Dynatrace, Inc. (DT) Earnings Call Transcript & Summary

All right, guys. So Ryan Mac, small and mid-cap software analyst here at Wells Fargo here for the 9th Annual Wells Fargo TMT Conference. The joke that I've been making, this is my third fireside on the stage is we should open the doors up. So I mean you just get a view of the ocean.

It'll be much better. Let's do that. That sounds good, although it was pouring down rain earlier, but now it's nice out.

That's what people keep saying. I mean, people are still used to perfection of how great it is here. This is my first year here. So I'm just like taking them all in. I'm blown away, but...

Pretty nice. You need to do the bluff path along the water.

Yes. And people have been telling me about the hike and everything. So I'm coming here for the weekend before next time. That seems like the pro move.

I think that sounds great. Now everybody is going to leave the room and go for a [ ride. ]

I mean...

Can't leave until we're done.

Maybe 1 year, we can just do a hike fireside.

[indiscernible]

Yes, I'll walk and talk. So look, guys, we'll be taking questions directly from the room, but if you have them, you can e-mail me at [email protected], and we get them in. Here from Dynatrace today is Rick McConnell. Rick, thanks for being here.

Just to start, we'd love to kind of hear about your year so far, a lot of change with AI, a lot of change with some of the go-to-market changes, but -- and something like the ODC and DPS. But overall, I would just love to hear kind of at a high level, like maybe what surprised you? What are you pleasantly surprised with for the year so far?

Well, first of all, very strong first half of the year. And I think the core messages from our Q2 earnings report that we just completed a couple of weeks ago were number one, really strong first half; number two, raised guidance for the second half; and number three, largely derisked that second half. So we felt good about those core messages. In terms of the core drivers of the business, we like what we see going in. Log management is a key emerging business for us, now very close to $100 million in overall consumption and growing at north of 100% per year. Well, when logs was growing north of 100% per year at $20 million on a $2 billion business, it has limited impact. But $100 million consumption business growing at 100%, a much greater impact as we expect going into the future and a huge beachhead for us as we look ahead. The overall strategic pipeline, up 45% year-over-year, very strong. So we feel good about the pipeline at the largest end of the segment, which is where we're really seeing the end-to-end observability and the tool consolidation. Our pricing method of our Dynatrace platform subscription continues to grow. That's now over 50% of our customers, 70% of our overall ARR. So we feel great about that. And consumption growth is super strong, more than 20% and that we believe to be a leading indicator of future net new ARR. So core growth drivers are in place. The business is strong, very healthy as we look into the second half and customers are expanding. And I suppose the macro [ uber ] comment at all is that observability is becoming more and more critical day by day, especially with the expansion to AI workloads, you just -- you need observability. It's no longer optional. It's become mandatory.

Yes. A lots to work with there, but I think the last part is really interesting. So can you just tell me in your customer conversations, like how do you feel like there's a higher priority place on observability? Like what does AI have to do with that? Like how does that change, I guess, versus like a year ago or in prior years?

Well, many different ways, I would say AI is changing the calculus on observability. The most basic one is that more workloads mean more observability. So you really as a -- especially in our target segment in the Global 15,000 and at the high end of the Global 15,000, you simply can't manage software workloads manually anymore. The number of resources required to deal with alerts, to manage those alerts to then triage those alerts is extensive. So the result of it is you need to be driving toward a methodology of automated processes to be managing software workloads. And this is where Dynatrace, in particular, comes in with our overall environment, which we can maybe get into, but some of the technologies underlying Dynatrace and the Dynatrace platform really lend themselves well to delivering an AI-powered observability platform that is increasingly driving autonomous operations. And I've been to -- I've been privileged to have been in, I don't know, a dozen countries in the last 3 months, meeting with many dozens of CXOs. And I would say that not only is observability top of mind, but figuring out how they can leverage AI to automate outcomes to drive lower cost and better success in software are very top of mind.

I think that would be a great place to start on the AI theme is in terms of the Dynatrace architecture, you've been doing observability at the largest of scale for some of the biggest of customers. So as people think about like where can they start on their own AI journey with the products that they're coming out with, how are your existing customers, one, looking to utilize you today? And then two, like what about your technological background makes it a better choice to start with for your customers?

Yes. A partner asked me the other day and this particular partner was of the mode of, geez, we're -- we've sold AI into maybe 10% or 15% of our customers and the actual deployment of AI is maybe 10% or 15% of those. And his question to me was, what percentage of Dynatrace customers use Dynatrace AI? And my answer is 100%. And it isn't 100% using agentic AI with autonomous operations where everything is auto correcting. That's still to come. But the foundation of AI, causal AI, predictive AI, generative AI, all analyzing an integrated data lake house of content, including logs, traces, metrics, behavioral analytics, business events, et cetera. That is then being utilized to deliver answers, not just dashboards, not just red, yellow, green alert, but this system broke here and you need to fix it in this way. And that is becoming more and more crucial day by day. And so -- and it is that AI engine of Dynatrace that hasn't existed for 1.5 years or 2 years or since AI became a thing, an exciting new thing. It's been around for more than a decade. If you then add to that the next phase of AI for observability and Dynatrace, it gets super exciting. And we can go there if you wish. The short form and to keep it as simple as possible is observability has gone through a number of phases. And I would say we're now approaching Phase 4. And to cover the first 3 phases, simply, I would do it in sort of single adjectives. Phase 1 was reactive. Something broke, you tried to fix it or you tried to figure out what was wrong and then you try to fix. Phase 2, I would say, was proactive. It was about automated root cause analysis. You wanted to know the second something broke, you wanted to know what was wrong, but you had to manually fix it. Phase 3 was predictive, and that's really where we are today. And that is taking all of the AI elements that we had before and applying additional machine learning and anomaly detection to anticipate issues to be able to resolve them before they became user impacting. Phase 4, though, is really where we all get super excited at Dynatrace. And this is around autonomous operations, as I mentioned. Well, what does that mean? That means integrating this predictive element into an agentic AI ecosystem that is essentially focused on delivering auto prevention, auto remediation and auto optimization, these 3 things. And in order to do that to auto prevent, auto remediate, auto optimize, we recognize at Dynatrace that we are a core, a necessary input to that equation because we can tell you precisely what's wrong. But we aren't always the solution to fix it. You may need to provision more storage on AWS. You may need to do a rollback of an application that was posted that had a bug in it, and that may be an operation for a GitHub or a Jira. You may need a workflow integration, and that would be with ServiceNow. So in this autonomous operations environment, what gets exciting is that we provide a core foundational element to what needs to happen, posted into an ecosystem of agents that then can take action through any number of these partners to then resolve the issue. And maybe a final point on it is you say, well, you're never going to be able to resolve 100% of the issues in an autonomous way because you won't trust the answers. That is true, but I've had so many CIOs, so many CXOs tell me, Rick, if you could resolve 20%, 30%, 40% of my incidents, you would save me, especially in our target segment of the largest companies on the planet, you would save me tens of millions of dollars.

And who doesn't want that? I mean, as you go through the phases of observability historically, right, as you kind of outlined where you went from reactive to proactive in terms of like how to think about your observability footprint, like what does that mean for how do you think about like which vendors you use or one platform versus best-of-breed for observability? Because it seems like to me in the dynamic you outlined on a go-forward basis, right, you almost want to have like one end-to-end platform in order to do those use cases. So do you think that changes with more focus from AI?

Well, this is a lay-up question that's going to sound self-serving an answer. But yes, I mean, the short form is we have seen material consolidation in observability. It used to be the case that you would have one vendor for application performance monitoring, APM. You would have another one for log management. You would have another one for real user monitoring. You would have another one for synthetic monitoring and another one for infrastructure and so on. And we -- there's a large airline comes to mind that we closed last year in the summer, and they had one of everything. And when they started on the journey with Dynatrace, the principal there even told me, gosh, I can't even imagine deploying Dynatrace because I've already got one of everything. How is that going to solve the problem? And in the end, they weren't getting the outcomes they wanted. So they ended up with an end-to-end observability road map that basically was Dynatrace, and it eliminated or supplanted all these other tools, deploying Dynatrace as a single integrated AI-powered platform instead. And that gave them much better outcomes than they otherwise would. Well, why is that? Well, if you have the underlying data types of observability, things like logs and traces and metrics and really user data, and they're all in different data stores managed by different vendors, then you are out of necessity managing those on a manual basis to try to derive insights. And the more complex the business, the harder it is to drive those insights and the harder it is to piece things together. And God forbid, you have multiple incidents at the same time. And then what do you do? And how do you cross-correlate this data? It's impossible. With the Dynatrace platform, we have a single integrated data lake house, which we call Grail, which houses all of those data types. We oversee and analyze that with a single AI engine called Davis. We evaluate that in the context of a single overall IT ecosystem topology called Smartscape. So the result of this is by providing a completely integrated platform, you have a comprehensive integrated picture that you can then act upon. And in fact, this is the motion that we see our customers driving and especially the larger the customer, the more complex the environment. The more complex the environment, the more they need a single integrated platform. And this is the directional heading. And if you look at a Gartner Magic Quadrant or a Forrester Wave or others over the course of years and you trended it, you would see that the point products are falling down into the left and platforms are very much present in the leader quadrant in the far upper right. And it is because, first, that's what customers are actually doing, they're betting with their wallets. And secondly, that's where you get the best outcomes.

Yes. I mean you always hear in software like, oh, we want one pane of glass. And it makes sense in theory and why it would be nice. But when you have like a more reactive use case, right, like archeologist, then it's okay to go platform to platform or vendor to vendor. But if you're trying to make insights and forecasting, right, that seems a little more difficult. We'll come back to the AI piece in a second. But just on the consumption rates that you talked about, around 20% consumption growth rates that's higher than the current revenue rates. I'd love to hear how that's trending and what's kind of driving that higher level of consumption?

Well, consumption rates are up quarter-over-quarter. So our consumption rates are growing in part fueled by the log business that we discussed earlier. And those consumption rates are, we believe, precursors of the opportunity for acceleration of growth in ARR and subscription revenue. And that is what I wake up every day thinking about is how do I help the organization as a whole to reaccelerate top line. And that is really with a keen focus on expansion of net new ARR. And the good news is we saw that with an acceleration of net new ARR in the first half and in particular, in the second quarter, a 16% net new ARR growth was very, very favorable result in the second quarter, 14% for the first half, very strong also. So those -- that really is the fuel to a reacceleration of top line. And why consumption is important is it's important to remember that we report revenue on a ratable basis, not a consumption basis. So consumption growth doesn't translate to revenue growth precisely for us. It is with a lag. So if the platform is being consumed more and more and more, then ultimately, that should lead to a convergence of those rates and that ultimately upgrades and ARR and expansions would catch up because customers can't consume in, for example, the mid-20s of consumption while they're only growing revenue 15%. I mean, at some point, it converges. And that's the expectation is that the faster we can drive consumption, the more demand there is for the platform. The more demand there is for the platform, the greater the opportunity for ARR growth down the road.

And that's the leading indicator you're probably most focused on, right? Like how much people are...

Yes. We've got 1,400 people in our services organization who don't think about net new ARR. Those people in our services organization, customer success, services, deployment, business insights, business metrics, these people are focused on consumption.

Excellent. So as you talk about a quarter-over-quarter improvement on consumption, what are some of the components driving that? Is it a better macro? Is it your customers doing more maybe with agentic AI? Or what is -- or like maybe they're adding more logs? Like what are you seeing to drive that higher consumption rate?

A combination of things. Number one, just growth of existing workloads. You need more servers, more logs, more whatever. Second is the inclusion expansion in new workloads, and those could include core level new workloads, consolidated workloads from other vendors or AI workloads, net new AI workloads. And then a final category might be new products and logs would fit in, in that category, for example, of where we see logs growth at 100%-plus growth year-over-year in terms of consumption, all of a sudden becoming material, and that really is driving new opportunity for consumption growth as well.

Excellent. So I mean, consumption definitely seems like the metric on a leading indicator you should focus on. Investors have been looking at net new ARR and ODC revenue and subscription revenue. But as we think about the most recent quarter, there were some dynamics where like you had customers do early renewals, and that's great and net new ARR improved, but then maybe ODC was a little lighter than investors were looking at. But that's okay. I mean you're encouraging the right incentives for your larger customers. But can you just talk about some of those early renewal dynamics and what investors should think about there?

Yes. It's funny since the earnings call, we certainly have got investors that have come back and that have said, wow, ODC was light, which is your on-demand consumption revenue, which is essentially overage billings. And I would say our response to that is sort of twofold. Number one, light meant $1 million to $2 million on almost $0.5 billion of quarterly revenue. So...

We're nitpickers here. That's how we go.

That's rounding here. And you always want to exceed expectations, and I've got it. I understand that. But we blew out ARR, net new ARR relative to even internal projections, let alone guidance. And it was simply a quarter in which the cohort of customers that were renewing in that quarter decided that instead of paying overages, they would expand. Well, every cohort in every quarter is going to be a little bit different. And we can't predict who's going to decide to pay overages and -- for a couple of months before going into the next year cycle and who's going to renew early. Overall, I would say, if I had to pick, I'd rather have the ARR than the ODC because ODC is, by definition, a onetime event and ARR is recurring. So an expansion -- an early expansion by a customer, let's say, 1 or 2 years into a 3-year contract is a very, very healthy signal that they are intending to stay with Dynatrace, intending to expand the deployment with Dynatrace and probably going to be increasing consumption down the road rather than as a one-off with just overage revenue for a month or 2. So if we had to pick, we'd like to see the ARR conversion versus the ODC. But ultimately, it's the customers' choice as to whether they want to pay a month or 2 of overage versus expand, and we give them that option.

And so what do you think is driving that decision to renew early? Is it just they've realized, okay, we're going to have a lot more utilization? So...

If customers are expecting, and this is why the dynamic toward net new ARR is preferable, let's say. If a customer is expecting and willing to commit to a higher overall contractual commit through a DPS agreement, then they should expand early. And the reason is because they're going to get lower unit price. So it certainly is the case, the more volume you do, the lower the unit price. Now the commit will go up. So maybe your original DPS contract was for $1 million a year, and now you're consuming at a higher rate, you go to $2 million. So you're going to spend more, but you're going to get more than double the capabilities off of that 2x improvement. So you might as well take advantage of the lower price point, increase the commit, make that higher commit. And the result is that you're going to move to an ARR expansion as opposed to an ODC single payout.

And so as I look at ODC versus the new ARR, that's just like a timing thing in regard to...

Just timing, I mean, ultimately, ODC as long as we -- ODC obviously ends up in subscription revenue. But to the extent that you continue to expand, that is a sign in and of itself that you have exceeded your commit for that 1 year. By definition, if you're paying overage in that year, you exceeded the commit for the year. And that is a healthy indication any way you look at it.

That's a pretty healthy signal. Yes. And since your DPS customers, 50% of your customers are on DPS, but 70% of your ARR, as you think about who's doing these renewals, those could be bigger customers as well. So like that would exacerbate the timing dynamic there if they had an early renewal.

Absolutely. And the largest -- the larger at this point on DPS contracts is because that contract vehicle gets them access to the full platform.

Excellent.

It also makes a lot of sense to be on a DPS contract for seasonal fluctuations. You take a look at commerce. And I mean it's timely because we're coming into a period now of the holiday selling cycle with Black Friday and Cyber Monday. Well, you look at our large commerce customers and their requirements for observability explode in the calendar fourth quarter. And if you have exclusively a server-based pricing model, then how do you manage that? I mean do you quadruple the number of servers you need for a quarter and then -- I mean you can't do that with us. DPS enables you to do that because you basically just sign up for the platform for the annual period. And you have factored in the fact that Q4 may be -- calendar Q4 maybe a large multiple of the other quarters, and then that enables you to grow, to expand and then contract. It gives you much more flexibility. And that drives greater consumption as well. That flexibility enables our customers to expand way beyond what they would have done otherwise.

I'm going to move on from this dynamic, but just so I'm clear here. So like if you're that type of like retail customer running into the holiday season and your consumption is already running pretty hot, it might make more sense to like, okay, let's do the contract renewal now because we already know we're...

I mean, if you're coming up on the holiday selling cycle and you're running short on your contract, you might as well expand.

Yes. That makes sense to me. So just on one of the comments you made earlier about the log business. I would agree with you, at $20 million growing 100%, it's like that's great. I'll see it in a couple of years. And then now we're here. So -- but still...

That was our fastest business to $100 million. So I mean, we really just began selling logs in earnest in October of 2024. So it's really been just about a year that, that business has grown to that size.

So can you talk me through like what you think the next 6 to 12 months looks like for that logs business? Like it's a higher ticket item, right? So are you winning from a competitor here or your customers looking to do like get their logs under one roof with AI? Like what's kind of driving like the next leg of growth for logs?

There are 2 primary drivers to our logs business. One is cost, which is -- I had a CTO of a large Australian bank tell me that their existing logs overall cost was meteoric. I mean the word that he used was meteoric. And yet they didn't feel like they were driving any additional value from that. They're storing more logs, they're ingesting more logs, they're querying more logs, but not a lot of incremental value. And they very much wanted to get a handle on that. And they need to find a path where they can manage the cost trajectory. The bigger element, though, I must say, is what I talked about earlier, which is your question, Ryan, on platforms, which is I'm not sure why, and I've been in this industry with Dynatrace for 4 years, not 10. But I have no idea why logs grew up on the one hand and observability data types and others on the other. It makes no sense. And the reason it makes no sense is what I described earlier, which is it is when you have all data types, inclusive of logs in one integrated data lake house that you can deliver the best outcomes, the more precise insights that are most actionable. If ultimately, your objective is really autonomous operations, that's when you need precise insights, you need precise answers. And to do that, you really want logs because logs add value to that. Moreover, it contributes on the cost front, but you actually don't need to store as many logs if you have trace symmetrics in many of the other data types because the incremental insights that you get out of all of the other data types combined actually helps compensate for not needing to have a log for everything. So it actually consolidates the number of logs that you have to track. It makes it much more intelligent. It gives you a better mapping and it drives better outcomes. And the result of all that is that our log business is growing because of those dynamics. And if we can save you a fair bit of money, and we can do so by driving a more integrated developed outcome out of an overall observability plane, then you're going to be inclined to move in that direction, and that's what we're seeing customers do.

So they probably have a logs provider today. And so there are some share gains as a part of this, but this seems like a more durable trend towards like shifting towards one platform than like 1 year of people shifting away from different vendors.

Yes. And it is because as you move to an integrated platform, it becomes much stickier. So if you're simply replacing like-for-like logs to logs with the sole objective being cost improvement, you're going to potentially run the risk of down the road, another vendor coming in and doing it for yet cheaper. And that then doesn't deliver the stickiness. In our case, it is -- I'm not sure I can even think of a case where we've done exclusively a logs deployment because that isn't why customers come to Dynatrace. So the result of that is you are doing logs, traces, metrics, you are integrating all those elements, you are applying it to applications, infrastructure, real user management, log management, traditional use cases and beyond. You are even applying it to diverse personas. You're applying it to AIOps, but also developers, SRE teams, platform engineering. They're all using the same underlying data stores, the same underlying AI logic. And that is resulting across the board in better outcomes. So you become much more embedded and therefore, much stickier as you look forward. So I think the likelihood that it -- a log workflow comes to Dynatrace and then leaves later on is much lower as a result of that platform-based approach.

Excellent. And maybe due to logs or maybe due to more broad-based strength or something else, but I thought it was interesting that your new logo land size increased 30% in the quarter. I know you guys had a big shift towards upmarket in your go-to-market. But, yes, anything that can call out for why those new deals are landing bigger at this point?

It is -- it follows, I think, Ryan, very logically from all the rest of the discussion that we've had so far, which is that end-to-end observability is becoming more [ proficient ]. And whether it is a large bank, a large insurance company, a large package delivery company, many others I can think of, these are landing bigger. And the reason they are doing so is because they are resulting from a pretty substantial tool consolidation and the tool consolidation is driving the bigger land. So instead of just an application performance monitoring, deployment that is land and expand on a single application, it is, no, I've got all these observability tools. I'm trying to make sense of it. I need to combine them. I need to converge them and then I need to attack it in that way. And that's what we're seeing, and that's driving the larger land size.

Excellent. And you talked about some pipeline strength with larger customers as well, but it's difficult to wrangle the timing on these things. Like -- is that just like enterprise buying, it's tough to pinpoint...

Yes. This provides some of the challenge in our guidance, to be honest, which is we delivered a very strong Q1, very strong Q2, a very strong overall first half. We raised guidance. But some investors have asked us, well, why not more strength in the second half guide? And the answer is well, trust, number one, that our internal plan is higher than the guide, obviously, without a question. But number two, that it is just the variability in these large lands, whether it is a large expansion or a large new customer land, it just adds more uncertainty. Now I think we've done an excellent job of managing that uncertainty heretofore with very good results. And so as I mentioned at the outset, we believe that we have radically derisked the second half as a result of this. And with the strategic pipeline growth at 45%, we believe that we've got the pipeline to cover some of these machinations of these large deals, but it is a reality of our business.

And some conservatism around close rates in the second half part of these deals?

That's exactly it.

That makes sense. Two more AI questions, then we'll wrap things up. Is there anything about APM that makes it a natural starting point for customers who are looking to observe AI use cases that they're about to roll out?

I would say in the AI native, it's a more logical use case to start with metrics and logs and probably infrastructure. So that's probably where they land.

You started first. And then large language model self observability seems like a very early innings for that. But are you starting to get more questions from customers around what Dynatrace can do in that use case?

Definitely. It has not been our starting point as a company. We have sold to the CIO or the CXO or AIOps for enterprise-wide deployments. We have, unlike some others in our market, not typically sold to developers. But the evolution in the platform that we're delivering that we've been working on now for the last year that is slated for release very soon is very much oriented to expansion in the developer space. And the result of that is that really provides us with much more fuel around AI native as a use case and a target customer base for the next evolution of Dynatrace beyond just the CIO, AIOps deployment. So we're very excited about it, and we bring, we believe, a lot of value in this agentic world in a number of ways. Number one, for AI observability use cases, for things like hallucinations and guardrails and all the integrations you need to have into AWS and Azure, GCP, ServiceNow, you name it, all of these vendors to provide a comprehensive deployment. Now having those capabilities really gives us the starting point. We have already deployed capabilities for AI observability into hundreds of customers at this point. So they're already using it today. So we're super excited about that. And we think that, that establishes the foundation for an agentic AI evolution into the future, which is this ecosystem of agents that can really take action where we believe Dynatrace is, in many ways, uniquely situated to take that on because of the precision with which we deliver answers, not just dashboards and not just data, which is really our superpower.

And with developers more focused on those as a customer, your shift to DPS and consumption makes a lot more sense in terms of they can scale more quickly on the Dynatrace at this point.

Exactly.

We're out of time. Thank you so much.

Thanks, Ryan.

Appreciate it. Thank you, guys.

Thank you all.