Good day, ladies and gentlemen, and welcome to the Autodesk third quarter fiscal 2017 earnings conference call. At this time, all participants are in a listen-only mode. Later, we will conduct a question-and-answer session. Instructions will be given at that time. If anyone should require assistance while the conference is in progress, please press Star then zero on your touch-tone telephone to reach an operator. As a reminder, this conference is being recorded. I would now like to hand the floor over to David Gennarelli, Senior Director, Investor Relations. Please go ahead.
Thanks, operator, and good afternoon. Thank you for joining our conference call to discuss the results of our third quarter of fiscal 2017. Also on the line is Carl Bass, our CEO, and Scott Herren, our CFO. Today's conference call is being broadcast live via webcast. In addition, a replay of the call will be available at autodesk.com/investor. As noted in our press release, we have published our prepared remarks on our website in advance of this call. Those remarks are intended to serve in place of extended formal comments. We will not repeat them on this call.
During the course of this conference call, we will make forward-looking statements regarding future events and the anticipated future performance of the company, such as our guidance for the fourth quarter and full year fiscal 2017, our long-term financial model guidance, the factors we use to estimate our guidance, including currency headwinds, our transition to new business models, ARPS, customer value, cost structure, our market opportunities and strategies, and trends for various products, geographies, and industries. We caution you that such statements reflect our best judgment based on factors currently known to us, and that actual events or results could differ materially.
Please refer to the documents we file from time to time with the SEC, specifically our Form 10-K for the fiscal year 2016, our Form 10-Q for the periods ended April 30 and July 31, 2016, and our current reports on Form 8-K, including the Form 8-K filed with today's press release and prepared remarks. Those documents contain and identify important risks and other factors that may cause our actual results to differ from those contained in our forward-looking statements. Forward-looking statements made during the call are made as of today. If this call is replayed or reviewed after today, the information presented during the call may not contain current or accurate information. Autodesk disclaims any obligation to update or revise any forward-looking statements.
We will provide guidance on today's call, will not provide further guidance or updates on our performance during the quarter unless we do so in a public forum. During the call, we will also discuss non-GAAP financial measures. These non-GAAP measures are not prepared in accordance with generally accepted accounting principles. A reconciliation of our GAAP and non-GAAP results is provided in today's press release, prepared remarks, and on the investor relations section of our website. We will quote a number of numeric or growth changes as we discuss our financial performance, unless otherwise noted, each such reference represents a year-on-year comparison. Now, I'd like to turn the call over to Carl.
Thanks, Dave. We're really pleased with our third quarter results. Q3 marked a significant milestone in our transition. It's the first quarter where we only sold subscriptions, and we're off to a great start. New model subscriptions grew by 168,000, new model ARR grew 91% at constant currency, and recurring revenue jumped to 76% of total revenue. More generally, we made progress on our two major initiatives, growing lifetime customer value by moving customers to the subscription model and increasing adoption of our cloud-based solutions. Given that this quarter was the most uncertain when we started the year, these are fantastic results. Let's look at the details. For the quarter, we added 134,000 net subscriptions.
New model subscription additions more than tripled year-over-year to 168,000, while maintenance subscriptions declined as expected by 34,000 during the quarter. Product subscriptions drove the vast majority of the new model sub additions. The launch of Industry Collections, the next generation of suites that include many of our cloud services, contributed to our strong growth this quarter. Collections are a great example of how we're simplifying our offerings while increasing lifetime customer value. During the quarter, we ran another promotion aimed at converting our legacy non-subscriber base to new product subscriptions. Like the promotion in Q1, legacy non-subscribers were offered product subscriptions at a discount when trading in their old perpetual licenses. The Q3 promotion added approximately 43,000 product subscriptions, significantly more than in Q1.
In addition, the overall average selling price was twice as high as the Q1 promotion due to a lower discount and the fact that customers favored higher priced products. Lastly, more than 50% of the subscribers turned in licenses that were seven years back or older. This reinforces our view that there are a meaningful number of active users whose licenses are more than five years old and are interested in moving to the latest software. While the promo successfully migrated many legacy customers to our subscription model, product subscriptions continue to attract a significant number of users that are new to Autodesk. Once again, new customers represented about a third of our new product subscriptions for the quarter. We believe some of these people were previously pirating the software and now have a much more affordable option with product subscriptions.
This is consistent with the fact that emerging countries are some of the fastest-growing areas for product subscriptions. In other cases, these new users have been using an alternative design tool and can now afford software from Autodesk. Overall product subscriptions grew by more than 30% quarter-over-quarter, with new subscribers coming from the legacy base, from people previously pirating the software, and from new customers switching from other systems. Particularly pleasing is that this strong performance immediately follows the end of sale of perpetual licenses for suites. We also had another record quarter for cloud subscription additions. They nearly doubled from the second quarter, which was also a record. BIM 360 and Fusion 360, once again, are leading the way, but we also have continued to have success with our other cloud products, such as AutoCAD 360, Shotgun, and Collaboration for Revit.
It is clear that we are building on our leadership in the cloud, and that cloud services are growing in importance with our customers who see the benefits of the cloud's inherent scale and connectivity. For example, during Q3, we signed a large BIM 360 contract with a global construction company. Last spring, they started a pilot program, less than 200 seats for new projects. Working with them on their implementation drove rapid adoption. Their results were so strong that they signed a seven-figure agreement expanding their use of BIM 360 Glue, Field, and Docs to additional projects, reaching over 2,000 subscriptions in just one region. They are planning to scale their BIM processes even further, eventually using BIM 360 for all projects in all regions. Maintenance is our other significant subscription bucket.
During the quarter, we included for the first time 13,000 maintenance customers from our acquisition of Solid Angle. As you already know, July 31st was the last time customers could add maintenance to a perpetual license, so the incentive to stay up to date with existing maintenance contracts is compelling. In Q3, the renewal rate for maintenance subscriptions was strong, ahead of last year. With no new maintenance agreements being sold, total maintenance subscriptions declined. As we've said before, we expect to see ongoing declines in maintenance subscriptions going forward. The rate of decline will vary based on the number of subscriptions that come up for renewal, the renewal rate, and our ability to incentivize maintenance customers to switch over to product subscription. This migration is good for both Autodesk and our customers as it moves them to our newest and best products.
While we are still working through the timeline, at some point, we will further simplify our offerings and the maintenance subscription and product subscription offerings will converge into a single program. Let me get into ARR. New model ARR growth surged to 91% on a constant currency basis and reflects the continued strong uptake of all of our new subscription offerings. Total ARR grew 15% at constant currency. ARR growth was negatively impacted this quarter by the allocation of our market development funds, which are accounted for as contra revenue and booked upfront in each quarter. Prior to Q3, these funds were allocated across all revenue streams. With the end of perpetual license sales, 100% of our market development funds are now allocated to recurring revenues in ARR. This impacted ARR growth by three percentage points, or about $40 million.
FX rates also continued to be a headwind for ARR in the quarter. The combination of FX rates and contra revenues reduced total ARR by about $100 million or seven percentage points in Q3, which is sizable. Despite those headwinds, we remain confident in our ability to achieve our FY 2020 ARR CAGR of 24%. I mentioned last quarter that some of our moves to drive subscriptions and ARR revolve around pricing. In early September, we made changes that rationalize the pricing of AutoCAD LT, AutoCAD, and the AutoCAD verticals. The aim was to drive customers to the right product for their needs and do so with higher billings and ARR. I'm pleased to report that these actions are motivating the intended customer behavior. More LT users are buying up to AutoCAD and the volumes for AutoCAD, and especially AutoCAD verticals, increased significantly quarter-over-quarter.
For the overall AutoCAD family of products, volume and ARR increased because of these price changes. We focus on subscriptions and ARR as the key metrics to measure our progress in the business model transition. However, we have received many investor questions on trends in annualized revenue per subscription or ARPS. I thought I'd spend a little extra time on it this quarter, and we'll spend more time on it at Investor Day next week. Let's start by focusing on new seats added in Q3, excluding the promo seats, as they have the temporary effect of depressing ARPS until their renewal, which will be at the full price. Suites and collections in the U.S. saw a year-over-year increase in ARPS of 24%. In Germany, the increase was +79%, and in Japan, it was +26%. AutoCAD LT in the U.S. saw a year-over-year increase of +9%.
In Germany, the increase was +55%, and in Japan, 50%. The AutoCAD family, which is the sum of vanilla AutoCAD and the vertical AutoCADs in the U.S., saw a year-over-year increase of 8%. In Germany, the increase was 13%, and in Japan, 78%. Because of the strong year-over-year changes for new seat purchases, the quarter-over-quarter ARPS change for the sum of AutoCAD Verticals, and AutoCAD LT in the U.S., Germany, and Japan were 5%, 35%, and 64% respectively. This was simply the result of more people buying vertical AutoCAD than vanilla AutoCAD, and more people buying AutoCAD instead of AutoCAD LT. Exactly the result we were looking for from the pricing change, and why we generated more volume in ARR from the AutoCAD family. These are very positive outcomes. In fact, except for a few emerging markets in Europe, we see essentially the same trend lines everywhere.
Why aren't these results reflected in the aggregate ARPS number we report? There are two primary reasons. First, the calculation is extremely sensitive to mix in the early stages. Small numbers exaggerate the effects of short-term shifts in term length, geo mix, promotions, et cetera, until it starts to normalize when the number of customers is large enough. Secondly, the method we use to count ARR in subscriptions is very sensitive to timing within the quarter. On an as-reported basis, ARPS can significantly understate the value of a subscription. The net-net of all of this is the fundamental business drivers are very positive, but aggregate ARPS for the company is declining because of various mix impacts, calculations around ARPS, FX, and the contra-revenue impact of market development funds. In fact, factors such as mix drove about 50% of the declining ARPS trend.
We expect these impacts to converge to a steady state. That's when you'll see aggregate ARPS begin to increase again. We will provide more insight into the components of the new model subscriptions as they mature into a bigger base next year. I know many of you are keenly interested in this area. We will spend a lot more time on this at next week's Investor Day. Total direct revenue increased as a proportion of total revenues to 29%, up from just 17% two years ago, and 25% last quarter. We continue to grow the volume of business with both our large enterprise customers as well as our eStore. On the expense side, we continue to prudently manage spend. Total non-GAAP spend decreased by 2% year-over-year. This marks the third consecutive quarter of decreases as we diligently control our spending.
We had already committed to keeping our FY 2018 spend flat with FY 2017. We're now committing to keep our FY 2019 spend flat as well. This will require some additional effort. We believe it's the right thing to do at this stage of our transition, and that we can do so without compromising the long-term health of the company. We remained aggressive with our stock buyback program in Q3, buying back 2 million shares at an average price of $68.74. We've now repurchased nearly 7 million shares this fiscal year and plan to remain active buyers through both programmatic and opportunistic means. This is another area we will discuss in greater depth at our Investor Day on December 6th. We were extremely pleased with our Q3 results.
We have increased confidence that the transition is working for customers, partners, and Autodesk. We're on track for the FY 2020 targets we set. Turning to our outlook. We've seen over the past several quarters, global conditions have held steady, with most of the mature markets performing relatively well while many of the emerging markets have been challenging. We evaluate our strong Q3 results and look ahead to Q4. We don't see any meaningful change in the demand environment. Talk of increased infrastructure spending and tax reform in the U.S. would likely benefit Autodesk. It's far too early to determine if the election results will have a material positive or negative impact on our business, domestically or abroad. Our view for the rest of fiscal 2017 remains positive. Our revenue and EPS results were slightly better than we expected in Q3.
Q4 typically experiences a sequential increase due to seasonality. We expect to see that again this year. Keep in mind that Q3 revenue benefited from $38 million in backlog from the end-of-sale of perpetual suites at the end of Q2. Since we've ended the sale of most perpetual licenses, backlog is now zero. It won't be a material contributor going forward in the subscription-only model. We feel great about the way subscriptions are trending. Now that Q3 is behind us, we are bringing up the bottom end of our previous range for subscription additions. We're on track for the spend target that we lowered last quarter. When we started down this transition path three years ago, the objective was the same as it is today: build long-term shareholder value and establish Autodesk as the leader of the next generation of design and engineering software.
We are accomplishing these goals in three simple ways. One, we're increasing the lifetime value of every customer by moving them to subscription models. Two, we're changing our cost structure by focusing our product portfolio and go-to-market strategies. Three, we're building the best cloud and mobile-based products and services in the industry, which significantly expands our TAM as the underlying technology platform shifts to the cloud. Next week at our Investor Day event, we'll provide further detail on significant items on our path to our transition goals, including $6 in free cash flow per share in FY 2020 and $11 in FY 2023. Some of the items we'll cover include an update on the transition overall, focus areas to achieve the 20% growth in subscriptions and 24% CAGR in ARR, the factors influencing ARPS, TAM expansion opportunities in construction and manufacturing, and success with our cloud offerings.
The evolution of our go-to-market strategies to align with the subscription business, recent operating changes that will increase our access to foreign cash and dramatically increase our financial flexibility, and our thoughts on capital allocation. To wrap things up, we couldn't be more excited to be finally in the subscription-only model. We've executed well over the past few quarters, and we're looking forward to finishing the year strong. Our vision and strategy are working. The business model transition is ahead of plan, and we're leading the industry in delivering cloud-based software for the next generation of design and engineering. With each quarter, we become more confident in our ability to increase shareholder value by creating a more predictable, recurring, and profitable business, and achieving the goals we set out for FY 2020 for subscriptions, ARR, and cash flow. Operator, we'd now like to open the call up for questions.
Thank you. Ladies and gentlemen, if you have a question at this time, please press star then the number 1 key on your telephone keypad. If your question has been answered, or if you'd like to remove your line from the queue, please press the pound key. To allow more of our audience the opportunity to ask a question today, you'll be limited to 1 question and 1 follow-up question. Thank you. Our first question for today comes from the line of Saket Kalia from Barclays.
Hi, guys. Thanks for taking my questions here. First, just sort of a housekeeping question for you, Scott. Can you just remind us what market development funds are from a high level, and what sort of dollar amount we should expect that to be on an annual basis?
Yes, sure, Saket. Market development funds are common to every company that sells through a value-added reseller channel. It's in effect demand generation dollars. It's money that we make available to our resellers for them to go off and drive demand, pre-sale demand for the product. It's not a new program. We've had market development funds for years. I think what's changed this quarter is the way those market development funds are allocated. They're not treated as an expense. They're treated as a contra revenue, meaning they're subtracted from our reported revenues. It runs $10 million-$11 million a quarter. Historically what's happened is that was allocated pro rata to the sales through the channel, which were heavily driven by license sales.
This quarter, of course, there's no license sales, all of the market development funds, same amount we've always spent, end up being allocated for the first time to recurring revenues. The $10 million-$11 million per quarter, when you annualize that, of course, is between $40 million and $45 million for the year. First time we've seen that effect on recurring revenue was this quarter. I would expect it to stay in that range going forward.
Understood. That's very helpful. For my follow-up, I'm sure this is going to be something that we talk about a lot more next week, how sensitive is the long-term model to mix? Meaning, can you just talk us through qualitatively maybe how different the mix is across the product ranges in this first quarter with full subscription, compared to what you're modeling in that $6 per share in free cash flow? Maybe just qualitatively, how much leeway do you have with that mix?
Well, let me start with this, Scott, feel free to jump in. We have a variety of mixes that we try to talk about. One was about certainly product mix. One is about geo mix. There's others like term length. There are very specific things, the difference between, for example, EBAs and maintenance. Those numbers look different as well as certainly the cloud revenues look different. Those things matter a lot. We have a fair amount of flexibility. We understand the price range that these things will come in. Right now, we're actually above the plan for where we expect it to be. One of the things I tried to do in the middle of the prepared remarks was give you a little bit of information about what selling on a like-for-like basis is.
Those were those quotes in the middle, all about how much more people are paying this year or this quarter compared to the previous year or the previous quarter. We thought it was important to really compare the like for like, so that people could understand it. One of the things we also mentioned is going forward, we'll start breaking this out for you. Having an average of averages of averages is way too confusing, as we've all seen. I think the folks at Goldman did a nice job at breaking out the model in their most recent report. I think they did a nice job of looking at the stuff we did. While it's not identically our model, it's close and certainly has the major effects in there and a good understanding of the kind of mix we anticipate seeing, and the price ranges.
Maybe that helps a little bit. We're going to spend a lot more time next week walking you through this, but hopefully that's a start on the problem.
Yes, Saket, the other thing that I would add to that is, I think mix is a bigger effect on ARPS right now than it will be longer term. You just need to look at the ARPS for maintenance, and you see it's far more steady because it's a much bigger base of maintenance subs. As the base grows, mix will become slightly less important. I think longer term, the more important metric than ARPS is ARR. One of the things we'll look to do is provide you better insight into where we think ARR is headed, so that the ARPS calculation becomes slightly less sensitive. Next question, operator. Karen, are you there? Okay, we seem to have lost the operator. Hang on one second while we try to reconnect to her and see if we can keep the Q&A moving.
This is a first for me. Of all the earnings calls I've done, I've never had this happen.
Sorry. No. I am here now. Our next question comes from the line of Steve Ashley from Robert W. Baird.
Okay.
Hi, thank you. Hi, thank you very much. I would just like to ask on the promotion, if you could give us some color on how much of that might have been direct versus indirect, and what role the eStore might have played in the performance there.
Yes, Steve, the promotion, Carl laid out the success of it, 43,000 subs that came through the promotion. It was almost entirely driven through the channel versus through the eStore.
Okay, perfect. I would just like to ask about collections, and I realize they have not been out very long, but with that, you introduced multi-user pricing, and that's kind of the first time that's been out there. I'm wondering if you've gotten any early response to that, or if there's any early color you can provide on that new pricing option.
It's still early. We're trying to replicate some of the things that we had in the previous model. It's only a quarter worth of data, we'll update it more as we go through next year. We know it is one of the things that customers really want. It's one of the important things. That whole middle bulk of our customer base has multiple seats, and while we offer lots of flexible offerings for people with large numbers of seats through the EBAs, people are also looking for it in here. We'll update you as we go.
Thank you. Our next question comes from the line of Heather Bellini from Goldman Sachs.
Hi, this is Sheetal filling in for Heather. Thanks for taking my question. First question was just on new model ARR. That increased by $43 million. I'm not sure if you specifically broke out what the MDF impact was for a specifically new model, first, could you break that out? Also, last quarter, new model ARR increased by $63 million. We're expecting it to increase in size again this quarter. Going forward, now that you're subscription only, should we expect new model ARR increases each quarter to get larger?
Yeah, Sheetal, the overall amount that ARR was affected by the allocation of MDF. Again, MDF is not a new program, the first time it's really been heavily allocated, almost exclusively allocated to recurring revenue is why we're seeing the impact so pronounced this quarter. It was about $40 million on the overall. If you split that between how much was allocated to new model versus maintenance, it's about six percentage points on a year-on-year basis, about six percentage points of growth for the MDF impact on new model, and about two percentage points of growth reduced by MDF being allocated to maintenance. Overall, it was three points, if you split it between the two, it was about six on new model and about two on maintenance.
Okay. I just had one follow-up on Industry Collections. Suites used to represent 30% of your revenues. I'm wondering, over time, do you expect Industry Collections to also end up representing 30% of total new model ARR? How long do you think it would take for that to ramp to that level?
My best guess is that it will be less because of the additional offerings we'll have with the cloud offerings and the consumption services tied to things like Industry Collections. I think it will be less, but not substantially so.
Thank you. Our next question comes from the line of Jay Vleeschhouwer .
Thanks. Good evening. Carl, I'd like to ask you my first question, a technology question stemming from Autodesk University a couple of weeks ago. Particularly if you could tie some of your technology disclosures from two weeks ago to the longer-term business implications. In other words, you had some very interesting things to say, I thought, about your future architecture with the Quantum Project you talked about. There were numerous references to what you talked about as a common data environment, new architecture, and so forth. Talk about the commercial implications and commercial plans for that as you showed two weeks ago, and what that means for the next number of years in terms of the business model, and then a selling follow-up question.
Yeah, sure, Jay. This goes back a number of years. We have a fervent belief that engineering software is going to move to the cloud. All design and engineering software will be in the cloud. We started to demonstrate that with a number of products. We've talked about some, like BIM 360. We've shown the basic architecture with products like Fusion 360 and how they take advantage of the connectivity and the compute power of the cloud. We started to demonstrate that, and we started investing in this area three or four years ago.
As I've said before, while everybody is rightfully focused on the business model transition of our existing business and how we're using it to attract new customers, what I think people will most be surprised about as you look out the next couple of years is the size of the cloud business that we built and how it really extends our TAM. We're going to spend a bunch of time next week showing a little bit more. You were at AU and got to see it. We'll try to put an investor look on it so that people next week can understand what we're doing and what the importance is and why we're investing in the cloud and why this really is the next-generation architecture.
I've talked in the past about it being a place for collaboration as well as it giving you access to virtually unlimited amounts of compute power, which is something our users demand. It's just a natural fit. It's unusual that engineering has been one of the slowest to move to the cloud, but we see lots of evidence of hitting the tipping point, not just in the U.S., but other places, in terms of customers' willingness to adopt it. Next week you'll hear a bunch more.
Thank you. Our next question comes from the line of Gregg Moskowitz from Cowen and Company.
Okay, thanks very much. Carl, thanks for all the additional information on ARPS during your prepared remarks. Having said that, given where you are today, you do have a lot of ground to make up to reach a 24% ARR CAGR by fiscal 2020. Can you elaborate on what gives you the confidence that you will get there?
Yeah. Part of it is the calculation method we chose gives great certainty and repeatability. It's very auditable, but it is a method that can greatly understate the amount of revenue coming in. One of the things we do, and we'll walk you through this example, but what we actually do is we take the revenue that comes in for the quarter, the recognized revenue, which if it's particularly back-end loaded, can be close to zero. We divide it by the number of subscribers at the end of the quarter. In the extreme, I'll use one example that'll illustrate two points. The promotions were very heavily back-end loaded. You have a low-priced offering that's discounted coming in very close to the end.
What you might get out of that is one day out of 91 worth of revenue, or call it four days out of 365, or slightly less than 1% of the revenue that that subscription is worth, we divide it by a whole subscriber. It greatly understates the value. As we go through this, it will normalize, as Scott mentioned, because the numbers get larger, and you'll get the same effect moving from quarter to quarter. In the beginning, it's very sensitive to mix. It's very sensitive to timing. We're ahead of our plan from where we want to be. While I understand that some people are paying lots of attention to ARPS, I think the better thing to look at going forward, and we've tried to stress it, is ARR. As Scott mentioned, we'll start giving you guidance on ARR.
On the ARPS one, we'll start breaking it out so you don't have to be guessing at the coefficients between what's in cloud, what's in EBA, what's in product subscriptions, and even in the subset that's something like the legacy promotion. We think we can help you understand that a little better. I'll tell you again, one of the reasons we gave the numbers in the middle of the prepared remarks about like for like was to show you the increase in pricing that we were seeing. There were some numbers that were quarter-over-quarter and some were year-over-year, and we'll detail that a little bit more next week at Investor Day so you can see the thing. What we think is the like for like is the best comparison.
Okay, perfect. Thanks for that detail, and agree that will be helpful, so look forward to that. Just one for Scott. Any update on the timing to potentially change the segment classifications on product subscriptions and EBAs from a hybrid approach to 100% subscription on the income statement?
Yeah, Greg, fast isn't fast enough on that. I would love to have that done now. I think I mentioned on this call last quarter, the complication with doing it is not the revenue side, it's the COGS side. We're working that very carefully, so the goal will be to end up with three revenue lines instead of the two that we have today, because I recognize the confusion it causes. The three lines would be one for maintenance revenue, one for subscription revenue, which ties to new model, and one for license and other, because there will still be a small amount of license revenue that comes in. We'll have to go through, of course, and carve up deferred revenue and then new revenue going forward. My goal is to get that done as quickly as possible. I'd look to roll that out next year.
Thank you. Our next question comes from the line of Keith Weiss from Morgan Stanley.
Hi, guys. Good afternoon. This is actually Stan Zloski sitting in for Keith. Thank you for taking our question. How much of the churn on the maintenance subscription base was driven from maintenance subscribers moving over to desktop subscribers? A quick follow-up. What are you thinking with the latest on pricing for the maintenance and enticing those customers to move over to become full product subscribers? Thank you.
On the first part, we don't know exactly. We use an estimation method that's fairly manual in order to do this, we'll try to explain this a little bit next week, but we don't actually know exactly how many except through a manual process. We don't automate it because that will make less and less sense going forward. Well, I think, Stan, the second question was about convergence of the programs?
Yeah, essentially.
Is that correct?
Are you driving pricing increases to entice people to move over from maintenance, or are they doing it voluntarily through seeing the greater value proposition of those products, or is there a third option that I'm
Yeah
we're not thinking about?
I think both right now and an ongoing basis, it's going to be a combo of the two. There'll be some incentives to move people forward who choose to go early. A lot of it is the value in the new offering, which is the latest and includes large aspects of the cloud in it, and gives users additional flexibility. We think, there'll be a little bit of a push and a little bit of a pull.
Perfect. Thank you very much.
Thank you. Our next question comes to the line of Richard Davis from Canaccord.
Hey, thanks very much. If I'm a firm that's decided to switch to the cloud, to what extent should we be thinking about, because this happened with Oracle to some degree, is there any risk that I would consider other cloud options? I mean, obviously, yes, but have you seen any kind of data that would say that's an issue, or is it more just like, "Listen, we're going to switch, and when we switch, I have all my models in Autodesk, so I'm just going to go to your new model." Thanks.
Yeah. On the cloud stuff, Richard, in terms of having mature, viable cloud products for both the AEC industry and manufacturing, we're really the only ones that have it. You can look out and see some small startups, but barely have any traction. We probably have three orders of magnitude, maybe four orders of magnitude, more users in the cloud than they do. I think for people who are deciding that the cloud has real benefits, I think Autodesk is the natural choice. We've detailed a little bit at previous times where we said more than half, and sometimes it was three-quarters of the users of Fusion are coming from other mechanical CAD packages. It's not just our users who are adding on cloud capabilities. It's really to attract new users.
The second part of it is, one of the things we did when we introduced our cloud products is we made sure they weren't only direct competitors with what we already did, but it was to extend the market. If you look at BIM 360, BIM 360 wasn't a replacement for Revit. It was to extend having model-based design go all the way out to the job site and reach all the construction workers. Those are the kind of things we're trying to do, is expand the reach of our products because of the more natural platform. I think you've seen the same thing in other parts of enterprise software with SaaS offerings, where people are expanding into parts of the company that their legacy applications would've never gone to. I think it's identical in the design and engineering world.
Got it. Great. Thank you very much.
Thank you. Our next question comes from the line of Sterling Auty from JP Morgan.
Yeah, thanks. Hi, guys. Carl, in your prepared remarks, you did touch upon the infrastructure possibility in terms of the new administration. Can you level set for us if you were to just give us a rough estimate, what % of the Autodesk business at this point is infrastructure related, and how should we think about that within the U.S.?
Just the construction or the broadly speaking AEC business, think of it as roughly around 50%, then you can take the fraction that's the U.S. part of that, and go from there. Of course, Mexico is going to be building the wall, so I guess we can extend to the rest of North America. Yeah. That's about the reach of it. The American part, which is about a quarter of about 50%, something like that, I'd put it in the 10%-15% range.
Okay, fair enough. Then can you give us an update? You touched upon the eStore a couple of times. At the analyst day last year, you talked about a lot of the improvements that you were making in North America, the initiative to move international with the eStore. Where are you on those initiatives, and are you at a point yet where you can start to break out how much of the revenue is actually coming through the eStore?
Yes, Sterling. I tell you what, Carl, I'll start, and you can jump in.
Okay. Go ahead.
We have, of course, eStores everywhere. We've launched our own eStores now across North America and in most of Europe. The eStore continues to grow. We talked about the proportion of our sales that were direct versus indirect grew to the high 20s this quarter. I think 29% up sequentially from 25% just last quarter and 17% a year ago. Continues to grow as a proportion of total sales. I think we'll continue to see that happen, not so much because we're guiding people there. When you get to a subscription model, buying direct is a style of interaction that customers want to have with the company much more so than when I renew my subscription to the Wall Street Journal, I don't walk down to a bookstore to do that.
I go to wsj.com and do that renewal, and I think we're going to see more of that business move that direction.
Thank you. Our next question comes from the line of Monika Garg from Pacific Crest Securities.
Hi. Thanks for taking my question. How much revenue in the quarter came from indirect channel, and where do you think channels could be as a % of revenue over the next couple of years?
Yeah. Monika, that's the data point I was just trying to give. The direct channel in the quarter was 29% of sales, and up from 25% last quarter. Where that goes longer term, it actually is a topic that we'll talk about again next week, but I don't think our view has changed much from the perspective that we had a year ago, which was as we get out to FY 2020, we think the channel continues to be a significant part of our sales. In terms of absolute dollars, we think there are more dollars of sales going through the channel in FY 2020 than there were in FY 2016. In terms of a % of pie, the % of the total sales that get done, it'll be a smaller % by the time we get out there.
I think we'll see a case where the channel has a place where absolute dollars of revenue grow, by the way, probably fewer channel partners out there than we have today, so it's a more profitable business for the channel partners that remain. As a % of total sales, it will come down pretty significantly from where it is.
All right. Thanks. I have a question on ARPS. I know Carl talked about a lot of questions on that already on the call. If I look Q over Q, ARPS for new model are down almost 10% in spite of the fact that collections were sold on subscriptions, which is a higher revenue per seat items. Why would that be the case? Seven straight quarters of ARPS for new model being down.
Well, generally speaking, we've continued to add lower-priced products as we've rolled this out. We started with a lot of EBA. We started with a lot of transactions going through the eStore. We also went through, when you have back-end loaded quarters, you greatly understate the value of that subscription. It's why during the thing I try to break out how the prices were increasing for like-for-like products going forward. I think that's going to be an important one. What you're getting is you're trying to look at an average and guess the component pieces. Rather than have you guess what we said we would do next year is break it out so that you can actually see.
In the interim, what we did is we gave you some numbers for showing that the prices are actually increasing, even if the calculation methodology in the mix is heading it down. As we said, some of the mix issues were responsible for 50% of the decline. Mix and timing play a big role here. Remember how I just described that calculation, that if you buy something at the end of the quarter, for example, on the last day, it greatly undervalues what that subscription is worth over the long term. In a quarter in which we had a fair amount of promotion, and the promotion was back-end loaded, you're going to see that.
I'll reemphasize again, the prices of like for like, so people buying the same thing that they bought last quarter or a year ago, those prices are going up significantly, and I highlighted how they were going up significantly across the world.
Yeah, Monika, what I'd add, and again, we'll spend more time on this next week. If you go back to new model subscription sales a year ago, it was a small number. They were largely in mature markets, a healthy mix of monthly and annual, as well as many of them came through the eStore. The channel was still largely focused on selling perpetual licenses at that point. What we've seen over the last several quarters is a pretty significant uptake of new model subscription sales, and I'm talking about product subscription sales now. Pretty significant uptake of those in emerging markets, which I think is healthy for the business, and certainly I would expect some of those to be recapture of pirates in those emerging markets. As that happens, of course, those are lower priced. That's a headwind on ARPS.
We see a lot more long-term sales of product and a lot less monthly. Frankly, we had a lot of toe dippers that were coming in monthly previously. We now have a lot more annual and multi-year. Again, good for the business, but the monthly subscription price is about 50% higher than the annual subscription price. The monthly times 12 is about 50% more than annual. As that mix shifts, again, good for the business, not necessarily good for ARPS. It's a headwind. As the channel has gotten really engaged and is driving this, a higher proportion of the sales of product subs come through the channel than did a year ago, where many of them were being sold off the eStore. Again, I think that's a positive for the business. It gives a scale, but it cuts into ARPS.
Three trends that are very positive for our business and positive for the long-term ARR growth, but all that cut against growing ARPS in the short term.
Thank you. Our next question comes from the line of Ken Wong from Citigroup.
Hi, thanks for taking my question. Carl, you mentioned the convergence of maintenance and rentals earlier, and I think you might have been heading down that path on one of the earlier questions, but could you give us some color on some of the moving pieces of that particular convergence?
Yeah. There are a number of things that we've talked about. One is, if you look out to kind of call it FY 2020, we want to be in a place where, first of all, we have really a single kind of offering with a single back office and infrastructure to support it. One that will be a combination of subscriptions, product subscriptions as you see them, plus a consumption model on top of it. That's where we see the business headed. Along the way, it's how do we motivate customers to move from one model to another? What's actually in the program? What are the price points? How does that transition work? We'll talk a little bit more about this next week. In our mind, getting to a single model is really important. It will give the best service to our customers.
It will be the most affordable for us to have. We can start getting rid of some of the systems that were designed for a different era, and be able to really concentrate on giving a world-class experience for what users are trying to do today. As we go through the next couple of years, we'll talk about how we do that and how we do that in the most prudent and responsible way.
Yep. Got it. Looking forward to next week. Building on Stan's question around maintenance and just renewals, any rough sense of the magnitude of renewals that are coming up in Q4 relative to Q3?
Yeah.
Sure, Ken. Renewals tend to be on the anniversary of when the license was sold. Of course, Q4 has historically been our biggest quarter. We've sold the most perpetual licenses over, if you look back over the last several years, we've sold the most in Q4. We have a significant renewal base that comes up during Q4. I would go back, though, and highlight what Carl mentioned in the script. Renewal rate is actually stronger this year than it was last year. Yes, we're seeing a decline in the absolute numbers, but it's not driven by a decline in renewal rate. It's driven by the rate at which the maintenance agreements come up for renewal. In some cases, we're incentivizing them and successfully moving them over to new model.
Yeah. We'll also see an effect in Q4 from people moving to EBAs. As we've talked about, we are really encouraging our large customers to move to EBAs. There's an issue in which we don't count the subscribers until the following quarter. You might have seen this in Q1 of this year. The same phenomenon will go on. Particularly if someone goes from being a large maintenance customer to an EBA customer, you'll see some of that tendency to depress the number of maintenance customers, and you'll see the corresponding increase, but you won't see it till Q1.
Thank you. Our next question comes to the line of Steve Koenig from Wedbush Securities.
All right. Thanks, gentlemen. I apologize if this question's been asked. Hopefully, it hasn't. I'm wondering if you could elaborate a little bit on your expectation for subscription counts which implies acceleration in the out years. Maybe if you could give us a little color on, or a ranking of the factors that will drive that, when you think about things like piracy and cloud, et cetera, what's going to drive that acceleration that you guys are looking for, and to what extent are you starting to see those things kick in?
Yeah. In the kind of existing business, what we're really looking for is the three places we saw it this time. One was legacy, one is piracy, and one is new customers. I think you've seen the effectiveness. We've been able to detail the effectiveness of the legacy programs very specifically. Piracy is a little bit harder. Unlike legacy, where people claim to be a legacy customer, people don't claim to be former pirates, so it makes it a little bit more difficult to identify them as former pirates. We do have our ways.
Right.
I expect to see more from that. Like we said, about a third of the customers coming are new customers. In terms of the other things that we see is growth amongst our enterprise customers in terms of EBAs and the size of those, and then the whole new category of cloud customers, and consumption. You'll both have cloud customers, and then you'll have consumption models where people kind of pay as they go for various services. Many of them will be tied directly to cloud products, but many will be consumption baked into things like the collections. For example, if someone's doing simulation or rendering in a collection, they'll get some amount with it, and if they want to use extra, they'll pay as they go. That's where we're expecting to see that.
Got it. Great. I'll leave it at that. Thanks, Carl.
Okay. Yeah. Next week, you'll see a bunch more about it. Right now, our cloud business is growing incredibly quickly. We're really pleased with the growth we're seeing in the cloud business. We'll talk more about which customers we're getting with the cloud business and why that expands the margin.
Great. Thank you.
Our next question comes from Brent Thill from UBS.
Thanks. Good afternoon. Carl, you mentioned a third of the subscribers are now coming from new customers. I'm curious what's maybe most surprising to you, what you've seen in terms of switchers from the competition, greenfield, old users, pirates. How do you put the community together and highlight where you're seeing the strongest number of new users coming into the platform?
Yeah. I'd probably put it probably in that category, I'd have to guess. Like I said, Brent, right now we're doing a bunch of manual work to figure out, of the new customers, how many we've never seen before versus those who pirated the software. We're trying to understand that dynamic a little bit more, and we'll talk a little bit more about that. As a rough guess, one of the ways to think about it, we've talked about this before, for every legitimate user out there's about one to slightly more than one user who's not paying for the software. My guess would be, as we made some of these offerings more affordable from an upfront perspective, we're attracting a fair number of the pirates. The piracy base is probably bigger than we originally anticipated last year when we laid out these goals.
The second thing is the legacy base is probably bigger than we laid out last year at this time. We saw it in terms of the distribution throughout the years. Again, the midpoint is around seven years. We're seeing a fair number of people who are historical users. On both those fronts, we're kind of pleased. The greenfield is a little harder to identify for certain. The other ones we have a better chance at. We'll give you a little bit more color on this as we go, we get the numbers. Like I said, at a certain point, it won't actually matter, but we are looking at both piracy as well as maintenance customers who are moving over.
Just a quick follow-up. You've been talking a lot about the make process. You can build it, you got to make it. You've, I think, mentioned in the past it hasn't been maybe as big a piece of revenue, could be a lot bigger in the business. Can you maybe just update us in terms of what you're seeing on that core area?
For example, we're seeing it in a number of areas. In the AEC area, particularly with products like BIM 360. BIM 360 really is communicating with all of our non-traditional customers involved in the process. If things like Revit in our structural analysis software for the architects and engineers, BIM 360 encompasses the entire workflow, all the way from architect to general contractor and subcontractors. Next week, Amar will go through a bunch of it, a bunch of what we're doing there and how it's reaching the people who are actually on the construction site building stuff. When you look at the other things like manufacturing, Fusion 360 as well as all of our advanced manufacturing products are really very specifically about the next part of the process.
I've kind of joked that we've helped people make beautiful digital prototypes, things that stay on that side of the glass, our customers get paid for making real things. What we started doing a number of years ago is starting helping them in that process of taking it from one side of the screen to the other side of the screen. We have lots of stuff going on in composite materials, a lot we have going on with machining, communicating on the factory floor as well as the construction site. Those are all expansions of the market, we'll spend a fair amount of time. We've generally spent most of our time, and probably appropriately so, as people have been very interested in the business model transition.
Like I said, this is a growing and very important part of the business, and as people look to the next couple of years, it'll become increasingly important. We'll give you some more detail, Brent.
Thank you. That concludes our question and answer session for today. I would like to turn the conference back over to Autodesk management for any closing comments.
Thanks, Karen. That will conclude our call today. As Carl mentioned, we do have our investor day next week, next Tuesday the 6th. It's going to be at our gallery in San Francisco. After that, we'll be at the Barclays Conference in San Francisco on December 8th. If you need to reach me, you can reach me at 415-507-6033 or email me with any questions. Thank you.
Thank you. Ladies and gentlemen, thank you for your participation in today's conference. This does conclude the program, and you may now disconnect. Everyone, have a great day.