Datadog, Inc. (DDOG) Earnings Call Transcript & Summary

So we're really happy to have the Datadog management team. We have Olivier -- CEO, Olivier Pomel; and CFO, David Obstler. Thank you so much both of you for [ coming ] today. Before we get into the discussion, really quickly, for important disclosures, please see the Morgan Stanley's Research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representatives. Thank you folks for coming today at the conference yet again.

I wanted to start off with what's been a pretty successful year that you just completed, grew revenue 63%; operating margins of 19.5%; serving over 23,000 customers. So a successful -- another successful year by all accounts. You did see some revenue [indiscernible] growth start to flow in Q2. And I was wondering, Oli, if you could sort of walk us through the difference, the nature of the slowdown that you're seeing now versus maybe what we saw in the pandemic? And to what extent is your -- how is your visibility compared today versus back then?

Yes. So first, thanks for having me here. So what we saw this -- over the past few quarters is a combination of 2 things. So one is we saw the growth of existing customers slowed down as they've optimized their cloud environment. And in addition to that, we saw interestingly the amount of new logos and new products [indiscernible] actually keep scaling and reached record levels in Q4. So we have this dichotomy between those 2 things. The last time we saw a slowdown, as you mentioned, was at the very beginning of COVID. I think we all forgot about it by now, but we had this very quick crash of everything, and everybody was scared of not having enough cash and was trying to save on any money that hasn't gone out of the door yet. And so we saw that. At the time, it was very brutal, very broad across the whole customer base. And we didn't have to ask ourselves to -- we didn't have the time to ask ourselves how much visibility we had into it for the future, because we -- before we're done looking at it, everything has started growing again. This time, what we started seeing started in -- starting in the second half of Q2 was a slowdown that is much less pronounced. That is also quite broad, but it's more predominant with the companies that are further into the cloud migration, mostly cloud native companies and digital native companies, that have a significant amount of their spend in IT because they tend to be digital and also for whom their IT spend is substantially just made of cloud spends. And so that's the part of customer base that we saw a slow down. We're not completely sure yet when that's going to be done. I think when, and we can discuss that a little bit more, if you want. But we -- when we gave guidance for the year, we didn't assume any recovery from that. We actually looked at the number. We asked ourselves in that are we seeing some optimization motions? And can we extrapolate what we see from the customers that are optimized and started going again to the whole customer base? We decided not to do it in the guidance we've given. We've assumed that a continuation of what we have seen throughout the second half of last year. And the reason for that is we don't think we have enough visibility into that. And we don't think our customers themselves necessarily know whether they've done yet. I think everybody is in a bit of a wait-and-see mode in terms of what's going to happen to the economy over the next year.

Yes. And with that understanding that you don't know how long this is going to last, both of you mentioned on the earnings call that cloud optimization is a thing that actually happens quite often through the life cycle of a customer. Can you give us a sense of what a cloud optimization project in terms of duration? How long does it last? Is it a 1-year project? Is it a 1-quarter project?

So we see routinely [indiscernible] it to the history of the company. Pretty much every single customer go through optimization. I would say, generally once a year, sometimes a bit more maybe. It tends to coincide with the end of the customer's contract, basically when they need to understand what they're going to have to commit to for the future. And everybody is going in the cloud. Everybody is adopting more and more of it. And so everyone is coming to this region at some point that they're spending large amounts of money, what is it going to do and all that. Let's see what we actually don't need and optimize a little bit. So the motion we see with customers is they'll typically optimize. They'll save maybe 10%, 15%, 20%, and then they'll start growing again. And the overwhelming motion has been every year, they share 10%, 15%, and they grow 50%, 70%, 80% after that for the rest of the year and then they rinse and repeat. They do something similar with the cloud providers, though it's not necessarily at the same time because the contract with the cloud providers are not contaminated with ours. And also, it takes a little bit longer to do that with cloud providers, because it has to happen at the level of the workloads. Customers are going to have many different teams, managing many different workloads. And so it's going to be more diffused and happen over a more prolonged period of time. So today, we see a little bit of both. We started seeing the observability optimization with us first. And we see it first because it's easy. We just have to go into our UI. You can decide for certain types of observability data that are volume constraints such as logs in particular, but also certain parts of APM that are transaction-based. You can decide to go and sample some of the data. We can decide to reduce the retention on some data decided. For example, for certain types of debug logs, you don't need to keep them for a month, if you can just keep them for a day or things like that. So it can be done very quickly. There's nothing that's customer-facing in there. So it is very little risk for our customers to do that. Then we'll see them do that, too, with the cloud providers, but it's going to take maybe a couple of months for them to do so, as they need to maybe downsize some workloads that might be customer facing. They need to release code to do that or release configuration and measure the output of that. So we -- it's going to take a little bit more time with them.

Yes, that makes perfect sense. When we look at the growth trajectory of the major hyperscalers and what they've sort of, given in terms of forward indicators, it sounds like Datadog is seeing a more pronounced slowdown than the hyperscalers are despite being at smaller scale. I was wondering if you could sort of potentially bridge the gap between what we're seeing at the hyperscaler level and what you're sort of guiding to for next year going from 60% plus growth to mid-20s growth in 2023?

Yes. So [indiscernible], it is not a one-to-one mapping between the hyperscalers and us, right? So on their end, they have a bunch of other things in their numbers that are not necessarily infrastructure. And I think that also depends on the cloud provider. I mean, you can argue that some of them have more different things in there than infrastructure and others. And on the other hand, we do more things -- and we do more different things over time. And we are not in a -- we're still not in every single market segment in geography the hyperscalers are in. So the numbers are not 1:1. That being said, actually, I agree a little bit with you because I think the slowdown is actually very similar between what we've seen from them and from us. So when you look at us, we're going from, say, 60% last year to a guidance in the mid-20s for next year. When you look at [indiscernible], I think they are going from 40% growth to -- I think, we're seeing forecast in the single digit -- mid-single digit for the next couple of quarters. They're actually slowing down in a commensurate way. And another way to look at it is that our growth as a multiple of their growth or the cumulative growth of the cloud providers is actually increasing. So we -- again, not supposed to be a one-to-one to start with, but I don't think there's anything glaring in there either.

Yes, that makes perfect sense. So just zooming back out and just thinking about the category overall, a common point of feedback I hear from investors is that Datadog is a great company. I understand the market opportunity, but there's so many alternatives. I don't know who's going to be sustainably the winner in the pack in terms of building a business that can both grow and grow profitably. How does the team think about competing in a market where there are customers do have multiple alternatives? And what's the core element of the strategy to stay ahead of the pack?

Yes. Well, so it's always been the case since we started the company. When I tried to raise our [indiscernible], I heard monitoring clouded market, you'll never win. And then we've seen the steady stream of incumbents and other companies and competitors of scale and new entrants. And we keep hearing about how there's going to be competition. They're going to do it better. They're going to be cheaper. If I had $1 for every time I've heard that, we'll have a $20 billion market cap. So there's nothing that's different in there. So it looks like competitive market. And then you look at the specifics of us, we're growing faster than anybody with scale from a larger base than just about any one, if you just accept 1 player. We're doing it with extremely high retention, growth retention. We're doing it with gross margin -- software gross margins, basically, that are very high. We're doing it also with the lowest cost of sale of just anybody out there and most software companies, which allows us to really out-invest anybody else in product development and differentiation. So at the end of the day, when you put all that together, it doesn't look like a competitive commodity market. It looks like a very differentiated market. And we're in the lead there and our strategy to compete there is to keep innovating, keep building on the differentiation, which we've successfully done today, and use our scale and our efficiency to out-innovate anybody and become the de facto platform for the space.

Great. And as we look at -- sort of heading into this tougher budget environment, there is an opportunity because the average enterprise has 10-plus different monitoring tools. There's an opportunity to consolidate monitoring spend onto the Datadog platform, just given just how fragmented the category has been over the last several years. What's been the early results of this initiative? And sort of heading into 2023, what's the plan to garner more momentum on this front?

So we're doing well. I mean, we're discussing that regularly on earnings calls. We talk about a number of new logo wins we're getting or expansions with customers where we consolidate a lot of their existing products. So we're happy with that. It is still not, though, the dominant part of our go-to-market. Our go-to-market is still mostly built on going after net new applications, net new workloads, starting small and growing with them and then [indiscernible] in consolidating. So our strategy today is to make sure that, in the long run, we are the consolidator of choice, that it's a complete no-brainer to bring everything on to Datadog because we cover everything end to end and we do it more efficiently and better than anybody else. And so for that, we're investing heavily in expanding the product set, covering more of the product spreads for our customers. So that's what we've done in the past few years. That's still what we do this year. I still don't expect our growth to be dominated by consolidation over the next year or so.

Great. And let's bring David into the conversation around the topic of pricing and packaging as well. And Oli, feel free to comment as well. But in these categories like observability and some of the database market where data is the driver of growth means a really exciting opportunity at times and software historically, that's also meant that customer sort of view a provider -- an infrastructure provider as like sort of a data tax. As CFO, to the extent you get involved in contract discussions with customers, what things are you doing from a pricing and packaging perspective to sort of avoid this perception?

I think there's some very attractive things the way we've always gone to market, which resonate in this market. First of all, we price based on usage, not in the case of data, not based on ingestion really, but on how it's used. We make it pretty transparent. And we've been investing in the platform to try to become even more efficient in the ability to use data. So whereas a number of software vendors come and try to get -- commits out in time, what we've always done is try to land and then have our clients grow with the product, whether it be with the host of the data. So as a buyer, I like that because I don't have to pre-commit. I can buy more as I go along. And I get discounts as I get into volume, which is a pretty attractive value proposition from a software -- for a software buyer in this market. Do you want to talk about data more?

Yes. And I think some parts of what we do are heavily direct constrained. And so I talked about logs and some parts of APM early on where customers can send basically arbitrary amounts of data, and they can generate amounts of data that are completely divorced from the growth of their businesses, which is one of the core issues in that domain. So what we do there basically is we invest a lot in the product to unbundle the various parts of the product and give customers levers so they can align what they pay with the value they get. And we see that play out, by the way, in situations where they actually do optimize, like we're seeing right now, it's by design. Like it's actually good for us, good for them. They get to tweak. They get to understand what's valuable to them, and then they get to grow over time as customers. So we keep investing in doing that. At the end of the day, when customers have data volume, that [ sale ] over the next 3, 4 years are going to grow by 2 orders of magnitude. The difference there is not going to be made by selling a product that's 20% cheaper. It's not going to make a dent in that. The real difference is going to be met by having the right feedback loops and the right levers. So customers can really understand what it is they need to keep and build a discipline in each and every one of their teams to do that and manage basically how much they already produce and how much they already use and for what. And that's what we've been building with them.

And sort of as a follow-up to that question, also sticking on the topic of pricing, the company has generally avoided doing broad-based ELA agreements. As we think about Datadog further penetrating the large enterprise market, given that many large enterprises are used to buying in this sort of ELA construct, do you think avoiding ELAs limits your opportunity in the enterprise?

Well, I think our product is basically based on the amount of use. So it is -- our clients are used to buying through the hyperscalers in the way that they've been buying Datadog based on use. It's unlikely that, in most cases, we're going to offer unlimited use of the platform because of the variability of the cost. But what we're trying to do is we're trying to provide more utility in the platform and complete transparency in pricing. So I think as Oli said, clients can calibrate themselves and understand where they want to use the platform and how they want to use the platform. So we don't see ELA is driving our pricing but more transparency, more functionality and more control of using the platform by clients.

And just to build on that, so we -- there's 2 things we like about not having ELAs. The first one is we get very good signal on what is valuable and what's not in the product. And look, whether you put in ELA or not, customers are paying for the functionality. It cost something to develop it. It cost something to ship it. Your customers are paying for it at the end of the day. By splitting it out and then bundling it, we get very good signal on what's variable and what's not and what customers are buying, how much they're paying for it and how we scale that. And that helps us drive our product development, which, if you think of what we do in the number of products we have, it's really, really hard to build the right thing by having this very, very clear metric of success, we're able to do that. The other thing I would say is if you just think also of the way customers consume infrastructure, and if we were to sell a fixed ELA deal with unlimited data for a few years, there would be no limit in terms of what customers can send, and it would bring bad habits. And in the end, it's not a good deal for anybody. It's not a good deal for us because we make that -- we lose money on it because customers are selling us terabytes and terabytes of data that they don't use, they don't care. It costs nothing, so we have all of that bad stuff out there. But it's also not a good deal for customers because after that 2-year deal, nobody wants to renew that deal. We don't want because we're losing money. And on their end, they have a huge mess on their end because they have grown all these -- their instrumental data applications everywhere to send all that data that is very messy and that's very hard to put under control. So it's not a good situation for anybody. We think that just the way the cloud providers charge based on infrastructure, having or building the user base is great for everyone in the end.

And that's been borne out by a very consistent unit pricing over time and essentially, which has maintained itself over many years.

This category is known for a high degree and high pace of innovation. So when we think of the tech stack more broadly, to what extent is Datadog have locked into its tech stack? Meaning that as new approaches to storing telemetry data, analyzing telemetry data in real time, the predictive analytics associated with AI ops, as that naturally evolves, are you anyway sort of limited by your tech stack? Or can you evolve that tech stack over time?

Yes. It's a great question. So under the hood, so we're not data stores centric. So we -- architecture has the federation of different data stores that are linked together by stack-based data model and by the way we structure the UI of the product, but we've been swapping in and out different components of the underlying platform and the underlying data. So which is fairly different from what you'll find in many of the products that were born in open source because these tend to be very, very data store centric. These tend to expose the implementation specific of the data stores to the end users. And it's very, very hard for the companies that get built around that to break out of that. So to get more specific, we have, in particular, 2 data stores that are very high volume. One is our time series database, which we use for our metric data, which is very high volume and high velocity. And we also have our EventStore, which is used for a number of products such as log management, APM traces or release on monitoring, for example. We had the fifth iteration of our time series database, right, the third iteration of our EventStore. If you're interested in winning some about it, we actually published a series of block posts on the architecture of our EventStore. It's called Husky, like the dog breed. We love dog breed names at Datadog. And we -- you can read more about basically the way it follows the latest evolution in terms of separating storage and compute, which opens different ways of scaling costs for acquiring and processing this data, different ways of packaging the applications that use this high-volume data, and we think is a good next step for us basically. But the point is we keep doing that, like that's something that we can do because of our architecture, and we kept doing it, and we expect to keep doing it.

I will definitely be checking that block first out, Husky. Awesome. Let's talk a little bit about security. You got -- the team has gone -- or data hub has gotten into the security market since about mid-2020. Can you sort of walk us through the progresses that you made? To the extent that you've seen any setbacks, what have been sort of the challenges in that time? And as we stand in early 2023, what would you sign the probability of success in sort of successfully moving into the cloud security market?

So first of all, we're very happy with the progress we've done in security. So we have thousands of customers using the product or the set of products because we have 3 main products in security. We have cloud team; we have application security; and we have a CNAPP product, which we call Cloud Security Management. And -- which is more on the infrastructure side basically. And the way we qualify our current success there is that we have a smaller number of customers that are using those products wall to wall and a much larger number of customers that are large, but users use that product in pocket. And so what we're doing right now is we're working on adding more coverage to those products in terms of technologies they cover, in terms of rules they cover, in terms of use cases. So that those products can be broadly applicable wall to wall to all of those different customers. So we have -- I would compare the progress there to where we were with APM when it went public. We had been on that product for a few years. And we had -- we're at the same level of success given tech, where we had a few customers, very large scale that use us everywhere and a number of others that were still missing a few things here and there. And the way we're developing that product is also very similar to the way we've been building the APM product. It is -- our secret point is that we have thousands of customers using it. And they're using Datadog all day long, and they have all their data in it. And they give us all the traction we need to make sure we understand exactly what it is we need to do to get to the next stage with them, and we keep shipping those features. So the expectation is that within a few years, we'll have a best-of-breed product on its own right. That stands not only on top of the data platform, but that can also play on its own in that category.

So heading in the right directions. I'm looking forward to seeing more progress in security over the next couple of years. Datadog has a world-class go-to-market organization. You mentioned the efficiency of that -- of your cost of sale. One of the things that the company doesn't really talk about is relationships with the global system. I guess whether it's the [indiscernible], Accentures of the world. Do you need them to effectively penetrate your market opportunity? And what is the status of the relationship with the GSIs?

Yes, I think we do have relationships with most of the larger GSIs, as it's not as a source of professional service revenues given our platform, but increasingly, clients, especially the larger enterprise clients are looking for them as solution providers. And we do have a significant presence in having those entities understand Datadog and recommend Datadog and look at Datadog as part of the overall cloud migration and digitalization. It is something that is not necessary in all clients, but it's complementary, and we've been working on it for some time in getting good results from it.

Yes. And it hasn't been a blocker. Historically, we didn't necessarily see eye to eye with most of the GSIs because they were mostly looking for billable hours, and our product was very low friction and it was not the best fit from that perspective. And it hasn't been a blocker. We're in 37% of the Fortune 500, and we're growing and that's fine. But I think today, the situation has changed quite a bit. The GSIs are much more focused on, as David was saying, in generating long-term value for -- in terms of the broader cloud migration of their customers. And we're a big part of that picture for them. And so we're working more and more with pretty much all of them to make that happen. So I would expect us to do quite a bit more of it over the future.

As we sort of wind down the conversation, Oli, I wanted to ask you sort of the long-term question, which is 3 to 5 years from now, are we be calling Datadog an observability company? If you think of what you guys have talked about, it seems like there's a theme going from systems of alerting to a systems of action. I was wondering if you could lay out the vision for us in the context of getting more involved in workflow, intelligence and security.

Yes, it's interesting. When we started the company, we called ourselves a data platform because we thought monitoring was the world of the '90s. And it turns out, people were still buying monitoring. So we call it monitoring in the end, and people understood exactly what we're doing as opposed to what you've been at data platform. But the ambition was always to be broader than just this very specific category. So monitoring today is part of observability, which is a broader category. There's more we're going to do around security, et cetera, et cetera. But really, the long-term focus is to be the true platform that you use if you're building, if you're running, if you're securing applications or if you have to understand what they are doing for the business or if you're trying to understand the business through the prism of the application, which, again, in a company that is -- or in a world where companies are mostly digital, ends up being substantially all of the business. It actually happens with the application. So the point is to be the platform for everyone. What we mean by the platform is you start your day. You log into Datadog. You end your day, you log off the Datadog. And that's where work happens. That's what ties everything together. So it starts with the ability. It includes security. It's going to include a number of developer workflows going to include ITSM. We think it can include a number of things around real-time BI for the business itself. There's more we can do there, but we want to be that true platform. That's the -- that's why you see us do all these investments in product. That's also why or how we think we become the unavoidable consolidator for everything else our customers are using over time.

It's a pretty exciting vision. Let me go to the audience and see if there's any questions for the team. If you just raise your hand. We got one upfront.

You recently launched a real user monitoring. So what's the strategy behind that? Is it going to be in -- used in developer workflows to do debugging? Or it's more around the customer observability side to understand the business further?

So all of the above. So the -- for starter, it's an expected part of an APM suite, like APM and following the real users that end -- start and end every single transaction you'll find in customer-facing applications. So that's number one. But also to your point, what's interesting to us about it is that it's a very nice bridge to getting into behavior analytics, understanding what your customer base or user base is doing with your product and then getting into further business analytics from there. So we've been heavily building on that for that reason. Today, that -- so initially, that run product was used mostly by developers for debugging and understanding what was going on with the application. Today, we built more around it that appeals to the product managers that are going to try to understand why things are used or how they're being used. We also have a session replay product around it that is being used by support teams to actually understand when something went wrong with the customer, what happened to them. Where did they click on, what did they do there. So we're expanding that to get into other use cases and other types of users.

And with that, we're all out of time. Thank you, David and Oli, for an exciting conversation. Really appreciate it.

Thanks. Thank you.