All right, good morning. We are here to talk Datadog and observability. We have CEO Olivier Pomel and CFO David Obstler. Olivier and David, thank you for coming to the TMT Conference once again. I think you've been here, I think, three or four years in a row now, so we really appreciate it. Let me get through the research disclosures. For important disclosures, please see the Morgan Stanley Research Disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. So, Ollie, we're coming off another successful year for Datadog. We're looking at 27% revenue growth, 23% operating margin, so you're at sort of Rule of 50 financials, serving over 27,000 customers. But we have been going through a sort of tech downturn, I guess, over the last several quarters.
To kick off the conversation, what did you learn about Datadog during this downturn, and anything that you were particularly surprised by?
Sure, thank you. First, thank you for having us. And just for the record, I want to say, the market was already bad before we started speaking, so it's not on us. In terms of what we learned last year, a few things. I would say there were a few confirmations. The first confirmation is that what we do in general, so in Infrastructure Monitoring and observability, and how we do it in particular, is extremely sticky. Even though it was a very, very difficult year in the market, a difficult year for our customers, a year where many folks were going out of business or really rethinking their spending, we maintain extremely high gross retention for our customers. So we disclosed some of that.
We have a graph that we share that our Investor Day a couple of weeks ago, but our retention is in the mid- to high 90s%. For enterprise in particular, it's around 98%-99%. So we feel very good about that. That was more of a confirmation. The second confirmation we got was that cloud migration is still here, is still happening. There was less volume last year as customers were also trying to optimize their spend. But in terms of the number of customers, especially large enterprises and traditional enterprises that were starting data migrations, that were going into the cloud, that didn't actually slow down. We kept seeing them come through and basically get started and adopt some of our products.
In terms of what was more, I would say newer and more surprising, we saw more consolidation than we had in the past. So, you know, we always built towards that. Like, we built a unified platform so we can serve a broad range of use cases and consolidate the whole set of needs from our customers onto what we do. But this became more important faster than we thought over the past year.
That's a great summary. And then I guess the other part of it is we looked at the last sort of investment cycle where, you know, the 2020, 2021. What parts of that growth playbook still remain valid going, hopefully, into the next investment cycle?
I mean, look, the playbook is simple. We're heavily in an expand. We acquire as many logos as possible as companies get into the cloud and get serious about moving into cloud environments. We land as fast as possible with few products to start, one or two products, usually in Infrastructure Monitoring and another product. And then we expand by having more use cases, and we grow with our customers on the cloud. So all of that remains valid. I think the parameters were a little bit different last year, but the model's the same.
I would say the one thing that we keep in mind, especially as we set guidance for the year and the future, is that we do not fully control the base rate of migration into the cloud in terms of raw volume of compute units that are being deployed or gigabytes that are being sent to the cloud. That, that's there. That's happening. It's going to happen for the foreseeable future. But we don't control whether the base rate of growth is, you know, 20%, 30%, 40%. We don't control that. So that's one thing we keep in mind.
Awesome. You mentioned in your previous answer some of the profiles of customers that were sort of in optimization mode. I remember last year you guys called out some digital natives, some consumer discretionary. More broadly, kind of the biggest spending, largest spending Datadog customers were the most likely IT to optimize. As we've seen some stabilization in the business for the past couple of quarters and sort of heading into 2024, how are these cohorts performing? And as we sort of get deeper into the recovery, which market segments or which cohorts do you think will lead the charge driving growth for Datadog?
David, do you wanna take this one?
Yeah. I think we said in our last couple earnings calls that that group of highly scaled cloud natives, which were the most intensive optimizers, had begun to stabilize. And in fact, in Q4, we said they, net net, grew again even at a higher rate than the overall customer base. So that's a good sign that that group has a better sense of what their business is and where they wanna invest. So they look like they're returning to more normal growth. As far as what might happen next, we think that group is positioned well, having gone through that optimization under the right economic conditions to begin to grow again. And as we've talked about, there's in the enterprise world and a lot of other areas, they are much earlier in their cloud journey.
We see them as leading growth in terms of returning to deployment of digital applications and investing in observability.
Yeah, you know, if I, if I were to single out, like, two—currently the two main drivers of outperformance in the customer base—I would say on the sheer volume of growth, there would be the AI-native companies that are going very fast. And then also on the, in terms of consolidation and adopting more of our products and spreading across the organization, there would be the traditional enterprises that David mentioned. So that these are the two parts of the customer base today that are outperforming.
Yeah, it makes, makes complete sense. Wrapping up the call like sort of the near-term conversation, you know, you came off a solid Q4, grew 26%. Guidance for both Q1 and for the fiscal year seems growth decelerating to the low 20s. What would need to happen for you to deliver more sustained growth versus decelerating growth, for the balance of the year?
Yeah. And to remind everybody, our guidance methodology has been taking the recent history and then discounting it. So, even though we see those signs of stabilization and return to more growth, we discount it. And, as many of you know, the main drivers of our business are usage of our existing products and new products. And so, if that accelerates, you'll see the net retention accelerate and more acceleration of growth. That gets all back to where the clients are in their optimization and then their deployment of new. And of course, we said all along that we've been pretty solid and pretty good in terms of new logos across the board, SMB to enterprise, throughout.
And so, if that ticks up, also that would, that would as Ollie mentioned. Other things that are out there include more adoption of our platform, consolidation, which has been an increasing driver of the platform, and then AI.
Yeah, makes total sense. A couple weeks ago, you guys had, what I thought was a really insightful Investor Day , and you kinda took us behind the curtains on how Datadog thinks about building product, how you think about entering markets. But I think the point that, resonated with me, the most was just core observability being a potential 5x-10x from here. Ollie, can you give us a sense of why, you have confidence that we're still in the early stages of observability, like, whether it's, you know, the size of the market or, you know, where your customers are in their cloud journey? What gives you that confidence?
I mean, look, if you just look at the numbers, so I think we saw on the recent Gartner report that we are the top company in market share in IT operations monitoring, I think, as the name of the category.
Yeah.
But we're still less than 10% of the category there. If you combine that with the fact that this category is only one of the several categories that are covered by observability, and then if you add to that the fact that all that's powered by the move into the cloud and that according to Gartner again, I think they owe me for quoting them so many times, the cloud market is going to grow, I think, an average of 20% in for the foreseeable future, 20% every year. I think when you combine all that, like, it doesn't take a lot of imagination to see a 5x-10x increase in the observability side alone without any of the other categories we've entered.
In addition to that, I would note that there are less and less large, independent vendors going after observability and innovating in the space. So I do expect that there will be some of the market share there will be up for grabs in the next few years.
Another interesting data point that came out of that Investor Day and sticking on, on the core observability opportunity was that, you know, roughly 40% of the base is on three pillars, which would be infrastructure monitoring, application performance monitoring, and log management. Given that customers who do adopt all three spend 10x than those who don't, what can the company do to get that other 60% to, to get them on all, on all three pillars? Is there a particular pillar that stands out as, as, as, as something that customers are slower to adopt out of the three?
Yeah. I mean, we're working on it. Don't, don't forget that we were 10 years ago, we were the infrastructure company. So we didn't have Logs and APM . And we announced a couple quarters ago that we had infrastructure over $1 billion but Logs and APM over $500 million. So that means this has been going on for a while. And we've been, either through greenfield opportunities or displacements consolidations. And, and we're working on it. So essentially, what we find is now most customers land with two of the pillars. It you know, it's hard to say. It depends where they are in their, length, and, and contract of either APM or logs. Maybe a little more logs, the barriers to entry and sort of barriers to adoption of logs. But then, so but it's very close.
And that's evident in the fact that both of them are over $500 million. And then what we're seeing is the motion of, once you land with infrastructure and adopt either logs or APM, there is the next motion is to adopt the other product sets. And as you said, then you have a much higher RPU for that customer and much more greater opportunity for growth.
Yeah. And you know, we stick to what works, you know. So we usually land with Infrastructure Monitoring first and maybe something else. That's because it's an immediate need for anyone who moves into the cloud that they need Infrastructure Monitoring. They need it wall to wall. It's low friction. It's very easy for us to train our sales force to always do that first. So it's always a great first product to have in with our customers. When you look at our larger customers, especially the more traditional enterprises, they typically are going to have one of the existing players on-prem for APM and all logs. So after that, the consolidation onto us is going to happen as a function of having the opportunity to do so, you know, when those contracts expire or they offer renewal.
Also, as a function of these customers having more and more of a footprint in the cloud, and the usage of our product reaching critical mass, which makes it very, very, like, basically a no-brainer to consolidate on us.
Yeah. Makes sense. We talked about the consolidation opportunity, and it's a theme that's happening across the observability market. Everyone's trying to be the consolidator versus the consolidated. What is the profile of the type of customer that chooses to consolidate with Datadog versus some of the alternatives of the market? And what are some of the benefits that these customers are realizing when they do so?
Yeah. Great question because essentially, our customers are not really looking for point products. They really are looking to remediate problems, increase efficiency, do it quickly. And that is a platform-type sale. So, those that, and I think it's broad. It's broad. It's SMB through enterprise. They are looking to have as much data in a platform. They want all the different effects and, and, and analytics to remediate and, and improve. And what we've seen and I think there's some case studies in the investor data referred back to. We've seen the benefits include faster remediation, therefore less revenue loss, greater development and operation efficiency because the platform makes sense of everything and they can remediate more quickly, less alert fatigue, etc.
You see, consolidation as you would in vending in going into one platform versus many point solutions and therefore, reducing cost or getting more functionality for the same cost. And then you also see a movement towards the platform, in some cases from open source where you are able to reduce the amount of developer resources that are required to maintain the platform and the operation.
Yeah. And I think I would really recommend that you check out the investor deck we shared a couple of weeks ago if you haven't. There's a few charts there when we show some of these consolidation case studies, basically. Every quarter we talk about, like, 5, 6, 7 deals that tend to be larger consolidation deals, like these are the headline deals. Every single one of those deals where customers go from 10 products to using us, it also involves an immediate, very visible, cost saving. So they go from, you know, having a cost of, you know, $20 million a year to spending, you know, $12 million a year with us. You know, that's the idea. That's always what happens there. It became extremely important in the past year to demonstrate this immediate cost saving.
In the past, you could get away with customers seeing the advantage of moving faster, maybe creating more opportunities to innovate and create more top line. Over the past year, they've been laser-focused on realizing return on investment very quickly.
You know, when I talk to investors, one of the, I think, simplistic frameworks people think about Datadog and its sort of place in the market is, you know, Datadog kills it in mid-market. They roll up customers in the market. Maybe some of your other competitors are more higher-end in enterprise. In fact, you guys have quite substantial, you know, enterprise business 'cause you're playing in all segments of the market. My question is for, you know, the type of customer, a large enterprise that's relatively early on their cloud journey and maybe they're using an incumbent monitoring provider, how do you get on that, you know, large enterprise's radar as they, you know, progress through their cloud ambitions?
And so as they get more cloud-native over time, what's going to be the forcing function to say, "Hey, we should think about moving to Datadog versus sticking with their incumbent"?
Well, I mean, there's the first of all, the way we go to market in general, which is a combination of bottom-up and top-down. So we are already used by many of the humans that actually make up those teams in those companies. They've used us at previous jobs. They use us on their personal projects using Free Tier. They're aware of us. On the top-down side, we actually have been building very successfully an enterprise sales force that covers the world. And, you know, that's I think now it's we're near, you know, 7 or 8 of, you know, this team having success and scale. And today, I think we're so we mentioned also at our last earnings call we're in more than 4 out of 10 of the Fortune 500. We're also in many of the Fortune 10 and not only the tech-forward ones.
So I think we have good success in getting in front of those companies. Still plenty more of opportunity, but we have good success there.
Let's talk a little bit about, you know, sort of mark-to-market where we are in sort of, sort of AI. You've got in the earnings call, you've mentioned, you know, you know, percentage of ARR coming from, from AI-native customers. In terms of, like, what type of workloads are these AI-native customers monitoring on Datadog? Is it more training or inference or any sort of profile of what these fast-growing customers are doing with Datadog today?
It's largely trained, oh sorry. It's largely inference today. And it's still, I would say, in terms of the raw usage, still fairly concentrated on a relatively small number of AI-native companies that really provide the tooling or the models or some of the platform-building blocks to the others. The good news is that that usage is growing quickly. And since this is inference, it's proof of traction of these companies. Like, basically, this is not just them developing models and hoping for the best. This is them serving models to customers of theirs that actually are using them for something else. We see that as evidence of future spread of AI. You know, we think right now, this is concentrated among a few of the model providers.
I think if you play it out a few years forward, you'll see a lot more usage across a much broader set of potential customers.
One of the points that you made, Ollie, at the Investor Day was, and the team sort of made, was AI bringing more complexity. And when there's complexity, there's a role for Datadog. And so when you think about AI being increasingly infused into customers' application environments, when do you think that starts to materialize as, you know, a boost to the monitoring opportunity for the company?
I mean, we see it today already, right? So we mentioned on the last earnings call, AI natives represent 3% of our AR already. And if you look at some of the clouds, so for example, in Azure, AI natives represent a significant part of our growth. So that opportunity is already manifesting itself here. I'd say in terms of broader usage of all of the various applications that are being developed right now among the much broader customer base that use AI, we don't know when those are going to be in production yet. And those customers don't know yet either. It's possible it's a few quarters out. It's possible it takes more than a year. Who knows? So that I think that's the biggest question now in terms of how that usage spreads, basically.
Outside of timing, some of the, you know, products that, you've introduced over the last year, you have Bits AI with its, you know, a sort of a Datadog Copilot.
Yep.
Correct me if that's the wrong term. You have LLM observability and, you know, capabilities to monitor the GenAI stack. How do you think if you take this set of capabilities as a whole and sort of rank them versus all the other opportunities which you're going for, which is vast, how much should we think about this part of the portfolio being contributes to growth, covering over the mid-term to long, longer term?
So the AI stack is a big contributor. Everything we develop in terms of LLM observability or integrating with all the various model providers and, you know, vector databases and all of the new components that make up the AI stack or the GPUs, this is to support our customers' adoption of the technology. And by the way, when I spoke about the opportunity of growing with AI, I think only a small part of it is going to be hard AI, as in, you know, a model running on the GPU.
A lot of it is going to be all of the other systems that are needed in order to leverage AI. To leverage AI, you'll need to have your data in. You need to be in the cloud. You need to be digital. And so it's going to feed a lot more of the, the usual, classic, I would say, digital transformation and cloud migration on the back just to feed this tip of the spear, which is going to be the models and the little bits of AI that run there. Now, if I look back at our products that are more, specialized around, brand new ways to, to, to build applications using, using models, I would say the market there is still, extremely early, and it's still moving very, very fast.
So when it comes to LLM operations in particular, what we see today might not be what ends up being needed in the products in the one year, two years from now when the, a number of those applications I mentioned earlier have landed in production. But there is, I would say, today a lot of usage and a lot of interest from customers, and everybody's learning at the same time, and is going on this curve. Bits AI is a little bit different. Bits AI is not about us supporting our customers' use of AI. It's about us injecting AI into our own product, so we can deliver better outcomes for our customers.
We've also spoken to that a little bit, you know, Investor Day in terms of going beyond observability and really becoming more of a system of action for our customers and what we call, on that day, closing the loop.
Right. Yeah. Which we'll talk about, as we talk about the Cloud Service Management opportunity. Let's talk about security first. So I think you guys entered the security space in 2020. So we're about three or four years in. Recently, you've introduced two new bundles, Infrastructure DevSecOps and APM DevSecOps. What do you expect customers will find attractive about these SKUs? And can these packages be the catalyst to unlock better monetization of the company's security offerings?
I mean, it simplifies things. And I think, again, in our minds, in a few years, it should be unbelievable that you'd buy your security separately from your observability. It's a little bit like, you know, trying to buy your, you know, the airbags for your car from a different person than the person who sells you the car. Like, that doesn't make any sense.
Mm-hmm.
I think those SKUs are a step towards that. They make it easier for customers to consider the security products and consider adding them on to things they already understand they need, which is the Infrastructure Monitoring or the Application Performance Monitoring. And on the flip side, it also makes it easier for our own go-to-market teams or sales teams to bring that conversation to the customer, present things. And it makes it easier for us to train them to do that. So the goal is to really fluidify the process there. It's still early, but with what we've seen in just a few months of having those new SKUs out is that they seem to actually resonate quite a bit. So we'll see. Maybe we'll talk some more about them in the future.
Awesome. Looking forward to it. When people look at the Datadog security ambitions, I think it makes a lot of sense, you know, from an architecture perspective, from getting everyone to work off the same data. I think the question a lot of people have is, like, who owns who makes the purchasing decision? Is it the CISO, the security team, or is it more on the app dev, DevOps side of the house? And to the extent that it's the security side making the call and allocating the budget, what are the initiatives that you have in place to get on their radar, in front of these security-focused decision-makers?
Yeah. So the playbook is the same there as with everything else we do, which is we combine bottom-up and top-down. Bottom-up is the same, for security. We rely largely on DevOps. For cloud team, maybe a few more security analysts, but still, the playbook is largely the same. On the top-down side, we do need to speak to the CISOs, which are our new constituent. To do that, we've also worked on being more present on the security market, being more trustworthy as a security company. We've built the security research team that is actually publishing quite a bit, and is very often quoted, you know, in terms of the interesting vulnerabilities and other things that are being found. So, I think we're on the journey there, but so far, we're on the right path, I would say.
Awesome.
There's a little bit of a nuance between infrastructure security, application security, and cloud team, I would say. Infrastructure security and application security, the vast majority of their users are developers and operations folks. Our usual bottom-up works really, really well with the current user base we have.
Mm-hmm.
The cloud team product has more of its users that are security analysts. And so we had to customize a little bit that bottom-up adoption as well.
As of today, there's no real reason for the company to build a specialist overlay security sales force?
No. And, you know, we've actually tested a few of those things over the past couple of years. At least we're not dogmatic about it. I mean, we're very pragmatic. We'll do whatever works at the end of the day, and we'll meet customers where they are. But what we've seen is that we had more success when we tried to broaden the, or open the aperture for our products as opposed to, which means making the products accessible to all customers through the whole sales force as opposed to constraining it by targeting specific customers and limiting the selling to a small subset of the sales force.
And the strategy, as Ali mentioned, is very similar to what's happened in APM and logs, which is to make it frictionless to adopt, prove value, get it integrated in the workflows of DevOps, and then, like, we've had more centralization and scrutiny of purchase in larger customers, particularly in enterprise. But at that point, you have it embedded in the workflows of the DevOps world, and it becomes difficult not to have it. And that's the way, you know, Datadog has grown, bottoms up combined with then from top down.
Yep. And you see that illustrated with the new SKUs we talked about. I mean, this is the example of broadening. These SKUs can be presented to every customer in every situation by every sales rep as opposed to, "Hey, there's this super, super specific security thing, and you need to be, you know, SEAL Team Six member to sell it to the customer." And.
Right.
The customer needs to know to ask for the secret menu to get it. You know, that's not like, we're going in the opposite direction right now. That being said, as I said, we're.
Yep.
We're pragmatic. There might be different modalities for going to market that might be adapted at different levels of product maturity, market maturity, certain types of customers. So if we need to do things differently in the future, we'll do we'll do them differently.
Great. You mentioned the theme of closing the loop, which was one of the themes I was also really excited to hear about at the Investor Day and sort of maybe the next evolution of Datadog. The term AIOps has been out there in the market for a couple of years, and that might mean certain things to certain people. But, Ollie, maybe you can just give us the vision of what moving from a system of alerting to a system of action looks like for Datadog. We're talking about a system that's automatically, you know, making decisions on behalf of operations and of the business. Are we replacing humans? What do you see the interplay between human agents and engineers and an intelligent AI system?
Yeah. I think you can think of it in terms of three different phases or stages. Or, the first stage is you point people in the right direction, and you keep track of them. And I think we do quite a bit of that today. Like, the system helps narrow down what to look at and points folks in the right direction and make sure they understand who else is working on it and who's responsible for what and how to get that done. Stage two is, the system just fixes the issue or is ready to fix the issue, but keeps the humans in the loop and asks for permission, validation, and things like that.
Mm-hmm.
And we do a little bit of that today. So we shipped some of that, at our conference last year. Say, for example, when we detect errors in production, we automatically generate a code fix that you can apply. And so there's a few more things that we're going to be doing like that. And then the stage three is, the machine just fixes it, and it tells you after the fact, "Hey, by the way, I fixed it." We, we don't do that yet, but we think we'll get there. And I think if you if you fast forward a few years, we'll see a mix of all of the above, because we're never going to fix everything all the time. There's always going to be new situations, new architectures, you know, bleeding edge, one-of-a-kind, situations our customers are going to have.
So there's always going to be a need for them to be involved. But we think in the end, hopefully, we can automate a good part of them, then solve with one simple validation the very vast majority of the issues they're having and then just leave a few for them to navigate on their own with some advice and some guidance from us. So that's the end goal there.
Yeah. It's quite a bold vision. You've recently introduced Cloud Service Management, and so it sort of aligns to this closing-the-loop theme. I think for Cloud Service Management, for Datadog, actually refers to a set of capabilities, that includes case management, which we recently launched, incident management, service catalog and resource catalog, as well as workflow automation and application builder. When many in this room hear the term service management, they think of players like, you know, ServiceNow and Freshdesk or Atlassian Jira Service Management. Is this how investors should think about the Cloud Service Management opportunity Datadog and the competitive environment that you're likely to face? Or are you talking about something that's somewhat different?
At a high level, like, if you zoom out and squint, it kind of looks similar. But if you look at the actual details of what we're going to do for whom, these are actually different markets today. What we focus on is production applications in cloud environments. And the participants are largely developers and security engineers and operations folks. Whereas most of these products focus more on corporate IT, on legacy IT environments, and don't really involve developers all that much. Like, these tend to be more of the sysadmin types that work on that. So, and by the way, this market we're talking about on cloud environments, and it's new.
It's emerging. It's forming. We think it's the future. We think that's where the action's going to be five years from now. But it's still new. So different markets, not a head-to-head situation with those companies. And by the way, today, we integrate a lot, you know, with the ServiceNows and the Jiras and all of the other tools our customer might be using because that is where they actually track the work. Like, they track what the teams are working on in those products.
So far in this conversation, we've talked about strong opportunity observability, you know, security, you know, you know, set up well. We're talking about moving from a systems of monitoring to systems of learning. I mean, I mean, there's several things we haven't even talked about yet. So when we think about, like, the timing of when all of this plays out, like, what is, you know, what is in your mind as you think about when, you know, these other pillars of the growth opportunity start to become more relevant for Datadog?
I'll take this one. Yeah. I mean, I think that security and is, is playing out, as we talked about, it's a combination of a build, which we've been at for two or three years. The build has to finish. It's layering in. So it's, and it also is dependent somewhat on the DevSecOps. The AI, I think we talked about, what we can control and not control, right? We're servicing the tool vendors. And what we're dependent upon in terms of realizing more value because we're setting up the platform to support whatever workloads our client has. So that's dependent on the pace of injecting AI into their applications. We see signs of it. We're optimistic. We talked about when technology changes and there's more complexity, we win. That timing, I think Ollie mentioned, we're following. We'll know because we'll see the workloads.
In terms of the, closing the loop, etc., we've been working on that. We have incident management, etc. And I think that'll layer in in terms of our platform. About half of our investment in R&D is in our platform. And in many ways, we've been extending the platform for a number of years. So in terms, we don't know how that's going to be monetized or when, but, you know, we think that's going to be a continuum of what we've been doing already, but we're increasing our level of investment in it to try to monetize and create more value and make our clients more efficient. Anything else you want to add on that?
No. That sounds about right.
Yeah.
Maybe another question for you, David. It's one that I get, you know, from, frankly, a lot of Datadog fans.
Yeah.
You know, people are really excited about the opportunity in front of the company. The question that I get asked the most is, you know, when particularly in this more cost-conscious environment, right?
Yeah.
You have the potential. There's customers who have the potential to spend $5 million, $10 million, $15 million, $30 million.
Yeah.
On Datadog over time. I think that there are worries that maybe in a given year, a high bill or, you know, sticker shock, quote-unquote, you know, gets some of these customers, you know, leaves a sour taste in their mouth. What are you guys doing to essentially play the long game for particularly for this set of customers that has the potential to be a $50 million customer so that pricing or a bill in any given year doesn't, you know, get them to think about diverting away from Datadog but instead to double down and go deeper into the platform?
It's a great question. And that's, we have substantial resources in servicing our clients, whether it's solution engineers, technical account managers, etc. I think Ollie mentioned how important it was to get through this period and, to be able to have a high gross retention because that means we're going to continue to grow with the customers. So we are working all the time with our customers to help them look at their use and get the right value out of that in some and we have, SKUs that are completely related to that, looking at how they use the product and helping them to, optimize or decide which products to use. So, we've been, I think, sympathetic with our customers in this time and helped them through it. We're trying to create, a long-term relationship.
And I think there's a lot of evidence that that was successful, and it's happening in, in our comments around RPO, which means that what clients are doing is they're committing longer to us, and they're committing in a more holistic way. So that's some evidence that this kind of effort of working with our clients to prove value and to grow with them is successful. And I would add on also, all of our product focus and trying to add more and more functionality and make the lives of DevOps and production professionals easier, is also contributing to that, meaning they naturally have more things to buy that make their life more efficient.
Yeah. Really, what we need to do is give our customers more and more levers so they can align, you know, with the pay-worthy value they get, which means we need to build the right feedback loops for them, and we need to give them levers in the product. We do that by unbundling a lot of the functionality so they can choose that. I would say there's also a big advantage of being a platform that covers a large number of use cases for them for that because going into a year with us, they don't need to understand ahead of time, know how much release or monitoring they need versus how many logs versus how much APM. All that's fungible afterwards.
And we can build the right feedback loops and have them adjust that over time, which is value they don't get when they have, like, five different vendors. I would say, though, we're very careful about and we've learned not to artificially try and make the problem go away. Like, there's a fundamental need in observability to have this feedback loop. You know, we have great retention, right? But in the very few cases where relationships went sour with customers, usually these were because we had at some point tried to make a problem go away by saying, "Hey, we're not going to charge for this at all," or, you know, you get two years free of that product. What it created by removing that feedback loop is that we had completely unsustainable growth of usage from those customers.
That always hits a head later on. That's when people get really, really, really upset.
Mm-hmm.
We've learned our lesson. We're not doing that. We're taking our medicine at the right time, which is when something's starting to look wrong, we try and work with customers to make sure they optimize, they do what they need right now, and we have the right feedback loop in the right place with the right people there.
Awesome. That seems like a good place to stop. Thank you so much, Ollie and David, for joining us at the Morgan Stanley TMT Conference. Really appreciate it.
Thanks a lot.
Thank you.
Thank you.