Opportunities, when are they going to close? There's a piece that comes through the channel that's more mathematical than it is sale by sale, but there's a lot, especially in Q4 with EBAs, that we have specific transaction-level detail on it.
The process itself kind of rolls up the same way. The degrees of variability are a lot lower.
It's Sterling. Just one quick follow-up call, Andrew, to the comment that you made earlier. We've got the increase in maintenance pricing as part of the promo. I think all of us, we look at our models in terms of the ARPS and the ARR, right? I imagine even though the price increase in maintenance goes through, have you thought about or is there a trend? You gave us the chart, the trend of the ARPS for new model. What does that chart or that trend look like for maintenance ARPS? Because I imagine it's going to be heavily impacted by the mix of enterprise maintenance customers versus others.
Yeah. Obviously the trend's up if you're driving price increases and moves to loyalty prices. The trend's inevitably up in the model I showed you right there for that maintenance base. Was that your question?
The ARPS is, there's lower price maintenance users that if you raise those prices, you raise the other, but it depends on the magnitude of the falloff. For example, if more of your higher-paying maintenance customers move to subscription quicker
Yeah.
your ARPS could plummet for maintenance ARPS, even though you raised it by 20%.
Yeah.
That's what I was asking, what does that maintenance ARPS trend look like?
Yes, there could potentially be some of those dynamics. We haven't dissected it to the level that you talked about there. I think we'll start to have a lot more insight as we start rolling out the loyalty programs and who moves earliest. I don't have that level of granularity in terms of who's going to move first.
Aren't you looking to collect the?
That's why you combine them.
Right. he's.
Yeah, no, he's asking a very specific question. He's asking a very specific, who in the maintenance space would tend to move first?
In aggregate, you're looking at ARR.
In aggregate, it's inevitably trending up.
Right.
Hi. RK Mahendran from HMI Capital. Can you explain a little bit better how EBAs work? Simply being, if it's a token model, is it that you pay $X for Y tokens, and if they run out of tokens, say it's a three-year model, at a year and a half, they have to come back and buy tokens? Or is it, "Hey, you used 2 times Y tokens, and next time you're just going to be priced up to, basically, your EBA will be 2X what it was last time.
Yes. The way it works is we actually run usage reports before we ever move a customer into an enterprise business agreement. We have an idea of what their run rate is and what their usage paradigm is. That's what we use as data to actually get them to commit to a certain bank account of tokens, is the best way to think about it. Each of our products have different token values. Higher-value products consume more tokens for use than lower-value ones overall. We make the entire portfolio available. We work with the customer to get as many active users that are using products and chipping away at that token bank account. Oftentimes, we do find customers running out of tokens before the end of the term, and we actually get a new transaction.
They actually will add more tokens into their pool until the term of the contract expires, then we'll actually do a new deal, which will have even larger amounts of tokens overall. Think of it kind of as a debit account of tokens that people are accessing daily with usage, and we're tracking it very closely, and they know exactly where they are and how much they have left. We work together collaboratively with them to make sure that if they're running low, they'll replenish before they run out.
You pointed out there's a 2.8 times uplift in tokens versus history. You're basing your pricing on history. Are they underestimating by 2.8x? They're coming back at year one.
No. One mindset change. The numbers I talked about are subscriptions, not tokens. In an EBA, we measure monthly active users. That's our measure of subscribers within a named account that's on an EBA. It's disassociated with the token bank account, so to speak. That's actually why Andrew talked about what happens with ARPS in an EBA, and he actually said that if we do our job well during the term of a contract, the ARPS goes down.
Yep.
We have a set amount of tokens that drive the ARR value that's fixed for a particular year. If the customer is expanding their monthly active users, the denominator in that ARPS calculation is growing during the term, which is lowering the interim ARPS until the time that they add more tokens or they renew the contract. The numerator goes up by a large amount, and we end up getting an ARPS increase. The 2.8 was comparing the number of subscriptions a customer had before moving to an EBA to the number of subscriptions, which are monthly active users, once they've gotten into the EBA.
I think it's not underestimating pricing as much as the level of interest that customers have when they get the portfolio is much higher than what they expected. Much more of our portfolio gets used by many more users once they sign an EBA. We've seen that time and again, and that almost always drives the earlier token buy and then the later true-up that we see.
Right. That's the important part of the customer success function that's actually driving that effective use and adoption with many users across their enterprise.
Maybe one last follow-up. Can you share some color on what percentage of EBAs are buying more tokens and what's the sort of uplift? Or if you've come across a renewal, are you seeing 100% uplift in value? Or how should we think about it? We've seen it, like Aspen Technology does this model really well, and the inherent growth of that model is much stronger than just a regular subscription model, I would guess.
Sure. We'll have a lot more information going forward. T-Flex actually rolled out three years ago. We're actually just coming up to the renewals in that space. What we have been doing is moving customers from our older EBA model, which was the MultiFlex, which was not a consumption-based model, to Token Flex. There, we've seen substantial increases overall. We'll have more information for you, most likely this time next year, because we'll actually go through a series of renewals of our first round of T-Flex, and we'll be able to give you some statistics on growth rates on those renewals.
Dave, we probably have time for one more.
Yep.
Great. Steve Ashley. In terms of looking out at the guidance for FY20, asking about the assumptions of the mix of business relative to collections, have you assumed much of a change relative to what Suites were? When you just look at what Suites had represented, you look out, have you made any changes when you look out into the out years?
We anticipate Collections to basically subsume the same kind of runway Suites had. That's the working assumption.
Sure.
All right. That concludes our session. Those of you in attendance here, please stick around for lunch with our execs. Thanks.
Yep.