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

We know it's an unbelievable conference, and it's Day 3. One of the last presentations. And extra-large room is almost full. So -- and we owe that to Datadog.

Thanks for having us.

Not us. It's you guys. And it's you too. So this conference, I was telling David that I never get tired of reading this. This conference, some of the stats, more than 3,100 [registrations], more than 500 companies and getting those 24,000 requests for just one-on-one anymore. It's many other ones or a small group on one. So thank you for your support. It's not possible unless we have the best of the best companies to present. And on that note, David, what a delight. I guess I should say welcome back to Goldman, right?

I know it's been a while since we've been physically. It's great to see so many faces that I've gotten to know on Zoom over the last 2 years but -- our meeting. So thank you for having us.

So I meant, you used to work at Goldman...

I did used to work at Goldman a long time ago. That was -- yes, I was an investment banker way back, yes. I admit it.

There are some of them.

Some of them around?

Yes. Some of them around, yes.

David, how would you describe what -- so if you came back to Goldman Conference, 2027, 5 years from now, what does -- what would you want -- what would you and Oli want Datadog to look like? What are your most audacious goals. It doesn't have to be a revenue target because nobody knows what the revenue is going to be in 2027. What do you want Datadog to be 5 years from now? And how do you get there?

Yes, good question. And we think about it a lot. So sort of the step back, Datadog has a -- am I coming through clearly? Is that okay? Okay. Datadog has an observability platform that combines infrastructure, APM and logs monitoring that is used by DevOps professionals to monitor customer-facing real-time digital applications. And so how do we think about this? We think that we are very early in this transition of workloads to the cloud and digital applications and that most companies are potential customers. So we want to be a centralized platform for professionals who are in DevOps to be conducting their business every day. And that includes monitoring that may include security that may include workflows that may include using BI to see what's going on in the business. But we want to be that platform that clients can get the value added out and constantly expand that functionality to be that one platform they can rely on to do their work in this area. And I think that's always been Oli's vision for Datadog and something that we've been relentlessly building towards through product innovation.

Yes. And I remember this when I met with Oli, I think you dialed in via Zoom initially, right? So you described the problem statement as just imagine the cloud is the next mainframe and all the things that you people did for the mainframe networking, systems management, database monitoring, it was in the small little contained world, but then you blow it up and it's the cloud, and that's a problem statement. And I heard them describe it that way. It just changed my perspective. The bigger the problem, the bigger, the solution, the bigger the company, right? So how do you guys think about what are the other problems or issues that customers face as they do this cloud thing that are opportunities? I mean maybe it's a product, maybe it's not a product, we don't know. But what is the angle of that lens that you're looking at? What are the other things outside of the purview of those 14 products?

Yes. And I think we agree with you when you think about what's happened, this is a generational re-architecting and in the complexity of the way software is developed, the speed and the requirements around that. And this has provided the opportunity for Datadog to get where it is today. And so when you think about how our client base in DevOps, these are, again, production environments, what are they doing in their day-to-day world, they are looking to see how the application is working, which we're doing. They are trying to identify problems earlier in software deployment, which is shift left, CI/CD and testing software going in. They want to be able to communicate, collaborate and order all of this information and remediate very quickly, that could be anything from BI tools, dashboards, scheduling, ITSM to we acquired a small company called CoScreen, which allows that group very quickly in real time to be able to go on a video service internal to the platform and remediate and meet. And then on top of that, it isn't just the design of the software and its deployment and latency, et cetera, what about security built into the software, what about instead of relying on only firewalls and gates, you're talking about looking into it and design and how it works. And that creates the whole security needs. So when you think about how in a democratized, ubiquitous, very fast moving, DevOps world, what will they need in the future, those are some of the things that we think need they need, and we're trying to engineer into the platform.

Got it. The unit of work becomes a -- not a virtual machine as in the past or a unit of compute, but it's a microservice and container. Then the whole world changes and you redraw all the boundaries all over. And so you mentioned the service -- video service monitor. Are you making a bit of a baby step into ITSM?

Yes. I think when you say ITSM, it's not the markets that, let's say, ServiceNow is doing, we're talking about enabling and making efficient the workflow of our world. And our world is real time. Basically, if the application goes down, if there's latency, if something happens, that's lost revenues. So everything we're trying to do is to increase speed and make the [ life. ] So I think if you think about that from the DevOps or the production environment professional, how are they going to fix problems, we're thinking about it that way. We haven't said, maybe we will, that we're going to be engineering and workflows and things like that. Part of that is because our problems need to be solved now. So it's really getting everybody on it and solving it. And that has a little bit of a different need than a service queue and it happens next week. It happens next week, there's no more delivery of the product, so you can't do that.

Got it. I wanted help us -- want you to help us understand what is exactly consumption about your [partly] consumption business model. If you just give a little primer. How much of it is fixed versus variable? Does it vary by product, vary by size of customer, number of nodes? Can you just give us a bit of a primer?

Yes. So stepping back, before we talk about contract, we're basically priced based on unit of infrastructure or host. That's one. There are a couple of different pricing models. So that would be the servers, the applications on top of that and some other functions. So that is sort of infrastructure-based pricing. And then we have data or testing type pricing, the amount of logs data that you're ingesting and indexing or the amount of tasks. So that's the basic denomination. We're not pricing based on compute or what happens in the application. Then contractually, most of our relationships are subscription contracts. They tend to be around a year. And they essentially are based on the amount that you're committing to and then they have an on-demand. And this is a little -- AWS and others, the hyperscalers do that. There's 2 major types of this. One is a take-or-pay contract that is denominated by, let's say, month, meaning if you don't use it in that month, you lose it and the other might be by year. So essentially, it is a usage-based model particularly in the denomination by year with an underpinning of a subscription that if they don't use it, they lose it, and we recognize it. So it's a hybrid model. In addition, we have -- if you go over that amount of usage, you go into on-demand, most of our customers, because these are greenfield rollouts stay a little short, then go into on-demand, and that's when they fix out. So we have that, which is associated with subscription, but is priced above that subscription rate.

Got it. So if a customer does not use the product all year long, I don't know why that would be the case, just hypothetically, do you recognize all of the revenue in the last day of the year or you go back and -- how does that work?

No. We -- for that, if they had all year to use it, so it was $1 million over the course of a year. And by the way, this is really rare, we have very few of these. But in that extreme case, that would be the recognition of $1 million in the last day of the contract.

Wow, bounty.

But -- but yes, but that would be -- that is a -- I mean I don't know of any case like that. So most of it is you recognize it along the way. The major motion because these are new workflows is that a client gets to 6 months, and so I'm getting near my limit. This is what normally happens. And then they do and extend, commit, et cetera, in order to avoid staying in on demand. But yes, if they did not use it then they -- it would be recognized in that period, Kash.

Got it. So in an economic environment like the one we're going through, a lot of companies, I know Snowflake's business model is very different than yours. MongoDBs very different than yours. But hypothetically, if somebody entered into a contract and their business was going through a bit of a tough time. But then their business started to pick up. The revenue recognition would be more volatile on your part, right? or...

Yes, it would be more volatile a bit. The type of contract that is denominated within the course of, let's say, a year would be more volatile. I mean it'd be slower and then pick up. If it's denominated by month, it would be more pro rata. So it depends on the nature of the contract. But yes, there is, I would say, the biggest volatility and -- would be the pace at which revenue is recognized off of usage in those contracts denominated over let's say a year.

By the course of the year. If economic activity starts to pick up, then things look -- there will be a positively variable outcome to the yearly contracts.

Yes, that's what happens essentially. That's the variability. For instance, we said in COVID, we had a pause. Many of us had a pause. And the consumption slowed down suddenly and then rebounded and revenues and usage in that period declined -- well, not talking about the declines, we're talking about the rate of growth of usage. This time, we said that in the second quarter -- and by the way, I'm speaking comments we made on our earnings call in early August, we said that we did not see that, but we saw some areas of more controlled usage, and we can go through that, resulting in sort of a middle ground.

Yes. Yes. So not like COVID. We're talking about a slowdown. I think quite a few software companies that have gone through the cycle of, oh my God, deal slippage, whatnot have come up with a more reliable forecast, maybe lower close rates, better pipeline and are sounding better coming out of that quarter when they reforecast, they sound more confident. How are you -- how did you approach the process of forecasting. Everybody got conservative with their forecast. What are some of the assumptions that you've made in your forecast for Q2? You [ did ] raise guidance, which was fantastic. But what were your assumptions regards the close rates, pipeline build, et cetera?

Yes. Well, sort of step back, what we said was that we saw a strong new logo accumulation. So we continue to see strong pipeline and new logos in our land and expand. We saw, I would say, the same amount of strength geographically, but we saw within our organic or usage, some weaknesses, some larger customers, particularly in consumer discretionary. So that's the background of what we saw. We always, and this is since we went public, provide guidance that has lower assumptions, significantly lower assumptions in both usage and in new logo accumulation than we've experienced historically. This is -- we've always done this. COVID, non-COVID, accelerating business, recession risks, et cetera. And we continue to do that. So we adopted a similar -- in this case, we pointed out that normally, we pass a little more of our beat through. And I would say, are not towards conservatism, this time was we did not put through as much of our beat. But as you said, that still resulted in raising.

Yes. Yes, exactly. So I suppose you sort of answered my impact of the economic environment. Qualitatively speaking, rates yesterday was a bad shock with inflation. And I know that you and I spoke after the quarter, and we talked about how some customers decided that they're going to hold on to their pocket but not sign a longer-term contract, that RPO growth slowed down and that revenue growth was actually pretty fantastic. After the call, I wondered to my myself, if that customer is really worried about inflation, they should be doing the longer-term contracts, but they didn't. And so as they realize the cost of capital is going, the option to [indiscernible] holding on to your money is just really, really bad tradeoff. Do you think, not just Datadog, the industry could see resurgence as these people say, let me sign these long-term contracts, you know what, I don't want to take the risk of some variability, pickup in consumption or price increase, whatnot. Why are customers thinking exactly the flip side of how they should be thinking?

Well, I can't comment on other companies and what they're seeing. We are land and expand and because our clients have been expanding over a long period of time, we're also not out there trying to get 3-year commits and things. So essentially, the question for us is in this normal cycle of you land, you scale up, you see what your volume needs are, why would they stay in on-demand, which complements revenues and around the edges, but rather than commit. And I would say decision-making maybe has taken a little longer because they're more cooks in the kitchen, people like me...

On-demand is like consumption, right?

On-demand is their consumption above the limit of the contract. And many companies, we do that with -- we do this with our own cloud consumption, we stay a little short in order to risk manage and you pay a little higher price for that. So I think that it's logical that it's a confidence thing as they get more confidence in the growth rate of their business, they will commit more. But I would say, it's a minor effect for us. And I think we said that in the subsequent quarter, we had seen some of those that stayed a little short, paid a little at the higher price, do the commit. So we'll have to update everybody as we see. That comment was very much more about billings than it was about revenues, as you said, because what happens, it affects billing, it's when the bill goes out the door rather than the revenue recognition. And we were trying to explain some more volatility around billing versus revenues.

Got it. Yes. So you're not alone because we had Snowflake and MongoDB that also were subject to barrage of consumption questions. Although Frank Slootman got away with it because he was -- his fireside was done by our CEO. So we didn't get too deep to consumption, it was very high level. But so let's say Datadog is compute, MongoDB is transactions, Snowflake is analytics, would you say you're more leveraged to compute and less to transactions and analytics like the other guys? Is there a way to think about the layer of that stack compute, transactions analytics, one way compute is more stable than transaction, maybe less stable analytics second derivative.

Yes. I think that long term, like we're levered to compute in that -- the number of servers. But short term, we're not priced. I just want to make sure everybody, we're not price based on compute in the variability. And so I think when you think about infrastructure, a lot of our products are priced on an infrastructure mode. The elasticity and the changeability of that in the short term is not as high if you would -- if you were pricing based on compute as it's used. So I want to make sure that's clear. And I think that -- what we've seen is that over a longer period of time, if a client's business is growing, they'll need more servers [ and were ] triggered there. So I think it's not as volatile in the very short term, but over the long term, how quickly the clients' businesses are expanded, could be as -- could affect that infrastructure and APM side. On the other hand, a lot of our products are priced based on remediation and what's happening if things go wrong, are you doing logs? Are you doing tests? And I don't think that recession sort of gives a path for, unfortunately, for things going wrong in your IT environment. So there's likely to be continuation of that. On that side, it has to do with some of the variability we talked about of how many logs you indexed or maybe how much you store.

Got it. I'm not going to name a company, but you mentioned APM. So we had an APM company here that said when we asked, how do you differentiate your products. And the answer was you guys do dashboards, which are very pretty to look at [ or as ] they offer solutions. Do you -- would you agree with that, that you're just a dashboard that this other unnamed APM company actually offers the solution and this is what our AI engine does and therefore would fix the problem?

No. I think that we have in our platform machine learning and AI.

Tell us more about this.

So it's called watchdog setter. And basically, what I think we offer is we offer a distributed, sophisticated but easy-to-use platform that has all the functionality in it. When you think about -- and it is easy to use across the organization. When you think about a couple of different models of software, you think about maybe on-premise or deliver on the cloud, but used by super users, so there's the group. And then you have something like salespeople would use Salesforce. They're using it every day in their business. They turn it on. Everything is there. That's Datadog. So if you're going to want your solution, your observability platform to be used by everybody who impact the software and the production and delivery, and you're going to want to increase speed of resolution because it's all there. And you're going to have both dashboards and some AI in it, that's Datadog. But we believe -- we don't believe that the IT world was going to be one of robots. We believe that there's experience and data and automation of data that is then going to be used by humans and that a lot of times, these pre-canned solutions give false signals. So we believe it's important to have intelligence in it, but it's not really what we believe, it's what our clients believe. Our clients believe that they want to get the information and then be able to resolve and that one size doesn't fit all. And I think Oli has said that to you and others. He said, he doesn't -- he believes that there will be -- this will not be handled in an automated one size fits all because our clients' applications and environments are not one size fits all, and that's the power of Datadog.

Got it. I want to talk about TAM and get your thoughts. So the CRM market in 2003, 2004, the famous, not to be named industry, funded very well recognized brand name said, it's $0 in TAM. And then several years later, they came out and said when ServiceNow has gone public, ITSM maybe $1 billion. And subsequently, everybody started to revise their TAM assessment, oh CRM, oh yes of course, it runs in the cloud. Why would you buy anything that does not run on the cloud. Where are we with the TAM assessment in your core markets today?

Yes. I think our TAM assessment is -- and the path of it is related to the path of both cloud deployment of workloads and the creation of digital solutions. And if you look at the percentage of workloads that are in the cloud and the percentage of applications that are delivered in that way and created through modern DevOps, it's still quite small. So...

It's outpacing cloud growth.

Yes, it's outpacing, yes. And we think that...

And that is your growth strategy.

We think that our growth is the -- underpinning it is cloud. But on top of that, you have deployment of modern development techniques, containers, micro services, et cetera. and you have the whole trend of DevOps and the democratization of data down to the users. All of those are growth drivers, and we think we're pretty early in it. So yes, we have TAMs that are created off of the major research, et cetera. But when we really look at it, we really look at those -- that percentage of the workloads that are in the cloud and what's going to happen in digital applications. And we still think we're very early in this huge opportunity.

Yes. So the question I should be asking is then somebody puts up cloud growth rate [ effects, ] what is the growth rate of your transformational workloads versus migrational workloads? If the migration slows down the transformation picks up, we should not be concerned, we should be -- I mean, concerned on Datadog's...

Yes. It's interesting to think about. I do think that the migration of static or, let's say, Microsoft Office, Microsoft suite into the cloud, has less effect on Datadog then that complemented by the use of modern DevOps. I think one of the biggest triggers as to whether they're going to be a Datadog customer is what is their software development strategy. And whether it's based on containers, microservices and whatever comes next.

The inside of the data, the inside our -- that's exactly...

Yes. That's exactly, yes.

David, do you have a rough feel for what percentage of cloud workloads are transformational versus migration. I mean I've heard some statistics from people, but I just wonder if you have a perspective?

I don't have, I would love to hear from, you probably have a better...

Yes, I heard like 15%, 20% is transformational and the rest is migration. So if that's true, then the transformational stuff is probably outpacing...

I feel like if you look at Datadog and you look at the intensity of the kinds of things we are monitoring, it feels like that transformational and not just the repotting but -- is growing faster. And I think certainly, Datadog's growth rate is probably indicative of that, although I can't say that we have -- we're not the research factory to -- for the world. Yes, I think it's a good observation.

You recently called out APM, log management as exceeding $0.75 billion in ARR. Those are some big numbers. They've been around just 4 or 5 years. What is driving success?

Yes. Two things are driving -- and by the way, we -- like you said 4, 5 years ago, we didn't have this. So I think it's 2 things. One, the compelling nature of the platform and everything being together. Our clients, we often say are buying various signals within one product, the platform. And so having it all together is a very strong value driver. So I think that's one. And 2 is us that we have really invested substantially in the product suite, not just individually but how it all fits together. And that investment has caused not only the pieces to be the best out there. But then the way they interact with each other in the platform is a very differentiated product.

I remember we caught up maybe a few quarters back, and you made the -- I said, how is your cost of customer acquisition so low? And why is it that your sales and marketing is so low relative to -- and you said because the products are so good, they sell themselves on their own, right? What is it about development in France that is so more productive than what we get here? Is there some secret sauce?

Well, I think it all goes back to...

Is it the Burgundy wine or...?

It could be the wine. It could be that -- there's sort of an aroma ahead that I could download with Datadog. Listen, it goes -- I've been at Data for years, and I happen to be lucky enough to join a company that was founded by Oli and Alexis and then Ana came. And they just relentlessly thought about the customer experience. And they often will say it's really how easy a product can be used that creates the value, not necessarily every widget. And this has been the bedrock. So I think that's enabled us to expose the product, get feedback and have it frictionlessly adopted and get that mind share. Once you get the mind share in an end market, you have your customers be your salespeople. They want more. And if anybody threatens their Datadog use, they say, no way, we got to have it. So I think that -- and it all goes back, I think, to the genius of the platform design and then putting a go-to-market against it, that was very consistent with that type of adoption.

That's great. Shift left, if you shift left, there are others that you're going to encounter, like GitLab, GitHub, there's a whole chain of SDLC companies that are waiting there and their view is that moving left is really, really hard, whereas for us to move right into ops is not that hard. You would content otherwise. Who's right here?

Well, we think it's hard to -- we think it would be hard to shift left, too. Like we're not -- so maybe we're seeing something different than them. Like we are not saying that we are targeting cold repository or essentially the act of software creation, where we start to come in, we think, is the border, where software is created and you're in test mode into production. And we think that makes sense for us because really, we're at the interface already of software as it gets developed into production. So I don't think we would be -- I don't think we would say that it would be easy to shift all the way left. We think that there's tangential market opportunity that makes sense for our customer base slightly left, and that's what we're aiming at.

Got it. Any -- I know you're a CFO, but you've been at the company long enough, and you're in the technology industry. So the competitive differentiation relative to the others in the space, we sort of touched upon the consumption model, the architecture, but it does look like everybody wants to be an observability company. Everybody is an observability company, forget wanting to be one. Do you hear that from customers, are customers coming back and saying, Oh, yes, I got to switch from 3 other companies. They're also observability, you're also observability. So I'm confused here. Do you ever get that? And if you do, how do you proactively tell customers how and why you're so different from every other half a dozen observability companies?

Yes. I think it's important to think about our core market. And when I mean our core market, I mean DevOps employing modern development techniques in container, microservices, et cetera. Real-time customer-facing with an emphasis towards having it used ubiquitously so that remediation can happen very quickly. Now -- you might say that's a specialized use case, but we don't think so. That is what any company that is going to have digital applications in the cloud needs to do. And so when you think about that, the real competition is open source do-it-yourself or some point solutions. And we've already said that we think our clients have told us that putting everything in a unified platform makes sense. So it really depends whether you say observability is this. You can say observability. I want to monitor on-premise and cloud. Well, that's not our market. That's -- and there might be others that might do that well or maybe they do on-premise. So I think it really depends upon the core market. We think we have a very high-growth market where everything is going and a market that finds value in this kind of monitoring. I would say that, it may have been more difficult, I think, in years past. But as Datadog has invested in its product become more of a name, has a very large user base and has performed for so many years, I think all of that investment is paying off in more sophisticated appreciation of Datadog's use for this particular use case.

I was wondering if you are the Chief Product Officer, Chief Marketing Officer and Chief Financial Officer, all rolled into one.

I love products. I must admit, I'm a Chief Financial Officer who knows where my bread is buttered and it's in product and selling it.

You're getting into it.

But -- no, I think that's it and -- yes, I think that there is confusion, there is marketing, there is -- but when you really go through the haze, you really see that there are solutions or platforms that are really, really best of breed in this particular use case. And in another use case, there may be another one. And as you think about that, I think our customers increasingly understand the difference.

Yes. I have just one more question before I turn it over to our clients. Maybe they're all too exhausted at the end of the day to ask a question, I don't know. But one thing that we have seen is logging is not a finite market. Practically every observability company that came through the Goldman conference has been talking about how surprise they are that logging is, it's like you're driving on this highway, you think you'd get to the end of the horizon and then opens up again. So logging is bigger and bigger. What are your observations on the logging market?

Yes. I mean, definitely, when you think about it, once you get the data to investigate the problem and understand you need to index and look at logs. So it's a very good complementary. And we have been surprised at how quickly that market has grown for us. I think our differentiation is that you can't just say logging. We're -- have logging that is deeply integrated with the other signals with the metrics and traces. And it's being used and think about the speed required to remediate. So that's where we differentiate. I think we've done a very good job in building the logging product, integrating it and designing a pricing that has really been very closely correlated to the client's indexes and use. But yes, the logging market had has very strong demand. And it's because it's one of the most important ways to do your detective work and figure out what's wrong.

Exactly. Anybody has a question feel free to raise -- wow, we got questions. You got answers?

They are still alive, yes.

Okay. Okay. We'll go to the gentleman in the light blue sweater first, if you don't mind, and then sorry...

Thanks very much for the time.

Jeremy, right?

Yes. Just in terms of the shift from DevOps to DevSecOps, how are you thinking about that sort of new market and what's your sort of security thinking around that side?

Yes. Good question. It's a new market. We think that it is essentially the bottoms-up use of security in a distributed way by DevOps. There are some progressive DevOps departments that are doing that, but it isn't the mainstream. What we're doing, making sure we talk about our product is, we're focused on that overlap. So that would be cloud and application security for that end market using -- looking at security signals designed in. So it's not endpoint, it's not network, it's not, et cetera. And we think it is a nascent market. There will be quite a few developments just like there was in DevOps in innovation of practices. We see some good signs. We have a lot of customers using it. But it's early on, and we are also midstream in building the product. So the combination of the market and us building the product will allow us to take it to market more aggressively and get some feedback and answer your question more comprehensively as time goes by as both those things happen.

I think his question is going to be similar to mine, but I was wondering in terms of -- I think you're referring to the application security testing product, you guys are building something now to compete with like a Veracode or Checkmarx versus maybe M&A, those guys into the portfolio? Or somebody like that?

So we have testing of the performance of the application from back end out to the website. We're not -- if you mean a testing market that's outside of that and in the use of monitoring real-time DevOps, that's not what we're doing. And if you're -- and on the sort of shift left side, we're trying to use the same signals we have in our platform to allow the intersection between testing and production see signals a little earlier. I don't know if that answers your question, but that's our strategy in testing.

We have time for perhaps one more question, one final question. If not, I'm going to jump in, we're out of time, minus 1, minus 2, minus 3. Okay. So maybe I'll ask one. Long term, what drives the free cash flow profile, operating margin profile, I know you don't -- you've not given long-term guidance. But what are the things we should be looking for that are good guides to the potential -- free cash flow potential of the company?

Yes. It starts with the gross margin and our ability to continue to innovate, yet maintain or maybe even grow gross margin, which we work hard on. And then to understand the scaling of the different lines, I think so far, we've seen some scaling and efficiencies in G&A. We do have a very efficient go-to-market. But we're investing in R&D, and we like that. So I think it's looking at all 3 of those and the timing of those that get to the art of the possible in margins.

Wonderful. That's spot on. Perfect. Thank you so much, David.

Thank you. Thank you.

Thank you for being here and participating in the conference. Have a lovely evening. And don't miss the reception drinks on me at 6:00.