All right. Good afternoon, everybody. Thanks for joining day two of Citi's Tech Conference. My name's Tyler Radke. I co-head the U.S. software sector here at Citi. This afternoon, we're happy to have MongoDB. We have Michael Gordon, the COO and CFO of MongoDB, as well as Serge Tanjga, who is the Senior Vice President of Finance and Investor Relations. I think I got that title right.
Awesome.
But, um-
Thanks for having us.
Yeah. Appreciate you, appreciate you coming. You probably have the shortest commute to the conference out of anybody.
Yes.
Um-
Love a good home game.
It'd be great to just give a brief overview of MongoDB. It, you know, certainly in the database landscape, where exactly do you fit in this big market?
Sure, yeah. Again, thanks for having us. Great to see everyone. Just to take a step back, so the database market is one of the largest in all of software. A little over $80 billion spent per IDC in 2023, with that growing to around $136 billion in 2027. So really a quite large market and growing at a fairly healthy clip. You know, one of the ways I try to think about this is, typically, I would think of the market as sort of a more mature market, and I would expect a market like that to grow maybe more in line with global GDP or something like that.
The reason why, you know, the, the market's forecast to grow, you know, $10 billion-$12 billion a year, right, sort of, you know, roughly 13% per the IDC numbers, is because databases are central to companies' ability to compete and innovate. You hear these phrases like, "Software is eating the world," or, "Every company becoming a technology company," or maybe even Citibank talking about having more developers than some big, large, you know, tech company. The reason for that is because companies are, are trying to drive competitive advantage through their internally built software. Packaged software is great, excuse me, but if it's off-the-shelf packaged software, anyone can buy it. That doesn't really convey any competitive advantage, and so I derive my competitive advantage from actually building software.
Every application that I build has a database at its center, and the agility, scalability, nimbleness of the database will determine ultimately your competitiveness. So that's why it's such not only a large market, but such a strategic market, and we're going after all the markets. MongoDB is the most popular and the leading modern general-purpose database, so we can talk and get into as much of the-
Yeah
... technology, you know, as people want to, but databases have been around for a long time. The sort of historic legacy technology is relational technology, with companies like Oracle being great examples of people who've, you know, really successfully captured the market. There are a lot of challenges with relational technology. It was built for a different era, kind of pioneered in the 1970s. Obviously, things have changed. Data volumes, data scales, data types have changed. People's need to innovate more quickly has happened. And that's really where our company and our product has come through with sort of a better mousetrap. And so the reason that we've been successful at gaining share is not just because of the better product, but that product resonates with developers. Developers are the ones who are building the applications.
And so MongoDB, you know, has kind of the hearts and minds of developers, and that's really enabled our success. All that, you know, combined with good execution, has us now. You know, the company was founded in 2007, so 16 years after founding, at around $1.5 billion revenue, kind of closing in on 2% market share, but with big ambitions. So I'll stop there.
Awesome.
We go a lot of different ways.
Yeah. Yeah. No, great, great overview. So I think one of the numbers that stood out, certainly from your Analyst Day, was, you know, the $100 billion TAM, I think, by IDC's definition, not necessarily your definition. But the question I often get from investors is: How addressable, how much of that market can you actually address? I mean, number one, some might argue the non-relational or NoSQL part of the market is a lot smaller. And then secondly, a lot of that market is almost, like, locked up in existing applications. Obviously, Oracle, IBM's, Microsoft's, et cetera, have a large business.
Mm.
So how do you kind of think about your ability to address that market today, and also in the context of some of the products that you have in the pipeline?
Yeah. Yeah, so there are a couple different ways to go at it. So I think first what I'd do is I would start off technologically. We've made significant investments in the product over the last several years, such that from a technological standpoint, there isn't actually a reason why technically you couldn't serve a specific use case. That doesn't mean it's the right thing, though, right? And so if we just go back to those IDC numbers, there's anywhere, obviously it varies a little bit year to year, but there's anywhere between kind of $10 billion-$12 billion of new created every year, new spend. Most of those are new applications. We're obviously particularly well positioned for new applications, you know, as the modern leader.
If we think about that $80 billion, you know, today, to your point, not every dollar of that $80 billion is doing an RFP every year, right?
Mm.
Like, if you have applications that are working perfectly well, like, you're not going to bother to go replatform them. So if we just use a somewhat simplistic assumption of saying a 10-year average application life, you know, that would mean that $8 billion of the $80 billion is kind of up for grabs, you know, in any given year. And so if you add that to the $10 billion-$12 billion of, of growth, you've got anywhere between kind of $18 billion and $20 billion transactable, you know, in a given year. And so that's kind of how I'd frame it up.
I think the other thing that I would call out is, while sometimes analysts will think about, not necessarily research analysts, but industry analysts will sometimes think about, you know, the relational market as separate from the non-relational market, or sometimes called the NoSQL market. I don't think that accurately captures the landscape today. I think that was helpful at a point in time when you had, some insurgent technologies, you know, including MongoDB, that were sort of trying for that, but served different use cases. But our general purpose nature allows us to compete directly. So we have, plenty of Oracle displacements, of relational displacements.
Mm-hmm.
That construct, while maybe providing a helpful organizing framework, doesn't quite match the market realities or the technical capabilities. From sort of a TAM and a market and a use case standpoint, we see the whole thing in play. We see use cases across industries, across geographies, across company sizes, customer-facing, systems of record, Wall Street trading applications, utility building applications, kind of hardcore mission-critical stuff, which wasn't true 10 years ago, right? We've evolved and we've invested a lot in the product, but that's kind of how I frame it.
Yeah. Awesome. So in terms of that incumbent spend in a lot of these legacy applications, you know, I think historically, Mongo has talked about a quarter of your business coming from, you know, rewriting or replatforming these legacy applications. As you think about generative AI and some of the code completion, code assistant technologies, especially combined with some of your own tools, like Relational Migrator, how do you think about the ability for legacy applications to kind of get remodernized more quickly-
Yeah
... especially as the tools become easier?
Yeah. So I think it's helpful to take a step back and just sort of think through these things. So first of all, if you have an application that is working just fine, no one, no CTO of Citi or anyone else wakes up and says, like, "I want to go replatform an application," right? You'd always rather build new functionality, improve user experience. And so what usually tends to trigger the need or desire to rewrite an application is you're seeing a lack of performance, right? You're seeing a slowdown, you're seeing a lack of agility, a slowness in your ability to innovate, and the application just isn't performing, right?
So then you say, "Okay, I have to go bite the bullet and go do these things." What are the key phases, I think, of migrating, particularly from relational? It's probably helpful to think through, then we can talk about where does Generative AI help in that. So the first phase is in the schema modeling and mapping your relational schema to the document model. Without getting deeply technical, the document model is much simpler, but because relational has been the standard, it is typical for people to think in these very complicated tables of rows and columns, and I think and I tend to orient my data that way, and it helps demystify and familiarize someone with the document model to be able to visualize what those schemas look like.
And to your point, we have a tool that we just released, GA, called Relational Migrator, that helps with that. The second phase can really be in the data migration, and we also have tooling that helps with that. Often, you'll run them in parallel to make sure the systems are correct. And then the third piece is the actual writing of the code, right? And that's where I think generative AI can help. That's a little bit more sort of on the horizon, and we haven't cracked the code there yet, but we're working on tooling and working with partners around that.
But I, I think that could be a real facilitator for some of these, because if you think about sort of the economic equation from a business owner standpoint, "I have pain, I need to go fix that," right? Replatforming to MongoDB takes work, right? That's real effort. And so anything that we can do to shrink down that cost will effectively improve the ROI equation or make it sensible for a higher percentage of those to actually be rewritten on MongoDB. If you're simply just looking for cost savings and your application is performing just fine, it will be faster to move to another relational technology.
Yeah.
Right? Because that's more kind of like kind to like kind. But as we can bring down sort of the cost, people will also get the application benefits of moving to our platform.
Got it. Sticking on the generative AI subject, just for a couple more questions, as I think about the other side, which is the new business side, your new customer additions have been, you know, hovering around record high levels for a number of quarters, despite the challenging macro. And I think you've talked about a number of generative AI customers who are using MongoDB. I think, you know, Hugging Face and some of these other well-funded AI start-up's are using MongoDB in some form. I guess two questions: Can you talk about the use cases, what type of workloads those GenAI companies are using MongoDB, and how do you just think about the significance of that growth opportunity over time?
Yeah. So, why don't we tackle this a couple different ways? So I think one thing that will happen, especially with these code assist tools, is I think it is safe to assume that we will see increases in developer productivity. People can debate today, and it's hard to, you know, construct a perfect A/B test to figure out, you know, what is the percentage. We've certainly heard estimates of 20, 30, 40%. But one way or another, I think we can comfortably assume that, you know, over time, and once you work out the hallucinations and everything else, we will see improvements in developer productivity. What that will mean is that developers will create more applications.
From what we've seen and all the customers we've talked to, there isn't anyone who's sitting here saying, "Oh, great, I'm going to get this improvement, therefore I'm going to cut my engineers." Most people have a robust and healthy and high ROI backlog of things that they can't get to.
Mm-hmm.
And so with greater developer productivity, they'll be able to crank out more applications. So first of all, we'll just benefit from that, right, by having more applications. Secondly, to the extent that they're sort of AI or AI-powered or have more AI capabilities, they will want and need the benefits of a modern general purpose database like what we have. And so I think that positions us, as well there.
Okay. And I guess the generative AI companies themselves, what type of use cases are you seeing?
Yeah
with MongoDB?
Sorry. Yeah, yeah, yeah. So you'll see it in a couple different ways. I think we're relatively early on, and we've certainly talked about a couple specific ones, but I'll try and sort of generalize.
Yeah.
There are plenty of companies... Yeah, we'd called up Hugging Face, I think, well before-... ChatGPT was launched as a customer, right? And so, it's not just all generative AI, right? But there's a broader field of AI, and these are companies that are sort of building the core of their business on MongoDB. Vector search is a popular topic, and maybe we want to get more to it in detail-
Mm-hmm.
But just conceptually, that's a newer area. We've recently introduced a product that's in preview mode, where people are incrementally building on top of that and sort of adding... basically incorporating the vector capabilities that help with some of the semantic search into the database platform, where we think that it'll eventually, you know, evolve over time, but we can talk about that as well.
Yeah. Yeah. Well, yeah, that's a great segue. I did want to touch on some of the new products you announced. You hosted a number of these events around the U.S., including one here in New York, MongoDB.local. You announced a vector search offering. You also announced a stream processing offering, which I think that one particularly caused was a bit of a surprise, but could you just walk through kind of your vision around those new products? What are the types of use cases or workloads that you saw a need for that in the market?
Yeah. So I'll, I'll try and run through a few of the things that, again, we can go into as much detail as, as you want to-
Yeah
... or is interesting to the crowd.
Yeah.
But why don't I take a step back and kind of paint the picture, right? So, first and foremost, what we're trying to do is respond to our customers and what developers want, and, and at the core, what we're solving for are developers' problems. And developers have a lot of different challenges, including, especially when you get to managing multiple or disparate systems or point solutions. And just simplistically, MongoDB offers that core operational, transactional, kind of OLTP database, which is at the heart of the application, powers the transactions and everything else. And any other point solutions that I add around it, not only do I have to manage the point solution, but I also have to manage some kind of syncing or communication infrastructure between the two.
And so the more things that can be brought into the platform, the easier the developer's life is, the greater the productivity will be, and the more value that they can create, create, the more they can focus on functionality and, and user experience. So that's kind of like the macro backdrop-
Mm-hmm
... from a developer and a customer standpoint. From a business standpoint, it's less about, "Oh, we can introduce this functionality, therefore, we can go address this adjacent TAM.
Yeah.
Right? We're blessed with an enormous TAM. As I mentioned, we're closing in on 2% market share, so we're not in search of additional TAMs. But really, what a platform story allows for is for us to win more workloads and to accelerate the winning of more workloads within customers, and to penetrate an account like Citi more completely, right, more effectively, more efficiently and faster by being able to offer a platform story that resonates. And so we can talk about text search, which has been out for a while. We just introduced vector search. We just announced in private preview stream processing.
Mm-hmm.
All different pieces that are around, in addition to time series, and other capabilities that we have. So why don't I stop there-
Yeah
... but happy to go in any of the directions.
Yeah, and-
At some point, I'll cut you in.
Yeah. And I know that we probably don't have time to do a technical deep dive into all those product areas, but I guess keeping it kind of high level, should investors think about those new products as incremental to the TAM numbers that you've already articulated? Or how do you-
Yeah
... kind of see that?
Yeah, theoretically, yes, they're incremental to the TAM numbers. That's not the primary driver from our standpoint.
Yeah.
For us, it's more about winning new workloads. I think that it will be more effective from an economic standpoint in terms of driving more new workloads. Like, yes, you could take an existing set of applications and say, "Okay, let's go attach some of these capabilities to that." And yes, theoretically, the spend per revenue, you know, per workload would go up as you had that kind of attach rate. But I would caution people and help people understand that this is a little bit different than sort of classic feature enablement, right? There are often other technological solutions, these point solutions, that people are managing, and therefore, it requires real sales effort if you're gonna say, "Okay, you used to be, you had an application, your application had text search," right?
You think about a website you go to where there's sort of filtering. You want to filter by size or, you know, geography-
Mm-hmm
... or, you know, whatever brand, whatever it is that you're filtering. That's search in action in an application. And, you know, you're doing that today. You have some product there. You have Solr, you have Elastic, you have something you're doing. And so it's not as simple as having some customer success manager say, "Hey, would you like to enable, you know, search? Great, that's now enabled." And all of a sudden, you know, revenue is up X% on that application. You've actually got to go displace the technology. There's a clear value proposition in having that integrated into the platform, but it does take real work.
And so, while I think there's uplift there, I think the big focus is on helping, having a platform story allow us to win more workloads more quickly, and therefore, kind of capitalize on a market opportunity faster.
Okay, that makes sense. So maybe pivoting to the recent trends in the business. You recently reported last week, I guess it was only a week ago. It feels like a long week.
A long week. Been a short week.
Yeah, exactly. But certainly really, really strong results, you know, across the board, whether you look at Atlas or the non-Atlas business. But I guess sticking on just the momentum that you're seeing, how do you feel like the macro environment has kind of evolved over the last few quarters? Do you feel like we're starting to... we're through the worst of kind of these macro headwinds, or... How would you just kind of characterize, you know, both the demand and consumption trends that you're seeing from customers?
Yeah, why don't I take a crack? So I think it's a good way you phrased the question because we tend to think of demand and consumption as, as sort of two separate drivers. So you've heard us talk about the new business environment, and that is basically ability to win new workloads, whether it is the first workload in a customer, i.e., a new logo or just subsequent workloads, whether it is EA or Atlas, we're, we're indifferent. And really, you know, throughout this macro slowdown that began over a year ago, we've been very happy with our ability to acquire new business, and I think that ultimately speaks to a few things.
It speaks to the size of the market, as Michael mentioned, it speaks to the quality of the product, and it speaks to our ability to execute and sort of keep laser focused on acquiring new workloads is kind of the currency that matters internally, right? So, and we haven't seen a change in that, right? Like, we've been able to execute through this macro slowdown, and obviously, we listen to what other software companies are saying, and I know that that hasn't been a universal experience for everybody. But whether it's, you know, the mission criticality of our product, the quality, sorry, the mission criticality of the market, the quality of the product, or the execution, but the new business environment has remained robust for us, and that's been sort of steady as she goes.
Don't take it for granted, but have been generally pleased with how that's been going. Now let's switch to consumption. So this is obviously an Atlas phenomenon. What we've seen very clearly, and we're among the first ones to call out, was at the beginning of our second quarter last year, we started seeing underlying usage of the apps on our platform slow down. And let me just pause and double-click on that because that's very important. So on Atlas, we see in real time what our customers are doing, and not just what they're spending with us, but also, how they're interacting with their databases, and that's what we mean by usage. And what we saw was the underlying apps grow at a slower pace, and that's very highly correlated with our growth and how much money they spend with us.
So we called that out when we saw it at the beginning of the second quarter. We sort of adjusted our guide accordingly. If you fast-forward, you know, five quarters later and where we are today, generally speaking, we've been living in this slower consumption growth world, and it's been. There's been some, you know, seasonal puts and takes, which we're happy to go through as needed. But generally speaking, it's been stable and slower compared to the period prior to the slowdown. The only other thing I would say on that front is that, as we observe our own business and as we, you know, plan and execute in our business, but also listen to what other people in the industry are saying, our dynamics seem to be different than other people.
Other people, whether they're cloud providers or other, quote, unquote, "consumption names," have been talking about optimizations. Have been talking about customers proactively reducing their spend with various vendors, and that really isn't a dynamic that we see. What we see is underlying application growth slowing down and our usage growth and our revenue growth as a result in Atlas slowing down, but we don't see customers proactively getting involved to change the way that they deploy Atlas in order to save money, in part because we're so tightly aligned between usage and the revenue that we derive, and that, we think, is a fundamental strength of our model. Customers build applications, spend precious developer resources in order for those applications to get used, and if they get used, they spend more time with us, then, you know, everybody wins effectively.
We don't see a change in the macro environment. We've seen it stable for a while now at this slower growth pace. That's what's implied in our guide for the rest of the year, and, you know, obviously, we'll update you as we see any change.
Yeah. Yeah, that's a great way of framing both the new workload and kind of existing consumption side. I guess to double-click on the consumption side, certainly it was a more dour mood here a year ago when we were up here post second quarter. But, you know, Q1, you talked about a pretty decent recovery in consumption trends. We certainly saw that at your Analyst Day and some of the slides you showed, and it seemed like it sort of continued in Q2, like slightly ahead of your expectations.
So now that you've seen a couple quarters of things not getting worse, I guess, what do you kind of need to see before you have the confidence to say that, you know, we're through the worst?
Let me take it first step.
Sure.
So, we look at the consumption, and you note—you're referencing one of the slides that we showed at our Investor Day, which tries to demonstrate visually what—how do we look at consumption. When we talk about consumption, we're talking about Atlas growth, run rate, week over week on a relatively short time horizon. So we're looking at a seven-day run rate, week over week. If you look at a business on a daily basis, you drive yourself crazy, but we think that there's actual value in looking at it on a weekly. And what we've seen is it slowed down in really sort of a very clear break in pattern starting in Q2 of last year, and then we've seen a kind of variation around that new mean.
And, the other thing that's happening in the business and the other thing that we're learning about as we go along is seasonality. So the seasonality in our growth is driven by the underlying seasonality in our usage. And what we've seen, and what you can see in that chart, is that Q1 and Q2 tend to be seasonally stronger than Q... Sorry, Q1 and Q3-
Q3.
Tend to be seasonally stronger than Q2 and Q4. And as we thought about forecasting the business going into this year, and as we sort of give you puts and takes and updates, that was taking seasonality into account. So in Q1, consumption surprised to the upside. In Q2, we had expected it to be seasonally slower than in Q1, and it was slightly better than that, but only slightly so. And as we sort of think about for the rest of the year, we expect Q3 to be seasonally stronger and Q4 to again be slightly seasonally slower than Q3. But the underlying trend feels stable.
Mm-hmm
... and has felt stable for a while now.
... you know, frankly, the way that we will know that we'll, that we're out of that, and that we're in some sort of faster growth environment will only be after we accumulate enough data points on a monthly or weekly basis, that we say, "This no longer feels like seasonality. Things have. You know, the underlying has changed and feels stronger." Yeah.
Yeah, and I would just add, just 'cause I think there's at least risk of confusion here. We talked about the Q2 results. They were down relative to the Q1 results, but that's what you would expect based on the seasonality-
Mm.
And they were slightly ahead of our expectations.
Yeah. Okay.
Just so people can kinda-
That's-
connect all the dots and all the frames of reference.
Are you regretting putting that chart up at-
No, no, no, no, no.
Yeah.
It's actually been great. It, it's come up-
Oh, people like the chart, so.
It's come up a bunch of times-
Yeah, yeah
... including since reporting again.
Yeah.
I'm glad it was helpful.
Yeah. No, that's great. So, so let's talk a little bit about go-to-market. I think earlier this year, you talked about, and you alluded to this earlier, but really focusing your, your, your reps, particularly at the enterprise, on getting workload by workload, moving away from, you know, pre-committed deals, pre-committed credits, right? Can you, can you talk about-
Yeah
... was there any disruption from that? Sometimes sales changes are difficult to go, but just update us on the progress and what the benefits you've seen so far from that shift.
Sure, yeah. First of all, it's been part of a multi-year effort to reduce friction in the process, which includes reducing the emphasis or the importance of commitments. And part of the reason for that is, if you take a new application that someone is going to launch on Atlas, if the rep has an incentive or is paid primarily off the basis of the commitment, I am going to hold out for the biggest commitment possible, right? 'Cause that's what I'm getting paid on. Amazingly, incentives drive behavior. The customer who hasn't launched the application has no idea what that app is going to consume, and so let's just say we think it's gonna be $100,000, right?
And even if the customer thought it was $100,000, they're probably not gonna commit to it. They might commit to $90,000 or $95,000 or, like, something lower than that. But they don't really know. There's gonna be this negotiation back and forth, a protracted conversation, and ultimately, they end at $70,000 or whatever they end at, and the reality is, none of that affected what the application actually is gonna consume. The consumption's gonna be whatever the consumption was gonna be, and effectively, that rep has wasted a whole bunch of time and effort that they could be going and selling new additional workloads, right?
To me, that's the key point, is when you think about this large market opportunity that we have and the fact that our quota-carrying headcount is measured in the hundreds, as opposed to the thousands or tens of thousands that our competitors have, anything we can do to effectively and synthetically expand the impact of that sales force, right, by freeing up time and giving them time, is valuable. And so that's where this focus on creating, that's where creating this focus on, you know, winning new workloads as the key thing, has really been sort of a long-term trend. And, you know, as you know, Tyler, you know, comp plans are complicated. Reps are sensitive to them. You typically only change them once a year. You're not usually changing comp plans mid-year.
And so, like I said, this has been part of sort of a multi-year phase journey where we said, "Okay, we wanna..." You know, and you can't, like, radically change the plan 180 in a year, otherwise there's too much change management. So it's been sort of a multi-year process to get to this point and feel pleased about it, and it's having the desired effects. Certainly from a, you know, change management standpoint, there has not been significant disruption. Like, you know, there's always changes, and there are always some people who, you know, are trying to find their way with a new comp plan, and we don't have any more of that than usual this year, but, so far, so good.
Yeah. One of the other pieces on the go-to-market that stood out to me, and I think you sort of actually might have reclassified some customers from direct to self-serve, and it kinda sounded like you found some new efficiencies in the business. And like, "Hey, why do we have a sales rep here if they're just going to transact through a marketplace-
Right
... or earn a self-serve motion?" Can you, can you talk about that, what you saw in the quarter-
Right
... the role that marketplace and
Yeah
... self-service can play going forward?
Yeah. So, the beauty of Atlas is that we have access to a tremendous amount of data, and we try to become more intelligent over time as we use that data and usage patterns to judge a customer's potential. And so it’s an ongoing process, but what we've particularly become more comfortable over time is our ability to call when we see that there's incremental workload potential in the account. And so if you think about our sales resources and, you know, just how scarce and expensive they are, we only want to deploy them against the greatest sort of target-rich environment in terms of acquisition of new workloads. 'Cause otherwise, you know, we are not maximizing the return on our investment.
So if you have an account that has a certain number of workloads, and those workloads are growing at a reasonable pace, but we see a relatively small potential for incremental workload acquisition in the near to medium term, we are better off putting it in self-serve, where the cost to serve them is dramatically lower, and more importantly, honestly, take them out of the rep's book so they don't have to spend time renewing the customer or servicing the customer, and instead focusing them on higher potential accounts where they can acquire incremental workloads. It's not purely a cost play. It's really an efficiency play.
And so we've noted on the last call that we've moved about 300 accounts from direct to self-serve, because we think that at least in the near to medium term, we can focus our reps' attention elsewhere to greater effect. I would say a couple of things, though. Number one is, it's not like a permanent state. Customers change, their opportunities change. That's why real-time data is helpful, 'cause we can revisit these decisions as we go along. And then the secondly,
They tend to be the low end of our mid-market channel, therefore, they are not that dissimilar for the average self-serve customer to begin with, and it's incrementally helpful. It's all in the process of sort of continuously looking for efficiencies and acquiring new workloads as quickly as possible. Frankly, we mostly called it out so that you can understand the direct net, the direct customer net adds, which looked lower relative to prior quarters, but it was a proactive decision in terms of managing between the channels, as opposed to any change in the underlying environment and ability to acquire workloads.
Yeah.
And frankly, it mirrors what the best reps do anyway, right? Like, if you're a good rep, you are efficient about your time, and you know where you can drive incremental outcomes because that's how you're going to max out on your comp plan, right? And so this is just sort of, you know, the natural consequence of that.
Yeah.
Did you want to touch on marketplaces, too, or...?
Yeah, I think that'd be great. Just maybe I can phrase the question a little differently.
Sure.
So I think this year there was a pretty big announcement or maybe two big announcements with Microsoft, but just talk about the evolution of the marketplace and, you know, I guess, do you have views on how much of the revenue or business comes through the marketplace over time?
Yeah. So, the evolution has been varied in terms of which... If you go cloud player by cloud player, but I think, all three, at this point, are in very, you know, healthy and good spots along the cooperation, competition, you know, continuum.
Mm-hmm.
Certainly, you know, they are competitors, but I would say the dynamic, for the most part, much more mirrors that of partners, and that's been great to see the evolution. Specifically on marketplaces, I think we're the only ISV who's on all three marketplaces in the console. And so I think that speaks to the popularity among DB the strength of those partnerships. We have go-to-market arrangements with all three. It varies and changes year to year, but I think all three, you know, have their reps getting paid. All three, if you're a customer and you've got a big commit with them, your MongoDB Atlas consumption counts towards those commits, and so these are, these are pretty, you know, good and close partnerships.
I think the dynamic varies customer by customer in terms of how much is, you know, cloud providers doing fulfillment versus jointly, you know, selling in the marketplace. Where we've seen the most success in the field and joint collaboration is when there's an incumbent cloud provider and one of the other two is trying to break into the account and locks arms with us to go successfully leverage the benefits of us working together. And I think that's times when we've seen that sort of most successful on a, you know, sourcing or kind of more than fulfillment basis. And we've seen that grow, you know, over time.
Still plenty more room to run and a lot more opportunity to coordinate, and collaborate in the field, but in general, those partnerships have been very strong.
Okay. Then moving to margins for a moment, you know, I think it's, Mongo is probably one of the few companies in my space that did not have to do layoffs, which is, you know, I think shows how, you know, responsible and diligent you have been in the hiring, but yet you've still seen pretty significant margin expansion, year-over-year, and certainly this most recent quarter, you've been able to couple that with a nice re-acceleration in growth. So I guess my question is, have you learned anything over the last year that you think maybe could drive even more upside to your long-term targets in terms of incremental efficiencies?
I'll take it. Correct. So maybe start with year to date, talk a little bit about the rest of the year, and then we can think about sort of the learnings and the going forward. So happy with the margin performance so far this year. In particular, Q2 was a standout quarter. We had a 19% non-GAAP operating margin. Two things to keep in mind why it came significantly better than we expected. The first one is revenue outperformance. We've done meaningfully better on revenue than expected, and on top of that, in Q2 in particular, we've done better on enterprise advanced and some of these large licensing deals, Alibaba, which comes with an upfront license components, courtesy of ASC 606, and that comes with very high margins. So that's the first reason for the operating margin outperformance.
The second is, our hiring plan for the year, has been back-end loaded by design, and over time, we've made it slightly more back-end loaded tactically as we sort of see availability of talent becoming better in the back half of the year. And so those two things combined are part of the reason why we've done meaningfully better than expected, in terms of performance year to date. If you zoom out and think about the full year, I'm still quite impressed by, you know, by our own ability to perform and execute. So we're gonna expand margins to 12 percentage points at the midpoint of our guide, so that's several hundred basis points better than expected.
I think ultimately what that speaks to is the underlying unit economics, the strength, the tremendous stickiness and sort of built-in organic growth that comes from existing workloads. And those are things that we've known about for a long time, and in fact, it's sort of a continuation of the trend since we've been public. We went public at negative 38% margin, and like I said, we're now gonna get to 12% positive, so that's 50 percentage points improvement. Nice round number. As we think about going forward, ultimately, we're trying to balance two things. One is the continued improvement in profitability, which is important to us, has always been important to us, and we know it's important to you all, but the other one is just the size of the opportunity.
Right, we are just closing in on a whopping 2% share. We have the best tech, technology in the market, we believe. We're fractionally in terms of penetration of our sales force.
... Our rep account is in the hundreds versus the competitors who are in the thousands or the tens of thousands. So there's plenty of opportunity to invest, and we don't want to shortchange the long-term potential for the purposes of, you know, near-term marginal expansion.
Right. In the last couple of minutes, can we just hit on your vision and kind of appetite for M&A? Clearly, there's been a lot of organic innovation, vector search, and the ability to, you know, stream processing, for instance. How are you thinking about M&A? What categories make sense? And then secondly, just in kind of some closing comments, what are kind of the three biggest things you're focused on just to kind of ensure this durability of growth?
Yeah. So from an M&A standpoint, we've done some small acquisitions. I think of them all as sort of, you know, opportunistic and offensive. That said, we have a big market. We have a large organic opportunity in front of us, and because our model is different in terms of the document model versus a relational model, there are not as many things that sort of make sense from a bolt-on standpoint, right? So like you talked about stream processing. We looked at stream processing, and we try to consider sort of buy versus build and all these things.
But when we looked at what was available in terms of Apache Flink, or ksqlDB, or things like that, they were much more rigid, in their structure and didn't think it made sense in the context of the document model, and what we're trying to deliver and drive for developers, and so that made sense to go build organically. But we'll continue to look, you know, broadly. Certainly, I wouldn't, you know, rule out, you know, anything, but I, I think we're mostly focused organically, you know, on the growth, in front of us. And maybe that's a good segue to the last part of your question, which is, you know, we have this very large market opportunity. We have this incredibly strong product market fit. What we really focus on is execution.
I know that sounds boring, and frankly, that's what we've said since the IPO, but that's really, you know, where it is. We don't need some big thing to happen to create some unlock. We don't need to wish for some competitor to do something or not to do something. You know, we've withstood, obviously, many different, you know, attacks and approaches there as the hyperscalers introduced their imitation offerings and other things like that, that we've sort of been able to withstand. We've raised plenty of capital, so we have the ability to go out and do what we need to do. And so really, when we look around the exec team table, the key focus is all on execution, to make sure that, you know, we really genuinely do capitalize on the opportunity in front of us.
Great. Well, that's a perfect place to end. Mike, thanks for joining. Serge, thank you, and appreciate the big crowd on an afternoon with a bunch of earnings. So thank you very much.
Yeah. Thanks for having us.