MongoDB, Inc. (MDB) Earnings Call Transcript & Summary

Well, thanks, everybody, for joining Day 3 of the 3-day UBS TMT Conference. We've got some fantastic companies today. Obviously, we're going to do MongoDB, but we've got Microsoft and Zoom and DocuSign and Informatica, really strong lineup. So thank you for joining. I'm sure most of you are familiar with the dynamic finance duo of Michael and Serge, who will walk us through the story and the numbers today. If any of you have questions, there is a question box in front of you. You're also welcome to e-mail me directly. So Michael and Serge, thanks for taking time out in the morning to help share the story.

No, thanks for having us. It's great to be here for having us.

Good, good. Well, maybe, Michael, I'll start with you. You guys obviously put up a big print, but it's been a crazy week with all kinds of conferences. So to the extent anybody listening in might have missed the details. You want to just take a few minutes and summarize what you thought were the highlights. And to the extent that you and Serge are getting like repeated questions of any sort, maybe take a minute to address them.

Yes, sure. No, happy to, just to sort of level set. So we reported a really strong third quarter overall revenue just under $227 million. That's a 50% year-over-year increased subscription revenue, up 51% year-over-year. Atlas, our Database-as-a-Service offering, grew 84% year-over-year and is now 58% of total revenue. It was really a very strong quarter across the board, both Atlas and Enterprise Advanced exceeded our expectations. Enterprise Advanced came in, you can clock it at just about 20% year-over-year growth, which was also an acceleration for Enterprise Advanced. And so fundamentally, what I think it speaks to or underscores is that we're going after this very, very large market opportunity, right? IDC has the database market at $74 million in 2021 going to $121 million in 2025. We have exceptionally strong product market fit as demonstrated by the uptake, the fact that developers love working with MongoDB, et cetera, et cetera. And we've done a really good job of executing in market, and we're increasingly seeing -- when Dev talked about in the prepared remarks, sort of the social proof effect, right, where as you look at sort of every industries and use cases and everything else, there are clear examples of people who have adopted our modern application data platform to go help them compete and innovate, and we're really succeeding in that regard. I think those are the highlights. They're probably -- it's certainly clean quarters, so not a lot of repetitive stuff, 1 or 2 questions on the early renewals, which is pretty in the weeds, but I'm sure we'll get to it because it will come up because it affects the deferred. But overall, a really strong clean quarter, the team continues to execute well, and we feel really good about.

Got it. Okay. That's a good summary. And yes, we will get to those early renewals later anyway. So maybe before we dive into the quarter and the numbers and interpretation, maybe I'll start with some broader questions for you. So I think in terms of identifying what the secular trends that MongoDB is tied to, it seems to me there are 3 major ones, but I'd love your take. So one, obviously, given that Atlas runs on the major public clouds, you're tethered to that broader migration of workloads to the big 3 cloud providers. Secondly, Mongo has always had a developer focus. And it does feel as if there's an impetus on the part of enterprises to equip developers these days with modern technology and the tools of their choice, sort of the similar trend that developer tool vendors like Atlassian and GitLab might be tethered to. And then lastly, it does seem that all the data companies, be it Snowflake or Mongo or Confluent or Databricks are all part of a broader secular trend. So am I hitting on the big 3? And is there one in particular that you would point to?

Yes. I think they're all relevant and they're all sort of tailwinds that we benefit from for sure. If I try and think about them though, in some ways, they're subsidiary to the broader theme where they're sort of supporting data points or sort of strategies about how to achieve the bigger objective that we're trying to solve for customers, which is really how do companies drive competitive advantage, like at the core of what we hear and what you see in the -- this trend, where it's captured in phrases like software is eating the world or every company becoming a technology company. Those are all shorthand for the fact that companies today are increasingly competing on the basis of their technology. Specifically, what that means is there are software applications. And if we dig it even further, it's really the software that they're building, right? If it's software that's available off the shelf, everyone can have that advantage, that doesn't really drive competitive advantage. The competitive advantage comes from software that you're building. And so what MongoDB and now just to close the loop, the database is at the heart of every one of those software applications. So where MongoDB comes in is sort of a nimble, agile, scalable, modern application data platform to allow you to go do that, to allow you effectively to innovate more quickly, that's really what we're providing. And so yes, cloud transformation or cloud adoption is a big part of the story. But moving to the cloud in and of itself really isn't enough and doesn't really help you adapt and innovate and compete. And yes, developers are at the core of it, for sure. And to your point, have always been part of MongoDB and increasingly companies are investing in their developers. And yes, developers want modern tools, and they want modern technology stacks, and MongoDB is a database developers most want to work with. And that's important. And yes, there are certainly increasingly large amounts of data that people need to harness to drive competitive advantage and to power those applications. And so yes, all of those are sort of relevant trends that we benefit from. But I think it's a sort of macro trend of needing to innovate more quickly and drive competitive advantage from software that you build that MongoDB underpins.

Got it. Okay. Perfectly clear. And maybe a second question I wanted to ask you. One thing we as a team are trying to figure out in a little bit more detail, and I think tech investors more broadly are is how the COVID crisis may have accelerated a lot of these modernization efforts. And the flip side and negative of it is that it may have created a little bit of a bubble and spend for some vendors. I'm not just talking Zoom, DocuSign, but for others that are now facing, sort of, tough comps, because that bubble in spending can't persist at these levels. I don't think that's the case with Mongo because you're not putting up numbers that would suggest that you're bumping up against tough comps or decelerating. But maybe you could address the, sort of, the arc of investment spending post-COVID and into 2022, how straight line that is? Or do you think actually maybe you're looking back over the last 12 months, you're coming off of an unusually robust demand environment?

Yes. So a few things. A lot of the trends that people talked about that COVID accelerated, we certainly benefit from. But as we were sort of describing, the application development life cycle is not quite as quick twitch as some of these other things. And so it doesn't sort of suddenly create this whole sort of burst of demand, right? It's not like everyone is suddenly working remotely and we need video conferencing software, we need collaboration software or we need these things. And we make an enterprise-wide decision and it's rolled out and some part of our future demand has sort of been pulled forward. To your point, that's not really our dynamic. I think what we said, and I continue to believe, is that the trends that existed before COVID, including what we're talking about in the last question, have all been accelerated. And so whatever pre-COVID, you thought our -- the outlook for MongoDB was 5 years from now, I think that picture is even more robust right, given the sort of underscoring of these trends. But in terms of the sort of day-to-day kind of quarter in, quarter out, if anything, I think certainly in the early part of COVID, we were worried that it would actually be harder to generate new business, not because the value proposition didn't resonate but just the physical go-to market motion was disrupted, right? People used to being in customer offices, on customer campuses, physically meeting with them, how would people who are adopting a new technology, how would you do technology feasibility workshop remotely, right, where you're doing lots of diagrams and playing with schemas and like how would you do that remotely. To our credit, we've really kind of powered through that well and certainly exceeded our expectations throughout the pandemic. But I certainly wouldn't say that there's any sort of significant kind of pull forward on a transactional or kind of on a quarterly basis. I do think this sort of future should be brighter as a result, more people haven't realized this is not anything that they can ignore. But I don't think that we sort of have that dynamic that you described. The last thing that I might add that maybe we'll get to depending on where the questions go, but I think it's sort of important is, in our Q4 guide, it's a pretty strong guide. We feel really good about the guide. And I think what it reflects is our underlying confidence in the business, the track record. But I think importantly, if you look at the last several quarters, we have now a pretty sustained -- as much -- as long as sustained as you can have during the course of COVID, a track record of executing despite the pandemic. And where we have looked at -- we've been very clear about this in our guidance that we were worried that that would be some hindrance, right? And that introduced some caution into the forecast. I think we now have enough sustained data points about our ability to execute that, that I certainly can't guarantee anything. And obviously, there are new variants and lots of twists and turns. But I think the level of concern that we had, that was influencing sort of our outlook, we sort of moved away from given these last several quarters of really strong execution despite a highly uncertain environment.

Got it. Okay. That's very clear and supported by the numbers. So maybe the last broader question I'll ask before we dive into some of those numbers. It's just on the competitive backdrop. I guess, we now have Couchbase as a public, next-gen database company. I'm actually doing a fireside chat with the CEO of DataStax in a couple of hours. More are hitting the radars of investors, and that always runs the risk of creating a little bit of confusion as to who's doing what, how they compete. It feels like, Michael and Serge, everybody's calling themselves a general-purpose database and rushing into the cloud. So how could you -- what's the framework that you can advise investors to parse through what appears to be similar messages from multiple database companies?

Yes. So -- and a lot of the technology is complicated against, and so I get the confusion. And to your point, there aren't 20 publicly traded database companies and so it's not a sector that people spend an inordinate amount of time thinking about, so I totally understand that. I guess I would say we aim to be general purpose from the start. That was really the outset and sort of the initial sort of value prop of MongoDB was to get -- to address all of the challenges of relational, but without throwing the baby out with the bathwater, right? There's a bunch of good stuff about relational in terms of performance in secondary indices and a whole bunch of other stuff. And so we've kept that. And I guess the best, sort of, proof that I can offer is certainly the 2 companies you mentioned, Couchbase and DataStax. When I came to MongoDB, coming up on 6.5 years ago, all 3 companies were roughly the same size. And here we are 6.5 years later, and I think we've out-executed in the market and the results certainly speak for themselves. I would really tie it back to 3 things. One, the underlying technology is truly general purpose. That's the benefit of the document model. The document model, we talk about as really a superset of all the other models. And so you could run graph, you can run time series, you can run key value. You can -- anything that you want to run, you can do given the flexibility of the document model. So I think there's a technological superiority and user friendliness that has clearly driven the success. Secondly, there's the developer mindset, right? That developers have seen that, developers know how easy and intuitive MongoDB is to use. And so MongoDB is the developer or the database developers most want to work with, right? So we've taken the sort of the technological superiority and advantages are translated into developers want to use the product. And then lastly, we've heard that with strong and solid execution in the market. And again, I think sort of the results speak for themselves and you can see it. I would just maybe make a couple of other comments because I think it's probably important. It is a very large market, right? As we talked about, the $74 billion going to $121 billion over the next 4 years. And so many people can be successful, right? There are -- because there's a different database for every application, large enterprises will have this sort of polyglot multi-vendor strategies. Increasingly, we are hearing people who are clear that they don't want to add a net new database for every new application that they have, right? And so being able to have something that's general purpose where you can standardize on is incredibly important. And so what I'd say is I think there are probably plenty of accounts where 2 or 3 of us exist in that account. And that might have been true 6 or 7 years ago when I showed up. But what's happened over time is we've added the workloads, and we've cut across the account, across use cases, not just like, oh, we're really good at caching or really good at like this one specific thing, but across use cases. And then one, then the modern standard that people have been able to align around given the general purpose nature of what we do. But it's a very, very big market. We don't particularly run into a lot of those folks where we wind up going after where there's like a lot of market share from someone like Oracle or competing against the cloud players. But again, it's a big market.

Got it. Perfect. Okay. That's very helpful. Let's dive in a little bit into the numbers. You flagged it, Michael, earlier on. But one of the interesting things about the quarter didn't really affect revenues, but was the big bump in DR and billings as a small handful of very large Atlas customers elected to renew early, make large multiyear commitments. Now I wanted to ask you about how to interpret that, because if it's merely a few Atlas customers essentially chasing lower price points, and it's kind of one-offish, that's one thing. But if it's sort of the start of something wonderful, where you might see this from a broader swath of customers, and there's more strategic reasons why they're doing this, that's a different thing. So maybe you could walk us through how to interpret it and why it's happening now.

Yes. So maybe I'll take the first crack, and then Michael, feel free to jump in. So first, just to make sure Karl you summarized it a bit, but maybe just to pause on it for a second so make sure everybody is on the same page with the mechanics. We are talking a handful of our absolutely largest Atlas customers. So that's the first thing. These are not the top 100 customers. These are the top 5, 10 customers when it comes to size. And they all had a similar dynamic. And dynamic was something like this: They were consuming already higher than their prior commitment. And they had confidence that the consumption would only go up from where they already were. And they approached us in Q3 to say something along the lines of my contract is expiring in Q4 or in Q1, but I'm already consuming above the commitment, and I know I'm growing from here. So why not just do a new deal now? And you're right. From their perspective, with a higher commitment come incrementally, I would say, better economics. And because they have high conviction: a, run rate is already higher than prior commitment; and b, they're going to continue adding workloads, why not make that incremental economic benefit and sort of lock in that gain? From our perspective, it's a couple of things. Number 1 is we want to be flexible and customer-friendly. Number 2 is there is value in having the commitment. But most philosophically, once we have a commitment that's higher than the consumption, then we're aligned to find ways to use that commitment. We're aligned to chase workloads. We're allowed to find new developer teams inside the company or new incremental projects input on our platform. And that's great. And that speaks to the whole standardization and sort of reducing frictions of internal adoptions. And that's fundamentally the reason why we pay the price with giving them better economics to sort of get that incremental access and incremental alignment. What that means from the perspective of financial statements, and you nailed it but I think it's very important that I just repeat it, which is there was no revenue. In fact, Atlas' revenue upon consumption, so nothing has happened in fact, because consumption was already higher than prior commitment, much of this benefit was already backward-looking, it was already in the revenue. However, when you sign a new contract, and these customers would build annually in advance, and again some were multiyear commitments, others were large single year commitments, then you have deferred revenue event. And we've been saying for a while, and we continue to believe, that a billing is a poor and an increasingly poor indicator of our business. However, we know there's a portion of the investment community that looks at it closely. And so we wanted to give the context so that people can properly put the calculated billings number in perspective as opposed to come to some other conclusion, which we think will be unwarranted. And so -- and that was that was well received. People appreciated the extra context that we had a number of sort of clarifying conversations. As far as what does it mean for the future and whether this is the start of something wonderful, I'll take a first crack, Michael, but I'd love to hear your views as well. Customers vary different widely in terms of how they feel about commitments. And it's actually quite surprising to me and to us. Because there are many customers who are perfectly happy to commit well below their run rate and remain that way. And when you ask them why you're leaving money on the table, the answer is usually something along the lines of, "Look, my app is volatile or it may become volatile. I don't like paying you upfront, even if I have the money." Or even it's as simple as it's the cloud. It's supposed to be pay as you go, pay as you consume. I'm philosophically opposed to writing some enormous check. So we're happy that it happened with a few largest customers is not really a surprise. They are already -- it's kind of like preaching to the choir, they're already among our largest customers. I'm sure it will happen sort of here and there going forward. I wouldn't expect it to be the beginning of a trend. And -- but it does in those specific cases, speak to sort of continuous growth in our strategic importance, and we're excited about that.

Yes. And I think the only thing that I would just sort of add is, I think, it's quite qualitatively interesting and relevant and sort of underscores the strategic nature, not just of MongoDB but of Atlas. I think we've increasingly seen Atlas get more enterprise and I think that's been sort of a recurring theme over the course of fiscal '22. So I think it's sort of qualitatively interesting and supportive of that trend and illustrative of that trend. I think it is not quantitatively interesting and it's sort of a one-off. And so it's part of like we're just worried that people would like see a number, do something in his spreadsheet and sort of go crazy. I don't think that's the interesting or relevant part of it. So hopefully, that just helps to put it in context.

And I guess we'll see the evidence of this also in the RPO number when your Q comes out, right?

That's correct, yes.

Got it.

Not -- you should just elaborate a little bit only in the sense that they are -- go ahead.

Yes, sorry. So our RPO, we're going with the practical expedient of just including contracts that are 12 months and longer. So for those that are 12 months or longer, you will see, sort of, both the short term and the near term and in the longer term impact of the RPO. But some of these contracts, large ones were actually 1 year in nature.

Okay, perfect. Thanks, guys.

So just to be clear, the practical speed excludes contracts that are 12 months and shorter. So it's only the multiyear deals or contracts longer than 12 months that actually show up in RPO, which is another reason why our RPO is not so useful or insightful for us.

Got it. So if these early renewals really didn't have an effect on Atlas revenues in the quarter, maybe you could talk a little bit about the Atlas growth you did put up. 84% growth was phenomenal. You've described in an earlier question and answer some of the broader trends that you're tethered to. Was there anything particular about the Atlas growth in the October quarter to call out as being a driver. any kind of unique shifts, maybe a new workload types, anything that you would flag or just steady as she goes, similar drivers to prior quarters?

Yes. Maybe I'll take the first crack at this one as well. So we're very happy with Atlas at 84%. Back to some of your prior questions and conversations with Michael, Karl, it's actually off of a tougher compare. Because as we think about the cadence of how COVID hit us last year, it was really like a slowdown in consumption in the first half followed by a normalization or rebound in the second half. So we're comping that normalization. And we were flagging that to investors like our comp gets harder. And here we are sitting accelerating on top of that harder comp. So we feel very good about it. The one thing that's worth calling out, and we did mention this in the script, is -- if you think of Atlas sequentially, which I think is helpful, and you should kind of -- investors should think about Atlas both on a year-over-year and sequential basis. But if you think about Atlas sequentially, 17% growth quarter-over-quarter, very strong number. Sequential growth more so than year-over-year growth tends to depend on growth of the existing applications that are already on the platform. Because you just -- you add workloads sure over all periods of time, but they tend to start small. So in a shorter period of time, it's really the base growth that drives the outcome. And there, we've seen growth of the existing application fleet, if you will, be higher [indiscernible] -- be at the high end of our observed range. And it's very broad back to your question like what's behind it. We sliced it many different ways. By geography, we saw breadth. By industry, we saw a breadth. It's not a back to work or reverse COVID or whatever you want to call it, benefits and some currents elsewhere. It's not like individual pockets of strength that we can extract away. It's just apps as a whole group marginally faster than they usually do. And there's always some element of variability. And with this time, we happen to be at the high end of the range, and we wanted people to understand that because that does benefit the 17% sequential and therefore, does benefit the 84%.

Okay. Perfect.

I would just add, in case it isn't obvious. It's at the high end of the normal range, but there's no reason to believe that it will persist. There's not something like explanation for why the high end should be the new normal, right? And so I think you just got to think about that as you think about sort of guidance and your own forecasts.

Got it. Let's talk about the EA business. To me, this is like the little brother who keeps getting overshadowed by the prolific older brother who gets all the trophies. But it was actually the EA business that, at least relative to our model, had more upside actually. So let's shine the spotlight a little bit on the EA business. And maybe you could describe why you saw an acceleration to 20% growth from mid-teens in the last couple of quarters?

Yes. So I'll take a step back here, Karl, and to remind everyone that while we provide a lot of the disclosure at the product level, we really run the business on a channel basis, right? And so if we think about the channels, self-serve is really almost all Atlas. Within direct sales, the mid-market at this point is mostly Atlas. Those tend to be early technology adopters, very cloud forward. And so the real determinant over the arc of time of the Atlas EA mix comes from what's happening within enterprises. So that's sort of like the first thing that people should be aware of. Secondly, in any given quarter, the underlying mix of EA versus Atlas really depends on just what deals are happening. And if I'm an individual rep, I've got a territory, some accounts will naturally be more cloud forward. Others might be for regulatory reasons, industry reasons, culturally, whatever, might still be in a self-managed environment. And in any given quarter, as a good sales rep, I'm just going to go to wherever the pockets of activity are where I can most easily transact, right? Like I don't -- we don't have incentives that say, "Oh, $1 of Atlas is worth more or less than $1 of EA." Our goal is to make MongoDB easy for customers to consume regardless of where they are in their cloud journey. When we started life as a public company, Atlas actually lagged Enterprise Advanced. There was less feature functionality capability. We quickly brought that to parity. And now you can say with the other things that we're adding in terms of the application data platform in Search and Data Lake you can actually argue Atlas is ahead. And certainly, the trend and the sort of future trajectory of where we think application development is going, we'll be much more sort of cloud centered. But there are plenty of industries and plenty of businesses that are still mostly or entirely on-prem. I think it's very -- it's very common or very easy, if you sit in a lot of conferences like this, everyone talks about cloud, cloud, cloud, to just think that most applications are in the cloud, and they really aren't, right? We're still pretty early on in early days. And again, there tend to be whether it's industries or geographies or just company-specific issues, financial services is heavily on-prem still. Health care, a whole bunch of other heavily regulated industries. And so our goal is just to make MongoDB easy to consume. And in any given quarter, like I said, the mix will be just based on where is the activity. I will say the one thing we are talking about that sort of a continued evolution is people are increased -- even if you are mostly self-managed and on-prem today and not in the cloud, I think people understand and acknowledge that over the next 5-plus years, they're going to start to move that way, right? And so increasingly, part of the value prop of adopting MongoDB and adopting Enterprise Advanced is that it's an easier transition to the cloud, right? It's almost like an on-ramp because the move from Enterprise Advanced to Atlas is a much more seamless move than if you're going to try and rewrite the application on some other technology or move to some cloud player or some cloud player's proprietary offering or whatever else it might be.

Yes. Got it. And what about more tactical reasons why it might have had a little bit of a better quarter. Maybe the renewal base was a little bit higher. Maybe you guys did a better job upselling. Anything like that?

I think that's less of a factor in Q3 and Serge, you can certainly amplify that. I think that tends to be more of the dynamic we would see in Q4. If you look at what percent of the renewal base does EA represent, the largest percent of the renewal base that's EA is in Q4. And for EA, really for any relationship, but particularly for EA given it's not consumption oriented, the biggest, most common point for sort of upsell would be at the time of renewal, right? We're discussing the relationship. We're evaluating things more broadly. We've identified maybe over the course of the time leading up to the renewal, additional applications, additional workloads that might make sense for MongoDB or perhaps the application itself has grown. And so you want to contract for more servers, right? So you tend to see not only is it the biggest renewal base, but it's the biggest sort of the upsell, the expansion, the expand part of the land-and-expand, particularly within EA given it's not a consumption-based product, tends to come upon renewal. And Q4 is actually the largest percent chunk of the real base for EA.

Got it. Okay. Well, maybe this is a good segue to talk about 4Q. Tell me if this is a reasonable way to assess the 4Q guide. So if I -- if I look at your 4Q guide and calculate what sequential total revenue growth it implies relative to the actual 3Q you put up, for this quarter, it's about a 7% sequential increase. If I look at 4Q last year, your guidance implied a 4% sequential increase. And if I look at 2 years ago, your 4Q guide implied a 1% sequential increase. So it feels like this is a "strong" guide. Now maybe that's influenced by the fact that Atlas is a larger portion of the mix. But maybe you could comment on whether that interpretation of 4Q guide feeling pretty robust relative to priors is appropriate.

Yes. So we feel very good about the 4Q guide. We think it's sort of a robust and sort of appropriate representation of what we're seeing in the business. I think sequentially is a good way to think about it, we can also talk about it year-over-year. Sequentially, both businesses ought to do better. Michael just mentioned the renewal base uptake, that's true in every fourth quarter. So that's like not new. That is more par for the course and sort of the nature of sort of how the business builds or has built historically throughout the year, and that helps. Atlas tends to grow sequentially, of course, as well. What's maybe different a little bit this year than in prior years, is this higher than normal consumption that we -- consumption growth that we observed in the third quarter because that has a kind of carry-on effect as you think about it. Because we think of an exception -- we analyze customer and existing application expansion on a weekly basis in terms of ARR. So that's like base dividends in the next quarter as well. So that's part of the reason why the guidance is good as it is in Q4. And then -- as you think about it sequential -- as you think about it on a year-over-year basis, again, the mix of business is different in Q4. So Q4 has more EA than any other quarters. So that plays into account as you think about how this shakes out on a year-over-year basis.

Okay, great.

The only thing I would just add, and I know I mentioned it earlier, Karl, but I think it feels relevant to sort of underscore it here, especially if I look at sort of this year versus last year. I think last year, we had a lot of unknowns related to COVID that was -- that we felt that we had to take into account when we were providing our outlook for the quarter. I think that while certainly there continue to be unknowns related to COVID, I think it's harder to ignore the several quarters of success that we have had in executing it against that. So when I think about kind of Q3 -- sorry, Q4 of this year versus Q4 of last year, to me, that's one of the biggest differences is there are a bunch of unknowns that we had to take into account back then. And while there's still unknowns, I think we've got much more of a track record of being able to execute through those. And so we've been able to sort of peel a little bit of that away.

Makes sense. I'll ask you one more, and then we'll turn to questions. The last -- the other thing that struck me about the quarter was the improvement in your non-GAAP operating margins, negative 1.5%. You guys are finally an inch away from hitting that breakeven. It's tempting perhaps to extrapolate from the last couple of quarters into next year. But Michael and Serge, what are -- I know you haven't provided guidance yet, but what are a couple of reasons, perhaps why it might be inaccurate to extrapolate linearly?

Yes. So I'll say a few things, and then obviously, Serge, feel free to jump in. But we've made the comment that, I think, for a couple of years now, we feel like we've made a lot of progress on operating margin. And because of the large market opportunity, because of the strong product market fit and because of the strong returns we're seeing on the investments that we're making, particularly in go-to market and in R&D, we actually have made more progress than we would want to. And what's interesting, I think, is sort of fiscal '22 relative to fiscal '21, provides a reasonably good apples-to-apples comparison in the sense of mostly COVID heavily affected both years when all is said and done. If you look at our full year guide for fiscal '22, at the midpoint, it basically implies a 4% operating margin loss, which is about 400 basis points of improvement versus fiscal '21. And so I think that's at least sort of like a normalized view because you kind of stripped out the COVID effect or non-effect because it's relatively an approximate in both periods. That's more progress than we wanted to make. Part of that is because of the strong revenue outperformance. And so I would say that's probably the next thing to say. And then lastly is we've tried to be really clear and transparent around incremental costs related to COVID. In our March call this year, we had said that with the return to more normalized operations in the second half of the year, we would probably incur $20 million to $25 million of expenses that we did not occur in 2021, right? So this is related to T&E, in-person events, office space, like things like that. Obviously, COVID's gone on even longer. And so in our most recent call, we updated that to make it clear that in fiscal '22, we only thought we would see $9 million to $12 million of that. Certainly, as we look at fiscal '23, we obviously haven't guided and we'll go through our planning and do all that kind of stuff when we have our March call. But I think what we'll do is we'll be looking to do the same thing we do every year on sales, which is grow the sales force as quickly as we can that's operationally responsible, right, because we have this very large market opportunity in this very thin footprint of coverage. Our biggest issue in terms of capitalizing our market opportunity or deals that we are not aware of or conversations we're not in, our win rate in conversations that we're in is exceptionally high. And so we want to increase the footprint, right? Like we're -- our footprint coverage is measured in the hundreds of quota-carrying reps versus the thousands or tens of thousands for our competitors, right? And so the governor or the constraint in sales and marketing for us tends to be operational, right? There are only so many people you can hire without lowering your quality standards. There are only so many people a ramping manager can ramp in any period without the quality of the ramp suffering and therefore, what certain levels of productivity or ramp levels to levels of productivity aren't productive. And I guess we've just seen far too many people who've had self-inflicted wounds or stubbed their toe. And so we take the operational side of that quite seriously. We'll look to invest in R&D to expand the application data platform, obviously, scale the rest of the business. And then we'll layer on top of those, sort of, other COVID effects. And our current view is that we're going to assume a renormalization of activity. And so we've given some specific quantification. Obviously, you could normalize the -- you can annualize those numbers and, of course, apply a rate of growth, right, because they're mostly headcount related because we're talking about T&E and office expenses and things like that. So -- and that will just be an add on top of whatever the kind of baseline COVID comparison would be.

Got it. We've got to -- we only have a couple of minutes, but we have 2 questions. Why don't I ask them and maybe we could hit them rapid fire.

Sure, yes.

One of the investors is asking what is the typical lag between landing a customer and then hitting your customer metrics and usage ramping up and recognizing revenues? And is that different for different sized customers?

It's different for different sized customers, it's different for different products. So if you have Enterprise Advanced and you entered a subscription for that, we're going to recognize the term license component of that upfront, right? We've said that's about 1/4 of the annual amount. And then the remaining 3/4 will be spread ratably over the course of that 1-year contract. Ultimately, regardless of -- usage because that's a self-managed environment. On Atlas, people tend to start using it, and this is where sort of the channel matters. If you're a self-serve customer, you're going to come in or even in the mid-market, you're going to come in, you're going to come in small and then you're going to sort of ramp over time. There's a similar ramping on the Enterprise. But if it's a known application as opposed to just starting out in the cloud, but there's sort of like a clear-cut application, it will take some time in ramping. But that depends on like is it a migration of something that I'm already running? Is it a brand-new application? Like all those things factor into it.

Got it. Last one is, why is application usage growth coming at the high end of observed ranges? What's going on with apps that the usage growth is so strong now?

Do you want to cover that, Serge, or you want me to?

Yes. So I think the key is sort of there's always a range and that we -- in various different quarters, we end up in different moments of it. In our observed sort of behavior or experience in Q3 that we were near the high end of that range. But to Michael's point, it's one data point, and we obviously see it every quarter, so we know that there's a distribution. So I wouldn't read too much into it.

Okay. Got it. Michael and Serge, we're up against time, but thanks for carving out time to help UBS' clients understand the MongoDB story a little bit better. And thanks for participating. And if we don't chat, happy holidays to both of you.

Yes. Same to you. Great to have you. Great to see you, and thanks for having us. Happy holidays.

Thank you, Karl.