[00:00:00]
Olivier: The biggest misconception about an IPO is that it's an exit. It's actually not. It's a starting line, not a finish line. Welcome to the Logan Bartlett show.
Logan: On this episode, what you're going to hear is a conversation I have with co founder and CEO of Datadog, Olivier Pommel. Datadog is a 40 billion business doing over two and a half billion dollars in revenue, and they got there while only burning 25 million in capital.
The goal with every single product we ship. is in the end, on its own, even separate from everything else we do, it should be best of breed. Get the hell out of anything out there. We also talk about how Olivier learned to acquiesce as a self described control freak and give autonomy to his different leaders, including his belief that you should hire executives that are more peers than they are subordinates.
People that report directly to me, I should be able to think As the people on my team, as peers, they're not people I manage. If I have to manage them or their career, it's probably not the right fit. Really fun conversation that you'll hear with Olivier here now.
Logan: Olivia, thanks for doing this.
Olivier: [00:01:00] Thank you for having me.
Logan: so maybe for people that don't know, can you describe what Datadog does?
Olivier: Oh, yes. So we do, uh, we do observability and security, uh, for companies that are into cloud or in the mix of cloud and on premise. So we sell to engineering teams and we sell to companies that are big and small. So anything from, uh, one person shops or students, uh, all the way up to the largest companies in the world.
And, you know, started a company in 2010 in New York, uh, went public, uh, in 2019. So it's been almost five years now.
Logan: And, uh, you all struggled to raise money in the early days. I'm curious, how did that inform the business that Datadog ultimately became?
Olivier: Yes. So that was, that was kind of horrible in the, in the beginning. So, uh, It was very hard to raise a seed round and it was a seed round in 2010, 2011, which was a very far cry from what you'd call even a pre seed round today. Um, the, uh, so in the end, I think in looking back, uh, anything that is a constraint or a hardship actually [00:02:00] is a, is a good for the future.
If you survive it, you know, you're more likely to make it later. Um, on our end, I think two things that we got from that was one real focus on solving a real problem. Um, so, you know, we never felt like we were winning early on, so we had to get constant reinforcement that we're doing the right thing. So we focused on that until today.
It's a core tenet of the company. Make sure we solve real problems. And the second one is, uh, we're always super scared that we would not actually manage to raise more money. So we wanted to, uh, to build a business from day one that was very efficient, um, and that would be profitable in the end. And again, to this day, like every single step on the way, even when it was a lot easier for us to raise capital, we've been very disciplined, very efficient.
So I think those two things, I really, uh, trace them back to the early days and the hardships we had.
Logan: Do you think that's a, uh, a universal truth that like anything that is a struggle ultimately proves to be a benefit if you threw [00:03:00] it?
Olivier: I, well, I think so, but on the other hand, you know, we're, there's heavy survivorship
Logan: Yeah, sure. Yeah.
Olivier: uh, Uh, but yes, look, if, if, you know, whatever you get asked, uh, if you could look back and, you know, change anything, would you change anything? Probably not. Like all the mistakes we've made, uh, actually they got us there.
Even before starting Datadog. So I, before starting Datadog, I was at an educational software company for eight years. Uh, I think, uh, like four or five years into it, I really wanted to start my own company. I could not at the time because I was waiting for a green card and I had to stay put for a few years.
I found that, you know, upsetting, you know, I really wanted to do other things. At the same time, all that time I spent there getting my team, growing with that company, uh, making mistakes, learning from my mistakes, I actually led me to be a better founder and a better manager. And, uh, in the end, I think we probably were more successful because of it.
So again, everything, uh, that came at a cost actually.
Logan: It's interesting. Yeah. It's hard to know the [00:04:00] counterfactual, but I assume starting in 2007, I don't, I don't know if you actually had the same insight at that point in time, but the market would have been, you probably would have been too early to what ultimately became data or it would have been a different business in some way, shape
Olivier: Well, I mean, and thank goodness we didn't start a business there. I w I was, uh, so I was starting with the same co founder and the idea at the time was, uh, a platform to look for apartments for rent or something like that. So I think, uh, and it, you know, we much better off not having
Logan: And especially if you struggled to raise in 2010, uh, 2007, 2008, uh, was not the best time for, uh, for fundraising or at least. Yeah. Yeah.
Logan: So, so what was that core insight in the early days of data?
Olivier: And the core insight was really, it was a problem we actually had, you know, which was, um, in our previous company. I was running a dev team. I was, uh, Alexei, my co founder at Datadog was running the operations team. And we started as a small startup. So we basically hired everyone on our teams.
We were very good friends to start with. We had worked in three different companies before together. Um, and we also tried hard not to hire any jerks. And yet, you know, we still ended up, [00:05:00] uh, down the road with, uh, operations that hated development, development that hated operations and teams pointing fingers at each other all the time.
The starting point was really, let's bring them, uh, into the same platform. Let's have them see the world the same way, uh, and let's them, let's get them to work together. That was the starting point for that at all.
Logan: and you did something unusual, uh, where you didn't actually write a single line of code for the first six months. Was that purposeful in going about building business?
Olivier: it was very purposeful. It was actually before we even struggled to raise money. Uh, and at the time, our biggest fear was not to, not to manage to build, uh, software. It was to build the wrong thing. Uh, basically solve a problem that didn't exist. And that was informed with some of our past experience, you know, so I had worked, um, uh, we had worked actually together at an email startup, uh, at the peak of the dot com boom.
And then at the, during a good part of the bust too. Um, and we had built that [00:06:00] great product, but we hadn't really talked to any users and, you know, it sort failed to launch in a way because of that, even though we thought the product was pretty amazing, we thought it was really Gmail before its time.
Uh, and after that, you know, building software for teachers really. It really drilled into us the fact that you can't just, you know, make up yourself what you need to build. You need to talk to users, need to understand what you're actually going to do with it. What goes through their mind, what's their sense of the problem, what do they do during their days?
Um, and so we were very scared when I started a company of building the wrong thing and we'll over index heavily on talking to customers, making sure that we. Solve every problem for them. Um, and that we would not just produce waste.
Logan: I heard you say something funny that, uh, when you don't have a product, everyone wants to talk to you and tell you about your problems. But when you do have a product, uh, they, they start shutting down or it becomes a little bit more, uh, you know, uh, you're on the other side of the table. Can you speak to that point?
Olivier: Yes. I mean, so here's the thing we, so we sell to, um, engineers. [00:07:00] Um, and engineers really don't like being sold to, and anytime you come to an interaction and it kind of looks like a sales interaction because you have something to pitch, it kind of, you know, it deepens it, it dirties it, you know, so people are not as open.
Whereas when you're here to really talk about problems completely from an open ended way, completely from the customer's point of view or the prospect's point of view, people are way more open. So when we didn't have anything to bring to the table. Everybody, including, you know, important people at large companies, uh, were very open to spend time with us and share what was working for them, not working for them, uh, what they hope the industry would become later when we had something that they might try, it became more difficult.
Logan: And so in the early days you, you launched with a alpha version of the product and then ultimately moved to a much more open beta. And I heard that that transition was particularly informative on like some selection bias of who was ultimately [00:08:00] getting feedback on the product. Can you speak to that?
Olivier: Yes, so we, we had a fairly protracted, uh, uh, closed beta. So, you know, we had the, uh, wait list that, you know, we'd select who we thought was the best fit to get in and then we'd spend time with them. And traction was sort of middling, uh, when we did that. I think we, we had a lot of people on board on the platform this way, spend a bit of time, forget to come back up to a point where we're actually worried that we were just not going to, it was not going to work.
So we said, you know, Hey, what the hell, let's open the platform. We're super worried that, uh, you know, it was still embarrassing enough as a product that, you know, the whole world would see it and, you know, and they would, they would be horrible. It was and get turned off and it would be the end of it. Of course, that was not the case, right?
Of course, when we open it up wider, uh, we actually let it to our, whoever was coming up on the, the product to, uh, self select and decide whether they were going to spend more time on it or [00:09:00] not. Turns out it worked a lot better than us trying to figure out who's the best fit at the right time with the right immediate problem.
And he's going to remember to come back tomorrow. Um, so that's when the, uh, traction, uh, started picking up. For anyone who's wondering, I mean, in general, and that's actually the advice I get to the founders I advise or, uh, communities I invest in. The chances that your product takes the world by storm and everybody realizes it's horrible are, are non existent.
Uh, what's most likely to happen is that. You'll get a little spike of attention when you open it up and then it will quickly die down, uh, but at least you'll get some users and be able to, um, to, to build upon that.
Logan: So, so, so in the early days, uh, you had this vision of bringing dev and ops together. And, uh, the cloud was, was the cloud just the, uh, derivative consideration of who self selected and opted in to that.
Olivier: Yes. So the, the starting point was dev and ops. Um, and then as an extension, there was this idea that we would build a platform and [00:10:00] really that we would bring different personas, different use cases into that platform. Um, we actually were very careful not to use the word monitoring initially, Uh, to us, it was a work from the past.
It was, you know, what happened in nineties and it was reduced, uh, reductive. Um, and we had broader ambitions than that. The fact that we served the cloud was just, uh, you know, cloud was what was new at the time, what the new companies were adopting, but it was still very much a toy, like even in 2010. Um, companies with any form of maturity or real scale were not building on the cloud.
That happened after that. So you could say that the, uh, the, the success we've had and the success that the cloud adoption has given us. That was largely timing and luck. Um, we didn't understand initially that, or we didn't dream that the cloud would be so broadly adopted. Uh, and we take over pretty much every single segment everywhere.
Um, and also the thing we didn't [00:11:00] fully realize initially. Was how, um, bringing dev and ops together was actually at the heart of that cloud transformation you know, being able to, uh, provision your own infrastructure yourself as a developer, uh, completely redistributed the roles and brought those two different teams.
Way closer than it were to be. So having a platform that. Uh, empowered companies to do that was actually super, super important at a time and it was a big key to our success.
Logan: And so you spoke about this, uh, or touched on this earlier, but you, you wanted to be, uh, you wanted the platform or to build out a platform and be able to be modular over time. What, what decisions did you make in the early days that set that up for success?
Olivier: mean, we, everything was always meant to be a platform. Uh, we avoided hard coded data models for most of it. We had this very flexible way of tagging to bring data together. Uh, we always anticipated having new data sets, new, uh, new [00:12:00] modalities for data to get in and out. And that was almost to a fault when we started.
It was an open platform and you could put all this data in and we started putting some monitoring data into it initially. Um, but we didn't call it the monitoring product. And as a result, it was difficult for our customers to understand what they were supposed to do with it and why they should buy. Um, so we had to go through a phase, you know, about it, I would say a few months after we opened up the product in a, in, in public beta, where we had to actually say, yeah, actually, no, it's, it's a monitoring product.
Uh, it's not a, you know, collaboration platform between dev and ops. Like everybody loves the idea, but you know, nobody is able to explain to their boss where they should pay for that. Whereas it is monitoring. Um, everybody understands, um, and that's straightforward. You make a case for it, you get it purchased and then off we go.
Logan: Did the trade offs in those early days of like being first principled and, uh, going to where the market was, was ultimately, or you thought it was heading over time. Did [00:13:00] that feel, um, maybe overly religious and not pragmatic? Uh, because there's the tension of just like solving the immediate problem versus building to where the, you think the market can go.
And so how did you think
Olivier: Yeah, I mean, there was definitely, and that's actually a, um, a concern that had been related to us on multiple occasions by the investment community, uh, which was, Hey, everybody who's successful in this space is building a, a theme, uh, product, like, uh, they start with a wedge. Um, and they, they build this slice when it's very specific role that does a very specific thing and that's where they get adopted.
And then maybe they can grow from there. Whereas, you know, we started with this very horizontal, you know, like we have this broad platform and it does a lot of things. Uh, but it doesn't just try to replace a thin slice of the ecosystem at once. Um, I think it took a while for this vision to really, uh, uh, prove itself.
So we did some of that. When our cloud monitoring [00:14:00] product became successful on the infrastructure monitoring. But even then, I think it was still seen as one category. Maybe we were redefining that category a little bit. Maybe when you look at our product and when you looked at the monitoring products from the 90s, they didn't look the same.
But still, it was considered one category. In the late 2010s, like before we went public, what really happened is that we were able to layer more products on top of that. We had we added a log management product, an APM product, Uh, a number of other products in different areas, and then different, um, personas with security and with, you know, FinOps and things like that.
And that's when we've actually proven that that, uh, platform vision actually works.
Logan: And you have an interesting view of the modularity of individual products, and there's this tension that exists between those best of breed solutions and the complexity, but full featured set that they have versus the, the simplicity and ease of use that, um, you're trying to offer within a [00:15:00] suite.
Logan: Can you talk about the tensions and sort of how you think about getting to a fully featured product in the Datadog world?
Olivier: Yeah. So, I mean, first of all, the goal with every single product we ship is in the end, on its own, even separate from everything else we do, it should be best of breed. So it should, you know, beat the hell out of anything out there, but it's going to get time, to take time to get there. Yeah. Uh, most of the categories we enter, uh, have a lot of existing players and they've been building features for these categories for, you know, 10 years.
Uh, and there's a lot, uh, and title stakes are typically pretty high to start playing there. And then there's a lot of stuff you need to build to be the best product there. But what we see and what we win is that when you have a broad platform approach and when the thin category you replace is completely integrated with everything else, all of the other categories, and they share the same user base as opposed to having, you know, Uh, you have five engineers using one thing and then five others using another thing.
And then, you know, they actually have [00:16:00] no idea what goes across these boundaries. When you have all of that, like it becomes irrational, uh, for customers, you know, to just not buy everything as part of a platform. So that's been the, uh, the, the, the path we've been on.
Logan: And the first initial, uh, couple adjacencies that you went into or additional products, you were solving for that same, uh, buyer set. And so you built out more modules within the same buyer set. And then ultimately you moved into different buyer sets. How did you think about that?
Olivier: Yeah.
So, I mean, we initially it was the same buyer. So the, uh, operations folks and the, uh, the CIO, CTO, depending on the company. Yeah. You're buying what we're doing today. We might have some different buyers, you know, the CISO, for example, but the motion remains the same. Like we still get adopted, largely bottom up on top of our platform.
When we sell, um, to security customers today, we, we typically land with observability and then expanding to security from there, I would say the same fundamentals remain.
Logan: if someone, if a founder is listening to [00:17:00] this or a team, uh, would you recommend focusing on, I mean, obviously, uh, some, some selection bias here, but focusing on that persona and building the functional areas within the persona versus taking the product, I guess the opposite would kind of be the service now model maybe where they they've personas over time with the same product.
Olivier: I would say at the end, you probably need to do both. Uh, but there's different levels of, of risks and different levels of reward, you know, with those. When you expand within the same persona. Uh, the risk is that, you know, you run out of things you can do for them, or, you know, you run out of incremental, you know, wallet share you can get, you know, and that can be problematic in the long run when you expand to different personas, like typically you see these big open spaces, like it has all these problems, all these reasons why we should be successful at solving them.
Turns out it's always harder than it looks. Uh, you have to discover the users, you have to build credibility with new parts of the organization. Um, so it, it takes more time like anything else. And when you, when you invest. [00:18:00] Uh, you have different amounts of risks and rewards for everything you can do. And it can take a lot longer for certain of those, um, endeavors to, um, to prove themselves.
And we've, we've seen that with our product set.
Logan: In structuring those new product teams, how do you, how do you go about, uh, thinking how to empower these individual teams on their own and give autonomy to make decisions while maintaining the same shared service or infrastructure or design or feel of what the product is that is Datadog?
Olivier: Yeah. So there's always a lot of push and pull when you have, like, like is our case, but half of the engineering team or product team works on the platform and the other half works on the specific products. But there's a lot, a lot of push and pull between those two sides, you know, so you need to be able to create new things.
Yeah. Uh, from scratch, uh, without having them be platform wide initiatives. And at the same time, you need to keep bringing things in, into the platform. Um, so that we don't have to have different ways of, of, [00:19:00] you know, doing the same thing across different products. So, I mean, the way we do that is by, um, having small teams, um, building products, focusing what matters the most for those products.
If what matters the most ends up being very platformy in the end and we'll, we'll, you know, push it back to the platform. If it is not the case, then, you know, there's, there's no problem. And we're fairly disciplined also in terms of who we evaluate the maturity of the various things we're trying, you know, so anything we do might start as an experiment.
Then we might try to scale it and we might try to standardize on it. And at some point we might try to, uh, uh, we'll stop using it and then retire it in the end altogether. So we have to, you know, be fairly explicit about where everything we do, whether that's a UI pattern or a data store that we've built, where it stands in that life cycle and you know, what we're going to keep doing, but it's okay to experiment.
Logan: Is there a single leader that ultimately gets tasked with kind of being the GM of, uh, those new products as they get going?
Olivier: So we have different product teams are looking after different products and we try, the way [00:20:00] we build product is we, we keep, um, those teams very small and we start building with customers as soon as possible. So we're not, uh, spending two years, you know, in a secret lab, um, with a team of 500 building a product.
And then we have a big reveal and voila, that's it. You know, um, the way it works is we're going to start with a team of 10 or seven or five. Um, and, uh, from the very first day, they're going to involve some customers, uh, to try and figure out. What problem we actually need to solve for them. And then there, as they keep building, uh, they keep adding more of these customers and spending more time with them and as those products.
More, get more, um, um, successful or as we get more clarity on what the roadmap needs to be and what value there is in delivering that roadmap, we can grow those teams and, and, uh, increase the velocity.
Logan: How do you think of the, the product success in the early days or the, what are the different inflection [00:21:00] points of more investment versus maybe pulling back
Olivier: Yeah, first of all, it's really hard when you have a, an organization that has a product that is very successful or several products that are very successful, which was our case, it's very hard to start new products. Um, because it's a completely different mode for people. They need to go from being mostly successful to being mostly unsuccessful.
Um, and, and it took us some time also to realize that, like, actually we need to treat that a little bit differently because you don't want people to feel like they're on the, uh, the end of their career because they went from that thing everybody wanted to, uh, okay, it's not working yet. You know? So is it because it your fault?
Um, so I think there's, there's a different, there are definitely a, uh, a different mode there.
Logan: and so in, in empowering that or, or enabling people to do that, you've also had success in Acquiring businesses that, uh, it seems like there's a profile or type of business that, uh, um, you know, something in a new [00:22:00] area that maybe the founders still want to stay on board and are excited about being part of the data dog platform versus just exiting for financial considerations.
I mean, how did you sort of build this and think about that funnel and when to acquire versus when to organically build out?
Olivier: Yes. So our approach to M& A is, is that, so there's a very broad roadmap we're going after. We're building this broad platform across different categories, across different, across different personas. So there's no shortage of things to build and probably more than, uh, we can build ourselves with a certain amount of time.
There's no specific part of it we think we should acquire. Um, so we're fairly open in terms of what we can acquire or not. Uh, instead we're constrained by the universe of companies we can acquire and that would make sense for us. So we'll build that funnel of, these companies correspond to things with what we're interested in?
Yes. No. Um, do we like what they're building? Yes. No. Uh, do we think they would sell for the right reasons? And the right reasons for us are they want to build, but with more leverage. [00:23:00] So we're never interested in, Hey, they've built something, but they're sick and tired and they want, they want out. The fine reason to sell, but it's not a great reason for us to buy.
So we, uh, we're looking for all that. Then we look for, you know, the cultural fit, all the other things, et cetera. So by the time, I mean, when you start, you have, you know, thousands of companies in your funnel. By the time you get to the bottom of it, like there's a handful. Um, and you know, maybe one every now and then works and that's what we've been doing so far. doesn't constrain us on the size of companies we're buying. So, you know, we, we look at companies that are tiny. We look at much larger ones. Um, but naturally there are many less larger ones. So the chance that you'll, all the planets will align, you know, with a much larger company are, you know, you're in far between.
Logan: And in post acquisition, you, you have a, uh, I mean, or at least I had heard you, you rewrite the product onto the Datadog platform and then try to ship something quickly just to prove that you, you can, I think, uh, how do you go about [00:24:00] doing that?
Olivier: Yeah. So, I mean, first we re platform everything we need to re platform. Uh, depending on how these products are built, maybe some components live on the customer Uh, environments that we don't need to rewrite all that, for example, but all of the back ends, the UIs, everything else needs to be tightly integrated with what we do.
So we replatform everything. Um, there's a cost that comes to that with that. Uh, but when you think of the value we deliver in the end, what we sell, that that's actually core core to what we are. And that's what we, uh, we make those acquisitions very successful in the long run. Now, when we buy a company, the only thing that matters.
Is what happens after we sign the deal itself is easy. Um, the all of that part fine, everybody will do it. Um, what happens in the year that's that that follows is where all of attention is even when we actually work on the deal itself. And for that, we light a very, very short fuse on the acquisition to start showing some [00:25:00] value.
And so, you know, we do it, uh, we do a three, six, nine month plan, you know, uh, the goal is we need to ship something in three months, which is a very short amount of time, you know, after an acquisition. And it could be anything, really. Could be something internal, uh, maybe something we show to some customers.
Um, it doesn't really matter what it is. It matters that it's in three months. Uh, and what it does is that it creates urgency from the moment we close the deal, both, uh, from the company we acquired, but also from the rest of our team to actually get together, integrate, you know, get things working and have something to show.
And it, it creates some confidence on both sides too. Like it shows to the new team that actually, yeah, I mean, we're welcome here and we can get moving and it's fast and it's, and it shows to the, the existing team, Hey, those people we just brought in, uh, to actually know what they're doing and they're helping us move faster because the biggest risk when you do M& A [00:26:00] is not that you waste the money you spend on acquiring a company.
The biggest risk is that you sort of kneecap the rest of your, uh, R& D organization. Because they think, okay, you know, why did we acquire this thing? You know, it's not helpful. These people are not working out or, you know, they're resting investing, you know, what, what, why am I working so hard? Um, so it's very, very important to build that confidence and show that value very early on.
Logan: for a very data driven, uh, company and product, uh, I'm told that you all are maybe a little bit more intuitive on your own insights on the product side and listen to, uh, more qualitative feedback from the customer rather than quantitative data, uh, from the customer, because maybe that's a lagging indicator.
Can you, can you, can you talk a little bit about this?
Olivier: I mean, first, first of all, I mean, we're still knee deep in data, right? Sure.
Logan: I imagine this is a very relative,
Olivier: Yeah, most of what we do, uh, and the products we build are informed by reams of data [00:27:00] and customer activity and what we see them doing and where we see the value. So there's, there's a lot of that. That being said, um, I am personally, um, very careful about not using data where it, you know, it is not meaningful.
Um, so. I'm, uh, you know, I torture people when they present, you know, percentages from a, it was an N of seven and, and two digits to the percentage and things like that. Like it's, uh, it's not because you have three numbers that it makes sense to compute an average basically. To your point also, it's very hard to understand for a lot of what we do, whether a number is a leading indicator or a lagging indicator.
And a lot of the tendencies folks have, you know, when they look at customer activity and things like that is to mistake the two. You know, so the, uh, The analogy there is, you know, it's a doctor that's, uh, you know, just start paying attention when you don't have any pulse anymore. Like I said, it's a little bit late, but most of the customer metrics you can get, uh, you know, kind of looked at that in the end.
So it's a, [00:28:00] one last piece that's interesting to mention is that, uh, data is not going to help you make your product simple. Um, it's a, you, customers are not going to, the moment you start getting into, uh, questions about, you know, is it simple and other things like that, it's Customers don't really respond well to this kind of questions.
There's not a lot of, of, uh, information in the answers they're going to give you because asking the question itself just, uh, creates more work in their mind than they would just looking at a product. And, you know, so it's very hard, um, I think to, uh, to keep a product simple just by looking at, at data. To simplify a product, I think you have to edit.
And I think editing. Uh, is something that is done in a more, you know, um, uh, let's say, uh, uh, opinionated way than it is now coming from the data.
Logan: How do you empower the teams to make those decisions? Because I can imagine you have an artisanal [00:29:00] to that and the team does. But at an individual level, how are they empowered to really have this design feel?
Olivier: Uh, I mean, look, empowering the teams is a, is this the only way to scale, right? Um, so I, I said earlier, we're building new products with small teams, um, and you know, those teams need to be able to move. They need to be able to change things. They need to be able to create new paradigms. Whether in the end we, we carry those paradigms over to the rest of the product, whether we choose to retire, to retire them after testing them.
Uh, whether we think something is just, just, you know, too confusing and, you know, we need to take another pass at it. I think that that's part of the, you know, product lifecycle, but yes, I mean, the, the goal as we scale is to keep pushing the decisions down as much as possible, uh, and empower people so they can keep running on their own, you know,
Logan: Early on, you, uh, you built a, uh, foundation for machine learning, I guess what we would call artificial intelligence, uh, today. What did you learn about scaling up a, um, AI infrastructure and [00:30:00] building in the early days or early ish days of what now would be considered that category.
Olivier: Yes. Uh, it's interesting, right? I mean, when we started, AI was a dirty word. It was mostly used as a bullshit buzzword. Um,
Logan: was pejorative in some ways.
Olivier: exactly. Yes. Um, and now, now it's fashionable again, but give it a few years. I think maybe, uh, uh, we'll, we'll, there'll be some back and forth on what's, what's fashionable and what's not.
Um, I would say the, um, the, the biggest learning from applying machine learning on the reams of data we have and with our specific type of user. Is that, uh, precision matters a lot more than recall for everything we do. The lie, so to speak, um, customers will tell you about AI in our field. What we tell us about AI, uh, is that if I have, if I'm given the choice between the false positive and the false negative, uh, I prefer the false positive because [00:31:00] you send it to me, send me the, uh, the ID, the site, you know, whatever the machine found out for me.
And I'll decide whether it's, uh, it's appropriate or not. The reality is, you send two false positives in a row and people just turn you off. And they never pay attention ever
Logan: They started ignoring.
Olivier: And that's been the, the, the challenge, the biggest challenge, actually, with incorporating AI in the, in the workflow, is the interface with the humans.
Make sure that you build the right trust relationship and that, you know, you, you actually, you know, help people get to the next step, thanks to that. And I would say today, that's still pretty much the frontier we need to cross, even with the, I mean, there's so many new advances now, uh, I mean, uh. I don't need to, uh, um, say a lot more about that, but, um, there's so many new doors that are open, new things we can do, um, higher expectations also from the, uh, the user base and the customer base.
I think the, the, the biggest issue is still, how [00:32:00] do we build that trust relationship with the user? Uh, and how do we have them actually lean hard on the, uh, on the automation and the AI?
Logan: Yeah, it's interesting because you see a lot of the creative that have like infinite iterations and are a little bit more abstract and it's, it's awe inspiring to see that, but you don't in the enterprise, you need accuracy, uh, and you can't have hallucinations. Hallucination in a, in an ideogram or mid journey, uh, picture is that's, that's art.
Hallucination in your system is an error that is going to lose a lot of
Olivier: Yeah, exactly. I mean, also, look, uh, if the machine generates a. A dog with three legs, uh, like you spot that immediately and you, and you reroll it and that's fine, you know, um, whereas if the machine gives you a complex interpretation of what's going on between your, your applications and, you know, and what's at fault, what's not, you're not going to be able to see the issue right away.
And so if it [00:33:00] turns out there's actually an issue, you're going to lose a lot of trust if the customer dials down, goes down the wrong path and spends, you know, half an hour doing that. It's perfectly fine for people to, uh, go down the wrong path based on what humans tell them. But they are very unwilling to do that based on what machines tell them.
So I think the bar is really high in terms of delivering value there.
Logan: I heard you talk about, uh, how a lot of companies end up being sales driven in their culture or engineering driven, and you all wanted to be customer centric or customer driven in that. Can you speak to, um, the trade offs of, of orienting and one direction of sales versus engineering and how you guys prioritize being customer driven?
Olivier: Yeah, I think it's, uh, it requires constant discipline and constant attention, because it's very easy for a company to veer one side or the other. Uh, you know, To caricature, uh, if you go sales driven, uh, you'll do whatever you need to close the next deal. Um, that's what's going to drive the roadmap, [00:34:00] um, that's what's going to drive the economics and everything.
Uh, whereas if you're engineering driven, um, you know, you start with a solution, you build this beautiful tower. Um, and then, you know, maybe you let the customers in, but only if it doesn't detract from the, uh, The beautiful, uh, journey we're on and the destination we set already. And I think for, um, what we do in particular, we can't have either of those.
Uh, we need to start with a customer. We need to start with, instead of starting with a solution, instead of starting with the money, we need to start with a problem. Like what's their problem? What, what do they have in their environment, in their life that is horrible? And how can we make it better? How do we show value there?
That's the, uh, uh, the core of what we need to focus everything on. It's, uh, it goes beyond what can we do for them today also, which is, well, what can we do that's actually going to be great two or three or five years from now? Which means. So for that to be great five years from now, we, we need to be a great company five years from now, [00:35:00] which means we need to have great economics.
We need to reinvest. We need to build like, you know, so it really, everything has to be seen through the prism of how do we, like, if we project ourselves, uh, five years from now with that customer who is everything amazing for everyone.
Logan: The manifestation of this within the company, I've heard there's some examples like engineering rotating through on support, uh, maybe for a week, a year or something. I'm not sure if that's, if that's still the case, but are there, are there things you've done from a cultural standpoint to, to enable this within the organization?
Olivier: Yeah. So we, we did actually for a, for a long time, we had engineering rotate into support. Uh, we don't do it anymore because it's, uh, uh, it was too many people running for the system. And, um, it was a bit too complex to run a scale. But in the idea behind it was really to close that feedback loop. Like we were big fans of feedback loops, uh, but to close the feedback loop between, Hey, I had a great idea and I implemented it in a product and then seeing what it actually works in [00:36:00] practice.
And it turns out like, like always, like, you know, some genius ideas are actually amazing and some others actually not that great. And you only realize that after the fact. Uh, everything, by the way, everything I keep getting, you know, everyone who listened that, but Everything you find that is horrible in the world around you started as a genius idea. The intent was not to make everybody miserable.
Logan: Your growth market initially was a bottoms up and I don't like the term products led, but, uh, that, that seems to be one of the terms that's in vogue today. How did, how did you make the decision to, to go about selling, uh, that way with the, I guess the alpha, the beta, and then from there
Olivier: Initially, look, we had no sales team, um, and we were selling also cloud companies and cloud companies in, in 2010 were smaller, newer type companies. Uh, so bottom up motion was just completely natural. That was the only thing that would make sense for that. And that's what we started with. It turns out [00:37:00] as we scaled and as the other parts of the market came to the cloud, like, you know, later on enterprises, big enterprises started moving to the cloud.
Things like that. When that happened, bottom ups was still mostly the way we went to market. And the way, and we've tried, you know, various ways of top down also, um, uh, you know, the earliest thing we tried in top down, for example, was when we started raising money, uh, from, you know, big VC investors, um, everybody was eager to make a lot of intros and, you know, get us in front of their best and brightest companies.
And it never really worked Like, and the reason for that, and the reason we still sell bottom up, I think, even in the enterprise, is that. We're the kind of product that is inherently has to be the, uh, the team's idea, like people need to want it. They need to feel a, uh, uh, specific relationship to the product.
The moment it's something that your boss's boss's boss passes down, like it's, uh, it's suspect. Um, so [00:38:00] that's why, you know, our type of product with our type of users. Yeah, much better bottom up.
Logan: And so then that sales motion, uh, in the early days, it sounds like you all held off on hiring like traditional VP of sales or sales people for a while. Is that fair?
Olivier: Yes. Yes. We, we were, we wanted to, uh, to figure out what the right way was to go to market, like we didn't want to re to rebuild an enterprise sales force from the very beginning in the same exact way every other company had built them. Um, some of that might have been, you know, we were just a bit naive and, you know, clueless, uh, um, but, you know, if you zoom, fast forward to where we are today, um, It actually yielded better economics and much, much, uh, tighter sales process in general.
So I would say, I don't know if I would recommend it to everyone to do exactly the same way. I think now you probably have a pretty good idea of how every single piece of software, whether that's for developers, uh, [00:39:00] cloud consumption, not consumption, everything can be sold. Uh, but at the time, at least it, it made a lot of sense and it paid dividends for us to do so.
Logan: One of the things that in, in going bottom up versus going more enterprise oriented is, uh, or outbound, uh, sales oriented is you, um, you don't have the same number of dials that you can pull necessarily like most of our enterprise businesses. You hire more reps and that's the way you, you grow the business.
What were the dials in the early days that you all thought about to actually generate more leads and make sure that you're at, you're able to scale.
Olivier: Well, so here's the thing. I mean, those dials still apply. Um, So that's why I don't like product led growth because it actually, it takes both product and sales. Um, the, um, the sales team still is going to get you in front of customers and there's still a coverage model and there's still, you know, a math of how many, how many reps do you have?
And you know, what part of the world do they cover and productive out there and all of that stuff. Uh, but the main difference is that you don't just talk to the, uh, the, [00:40:00] the C, the, the CIO, the CTO, the CTO. You actually talk to the people in the, in the trenches who do the work and they're the ones adopting and they're the ones growing the product and all of that stuff earlier on very early.
Um, we were like many super early companies were inbound driven, so we would produce marketing, um, in the form of events or blog posts and things like that. And then we get people signing up on the product and then adopt it and they'd go. I think that model for us and pretty much for everyone at some point plateaus, like you reach a point where your natural reach with the content and the events and everything else is limited.
Uh, and then you need to go further to actually find customers in wherever they are. It's actually a pretty traumatic, uh, transformation to go from inbound to outbound. Uh, most companies struggle with that. Uh, because it's a different culture, different sales culture. It's, uh, um, there's different metrics, different everything [00:41:00] that relate to that.
So, uh, I remember it was a hard time when we did that, but again, to scale past a certain point, I think there's no, no other choice.
Logan: About how big were you when you started layering on the outbound?
Olivier: Uh, I think we, we did that, uh, um, maybe we were a hundred million revenue, something like
Logan: Okay, so pretty late on a relative basis. Um, in generating the content, a lot of technical companies will try to do rotations for engineers, produce one blog post a month or whatever, and inevitably it falls down the stack rank and then they're not happy with the product and, um, that they're actually putting out there.
And so you guys had a, uh, a different, maybe tried that and then. Ultimately hired people specifically dedicated to producing content.
Olivier: we did. And the problem was, um, it's actually hard to operationalize having engineers produce content. And what we found also is that it's not. It can be very, a very, uh, unproductive use of people's time [00:42:00] because people might be excited about writing, uh, but, uh, they might not actually enjoy write all that much or writing all that much.
So you'll see people. Um, you know, spending their day in front of the, uh, in front of the editor typing almost nothing. Um, and just because it's painful to stop writing. So we, we decided not to do that. Uh, we stopped doing that. So today we have a team of writers, uh, that are at the intersection of tech and journalism.
Typically, we look for people with both experiences and they're the ones, uh, they actually can't be marketers. They can't be non technical. Otherwise the content just doesn't come out right. Uh, and they're going to spend time also with the internal teams. They're going to spend time interviewing people internally.
They're going to spend time, uh, validating everything they put in there. They're going to spend time researching with the data. Uh, and they're going to produce content there. We still have engineers on a occasional basis, um, publish, uh, but that's more, it's more of a nice to have, if [00:43:00] people are motivated, if they have something they want to say, if they're, uh, if it's not too painful to them.
Uh, but we don't rely on the team to do that too much otherwise.
Logan: In building a go to market motion as a technical founder, was there something that was particularly counterintuitive to you? Uh, that that you learned along the way?
Olivier: I don't think there is anything that was so counterintuitive. I think, look, the way I see it, these are all systems, um, and your systems are feedback loops and they're instrumented in some ways. I think the thing that's the hardest to, um, um, come to grips with as you build the sales systems. is that while they are extremely well instrumented, like everything that happens, uh, is captured in a CRM, all the interactions, you know, you have increased numbers and just about anything, uh, the numbers are pretty lumpy pretty hard to interpret over short periods of times.
So even though you can measure to, uh, with, you know, with [00:44:00] many decimals, actually the productivity of your team and the conversion rates and all of that, that you actually need to let things play out for quite a while before you get an idea of what's working and what's not working.
Logan: Yeah, the architecture point, uh, that, that your sales organization has an architecture similar to your engineering, uh, team or is building an architecture is architected in some way. It's a, there's a lot of parallels between the two and making sure the systems don't have break points or single points of failure or anything.
Olivier: Yes. And, and look, it's very important to build the feedback loops everywhere. Uh, so how do we know that what we're doing is working? How do we know that we're spending our time in the right place? Um, if we did something horrible, how will we know about it? Like it's, uh, it's hard. And when? You know, When the one thing, um, we've done for example, very early on is that we had these, uh, very short term contracts with customers.
And we still do that today. You know, we can buy our product month to month. Even if you buy on our, if you're in a multi year contract with Datadog, uh, you can use any of our product for, you know, a [00:45:00] month if you want, and then churn out of it, uh, and use another one instead. Um, I think what it gives us is this very, very, very quick, uh, feedback on whether the product is good enough and valuable enough.
And that helps us. Improve very quickly, it helps us get bad news, uh, in an, in a way that's, uh, impossible to ignore because, you know, as, as you grow as a company, uh, you train yourself to ignore bad news. Um, and, and you have this opportunity right now to fix it. The flip side is, or the, um, uh, the alternative is you, um, you sign everybody on a three year contract.
Um, and if they stop using the product in month two, you don't have to face the hard truth, you know, for another three years. Um, and by then it's too late to actually fix your product. You haven't, you know, put that into the, uh, into the feedback loop.
Logan: heard you describe yourself as a little bit of a control freak in the early days. How did you learn to [00:46:00] empower teams as you scaled?
Logan: We touched on it a little bit earlier, but it maybe didn't come naturally.
Olivier: Yeah. Well, I mean, look, I think many founders can relate. Like typically when you start a company, you might be a bit of a control freak. Happens. I've seen it a number of times. Um, I think for me, it was a very new, um, and, and good experience to learn to, uh, give space people around me. So I, I knew how to do that already with my co founder.
We had, we knew each other, we had worked together for a very long time. That was not an issue, but adding more people to the mix, I think was more difficult. Um, very early on, like we had hired the, uh, um, the head of product. Uh, and I was butting heads with him all the time because, you know, he was a control freak, I was a control freak, we were going on each other all the time, and it, I think it almost blew up in the first three months, at which point we decided to hit reset.
Um, and that's actually how I [00:47:00] learned to, uh, to give room. And that's something I've kept doing after that, as I've added, I've added more, um, uh, people to my senior management team. So people that report directly to me, I find that I should be able to think as the people on my team, as peers, they're not people I manage.
If I have to manage them or their career, it's probably not the right fit. Uh, I should let them do their thing and we should talk about, okay, what's best for the business. What do we do next? Et cetera. Maybe we can disagree on some things, but you know, they're doing their job. Um, their peers, I trust them and have the space.
I shouldn't be making decisions for them. Um, I encourage everyone who's building a management team to try and think that way. I think it's liberating. And I think it's, uh, also helps you set the bar, the right level when you hire him.
Logan: Is that early product leader still with the company today?
Olivier: Yes. So he's retiring at the end of the year.
Logan: This isn't Amit,
Olivier: This is Amit.
Logan: Oh, wow. So, so, so, I mean, for people that don't know, Amit was a quasi co founder in the [00:48:00] early days. I've heard, uh, and he, he, a lot of stuff rolls through him. It's not just chief product officer.
Olivier: he did, he did a lot of the customer stuff for a lot of time too, and which for us is very related. Uh, again, the product is all about what the customer thinks of it and the pro and the problem we solve. So we thought actually the two were very, very related and, uh, yes. And so that, that was a very productive relationship for the company.
And, uh, and that almost blew up, completely blew up, uh, in a, in a firework, uh, in about two months in.
Logan: Wow. That's, that's amazing.
Logan: Um, You still do things though, to stay close to the customers despite, uh, or close to the business, despite the empowerment. I've heard there's some tactical things you'll do going through, maybe still support tickets or reading employee reviews. Uh, can you speak to the ways at which you stay close while still empowering
Olivier: Yes, I think it's a, so there's two things. Uh, one is. Um, data should flow up to management from everywhere. Like there shouldn't be any [00:49:00] filtering. Uh, of course you'll, you'll have these filter views anyway, like people will bring, uh, will build retrospectives and reports and things like that. You should be able to get data from every single layer of the organization.
I see it as important to sample data from just every single part of the org. Uh, because that's, at least to me, that's my contact with the, with reality. Like, what does the fabric really look like when you read customer support issues, when you read transcripts of sales interactions, when you read, you know, what's, uh, what engineers are complaining about, complaining about on the engineering team, like, what does it look like?
What do they talk about? Does that actually agree with the numbers and the reports and the other things I see, like, are these, is it part of the same reality? I see that also as part of training the model. Like, you know, you. You, your machine, like you keep, you keep seeing those things, what you see actually completely informs your perception of the world.
So you can't just look at a nicely polished, simplified, a little bit, you know, uh, [00:50:00] rose tinted, uh, versions of the world that you might otherwise see as a, as a, as an executive. Um, so that's that, so there are all the way up. Uh, on the way down though, it's very important to go through the management team, so you can't just go and, uh, go find somebody on the sales team and tell them, no, no, you do this, you do that, that you actually have to go through the management for that.
Otherwise, when you don't do that, like you completely. And you, you notify, I mean, you basically, you disempower your management team. People know that they shouldn't listen to their managers. Instead, they should listen. They should wait for, you know, whoever for the top to come and tell them something. But I think it's very important to, uh, to have this asymmetry.
You can get data from anywhere, but you can't act, uh, on anybody at any time. The other thing I found is that when you, um, sample a lot of the data this way, uh, everybody in the company starts caring about the details. Um, so instead of looking up and trying to think, okay, so what do I need to show my boss?
They start looking down and think, okay, so [00:51:00] what's actually happening here? I better understand that, you know, the first time I, uh, people see an email from, you know, um, me or C level or VP level about a very, very minute issue with a customer and asking a question about it. The first question people ask themselves is why, why is he or she looking at that?
Uh, and then the second thing that comes after that is, okay, maybe I should probably know more about it than, uh, all these people that are up the chain. So let me look into that. And create this culture where people care a lot about what's actually happening more so than what they're trying to pass up to the hierarchy.
Logan: Is it right that you all only ever burned 25 million roughly,
Olivier: Yeah.
Logan: and now you're approaching 3 billion run rate or something,
Olivier: Anyway, still, uh, two and a half billion, but
Logan: two and a half billion. So approaching, I think that's, I can say that maybe you can't, um, outside of, uh, simply strong product market fit in the early days, uh, what. What enabled that culture [00:52:00] of, um, cash, cash conservation and, and, uh, efficiency in a way that maybe isn't obvious.
Olivier: Yeah. I mean, look, it's a, it's a culture of, of discipline in general. Um, you, so first whenever, like the easiest way to spend a lot of money, um, as a company is to have too many things in your sales process, uh, because it's tempting to, you know, you spend more on marketing, you spend more on sales. You can, you can, there's always more things you can do, more people you can involve in your sales process.
And so our approach there has been to be, to always understand before we add something to the sales process, um, what's actually load bearing and what's not in the sales process. So maybe we're going to try one thing. We'll try it small in a small fashion first, or we'll try it and then we'll remove it and then we'll see, you know, does it actually make a difference?
Um, the problem if you don't do that is you keep adding things, you know, typically in zero interest, uh, worlds. The pressure is just [00:53:00] to grow faster. So of course you're going to spend more and you're going to, you know, weigh all the different things. Um, problem you get at the end of that is that you have no idea what part of it you can remove on up and you will scare because the pressure is still to grow after that.
So you have to save money and you have to grow, but you don't push you can remove. So it's a bad situation. So on our end, we're always very disciplined about, uh, deciding what to add, what not to add, and keeping those economics fairly, uh, um, fairly straightforward. We've also been very disciplined about not doing bad deals, you know, so we, when we look at.
Um, uh, community situations or customer situations, uh, just because somebody else wants to do something that, you know, we think stupid doesn't mean we should be doing it. Um, and again, if you look back at the, what I said earlier about, uh, solving problems, uh, for customers in the, in not just for today, but for the long run, like this is the right thing.
Like if we want to be, [00:54:00] um, the best company, Uh, and the best solution for our customers and make a big difference for them five years from now. Um, we need to keep being investing in our product for those five years. There's a cold hard math to that and I can come back to that. Uh, and that means we shouldn't actually be doing bad deals.
That's part of it.
Logan: so that.
phrase, uh, load bearing to make sure I understand the point. Uh, so you would keep the sales organizations more nimble and figure out what were the mission critical functionalities within that and not layering with, um, sales engineers or solution architects or whatever it is around the process, because then it muddles the, what is actually mission critical.
Olivier: there's sales engineers. Of course, you always, I mean, you're going to have sales engineer, you're going to have all of those things, right? But then you can, you can say, uh, hey, uh, We need, for these products, we need specialists that are going to sell them because they need to understand the product better.
So then you need to scale another organization on top of it. And you have these more people that show up, you [00:55:00] know, in the customer meetings, maybe you're going to say, Hey, um, we, we do a better job with customers when we can, um, do an assessment of the business value. Which also do by the way, but then the question is, how do you, how much do you scale that?
Do you also end up having one more person that comes to the meetings with the customer every single time because they're here to assess the business value? These are all these different things you can do and they all make sense, you know, individually. Uh, but if you're not disciplined about what, uh, understanding what makes, what actually makes a difference when for which customer in which situation, you end up scaling all of those organizations at the same time.
And you end up in situations such as, you know, before I studied Datadog. In my previous company, uh, we tried to buy a service from, I think it was Sun at the time, uh, it was a server. They, we were just happy to order it online. You had to contact sales. We contacted sales and then 10 people showed up. Uh, I just wanted to [00:56:00] buy and knew exactly what I wanted to buy.
And, you know, 10 people showed up, uh, and you had all these different functions and they produce a memo, uh, you know, where we're going as a company and all those things, none of that was required. All of that was, I'm sure, incredibly expensive for them. And is it just a by product of, you know, uh, an opposition that has grown, you know, without understanding what actually was needed or what made sense from a customer perspective.
So we try to be very, very disciplined, uh, with all that.
Logan: so, uh, pricing in the early days of data dog, you want to make it really easy for, for customers to get going. But there's this, this balance of, of what type of business you want to be the premium product or, uh, maybe the, the cheaper solution out there. How did you think about where to fit on the pricing curve?
Olivier: Yeah. So there's two things. I mean, one is what, what kind of, um, uh, what kind of cycle do you want to have with your, with your customers? Uh, do you want a cycle where you [00:57:00] charge for your product? Uh, every year? Uh, your customers tell you, well, that's a lot of money. Um, and I need to deliver value for that.
So here, these other things I want you to do for me. And by the way, the part, your product is not valuable enough, so you need to fix it. That's a good interaction because it means it pushes you up. It pushes you to do more. It pushes you to solve more problems, pushes you to be better. Um, that's the interaction we chose.
The alternative is to be a price disruptor, which is, um, you just cut, select the customers on that. And then the commercial with the customer is why it's, uh, well, you know, if you remove this and this and this could be even cheaper, um, and that one pushes you down. So that's, I think in a, in a short term might work.
Maybe some business models. Things that are more consumery, I think it works, you know, when you, when you, uh, you can get mass adoption that way, for example, uh, I think for what we do, it doesn't work, you know, in the long term because there's a, there's a, um, [00:58:00] a fairly, uh, cold math in terms of what it takes to build a company, you know, to be successful in the long run, there's a, um, um, uh, like you need to be able to, first of all, you need to be able to sell your product, you know, so 10 times cheaper.
Uh, you still actually pay a sales team, um, to sell it or is it so cheap? Because the cost of sales is fairly fixed. Like, you know, you still talk to people, you spend time, et cetera, et cetera. Uh, you market them, you have to find them. Uh, so how much of your revenue is going to go to sales as a, as a result of that?
And then as a, as a consequence of it, how much do you still have to invest in product? And then where can you be, you know, two, three, five, 10 years down the road, based on that investment you're going to make. So we chose the, um, the cycle of, Hey, we're going to be pressed for value by customers and we're going to be on the hunt for delivering more value and doing more for them over time, as opposed to, you know, hiding at the bottom of a bill somewhere, hoping to be forgotten and [00:59:00] having, uh, and, and shaving from it, you know, perpetually.
Logan: Yeah, competition. I mean, this is an old adage. I don't know if you agree with it, but competition will typically come from some orthogonal area and it's It usually starts as a low cost option or a lower cost option, and then it builds more full featured set over time. How do you, one, do you agree with that?
But then two, um, how do you make sure that you don't get disrupted from a adjacency with some lower price option?
Olivier: I mean, look, so we, so first of all, we're always careful about what customers actually use, right? So the way to, the way to look at competition is to read computer websites. Uh, cause that will drive you crazy. I guarantee you that, especially in a field like ours, you know, where there's a steady stream of new companies.
And when we, when we started that up, there was also so many incumbents already, um, But the way to understand competition is through the eyes of the customers. So what do they say? What do they see? What do they like? What do they not like? What do they think is viable? What do [01:00:00] they think is not viable? So that's completely the prism to use, uh, to look at competition.
Then when you look at a disruption, uh, so you have to look at what's good disruption, what's bad disruption. Uh, so bad disruption, uh, is are, are they disruptive from a, from a price perspective, because. They have a poor business model or, you know, they, you know, they raise a half a billion in zero interest and they're using that to subsidize their customers.
Are they building a business where they can't actually, um, uh, they can't actually invest and they're not going to be around in a few years. That's bad disruption. That typically this one, a little bit painful because, you know, these people out there and, um, offering deals, but, you know, you know, and we've seen that many times over, like, you know, these are not going to be around for the long run and those deals.
Of a short, um, lifespan, the, the good disruption, like new technology, new modalities, uh, new, new topologies for data, things like that, uh, that [01:01:00] we actually try to bring upon ourselves. We, whenever we see something in the market or we can think of something, we try to read ourselves. An example of that is we released a new product at our conference last year.
It's called FlexLogs, um, that, you know, makes, um, log data, basically a lot of managing cheaper, um, And, uh, we aim to do more and more of that over time, you know, for our customer base.
Logan: I've, I've, I've heard, uh, data dogs culture, uh, described as low ego slash drama. Um, how do you, uh, how do you actionable eyes that or, or make sure that, uh, that remains the case?
Olivier: I think it's a, so that's the stance of the company. That's the stance for, uh, the management team, my co founder, myself. Um, I would say the, the biggest thing we look for in people and the biggest threat of the company that leads into that is that, uh, we need to be, you need to have a growth mindset. We need to be humble where that's in [01:02:00] front of the customers or internally.
You know, it means when we go meet with a customer, we, we here to, to learn about their problem. We're not, we're not here to teach them what to do things that we want to be taught. Um, It doesn't come naturally, mostly like when you've been building something for a long time, you have a lot of opinions about what works and what doesn't, that you actually need to be, but when you meet with a customer, you need to be here to listen.
Um, so that's humility, the same thing works internally when, uh, say, you know, you're a star engineer, um, you've been hired because you're super smart and you're super good at your job, that's a given, uh, you build something, uh, but somebody else on your team, Um, who's been here maybe a little longer or, you know, he's also very good or he's, you know, just basically someone who's maybe your manager or, you know, whoever is telling you, no, no, actually, you know what?
It's not, it's great, but we shouldn't be doing it that way. Let's try it again. It's all the way, um, we want people who, when they are being told that will, [01:03:00] maybe they will, um, uh, explain, discuss, et cetera. And then they say, you know what? I'll do it, I'll try it again, as opposed to, Oh, you know, this is bullshit.
I know what I'm doing, you know, et and I've seen many, many smart people, very smart people, uh, falling the trap of being too aware of being very smart and reaching a plateau because of that. Um, because you know, they're not open enough to the, uh, to trying again, learning, um, not having necessarily that work mindset. when you forced, um, to be humble, when it was a customer and we forced to be humble, Uh, internally, I think it lends itself to a culture where, you know, you take yourself too seriously. Um, and, you know, leading to the low drama and, and low ego side of things. And I think it's a great thing. For the future profiling organization, I think it's great.
Logan: in an interview and it's kind of innate to people, or is it something that people can [01:04:00] learn when they, when they come in the doors?
Olivier: It's, I think it's hard to assess in interviews. You can get, you get some signals from it. Uh, I think in interviews you understand, uh, you know, who smart people are, um, how well they understand their job. That's good. When you look at their experience, you get some sense of who they've progressed and who they've grown.
Uh, but the two things you, uh, you going to, uh, to learn to see only when they're on the job is, are they actually productive? You know, some people are very smart, very good, but not very productive. Uh, and, and do they actually have the growth mindset? I would say early in career, uh, growth mindset can be trained pretty well.
So we do a lot of campus recruiting. Um, and if you have the right organization, if you have the right managers, the right mechanisms and culture inside the company, I think you can really train. That growth mindset and, and as a result, you know, we've been very successful with seeing people who joined a company as interns, uh, and who now run [01:05:00] huge parts of the organization, uh, just because they they've always on the hunt for the next stage and they want to grow and, you know, they're happily.
Uh, I'm going to take the next problem, make it disappear and then grow one level.
Logan: there quantitative ways that you think about, um, measuring the culture or how do you go about making sure that there's not drift from these principles that you, you espouse?
Olivier: hard. Uh, I don't think we have any hard data on that. Um, and you know, one thing that's specific to us is that we've never written in doubt. So we've never, you know, we don't have like, you know, you know, to have principles or, you know, seven values or any of that stuff. We've considered doing it a number of times.
Uh, maybe we'll do it in the future, you know, as we need to, you know, hire and onboard, you know, thousands of people every year. Uh, but so far we haven't done it. What we do instead is we do the traditional employee surveys. Um, but one thing that we, we still do to this day is [01:06:00] I will personally read every single comment in the employee surveys that we did them twice a year.
Um, and they're a great way of just getting a sense again of the, The fabric of reality, like what are people thinking? What did they complain about? How do they speak about the organization? Who do they, what's happening in there? Um, and is there, do we see some change over time or are there some things, you know, when you slice it by team or geography or things like that, are there some things that we should be paying attention because it looks like in this part of the world, maybe people behave a little bit differently or think a little bit differently.
Logan: I've heard you describe culture, and this is a good reminder as a as who you hire, who you fire and who you promote. Can you speak to that? A little bit of how you how you think about
Olivier: Well, I mean, who we promote is people who are, um, a real growth mindset, humble in front of the problem. Um, low ego, low drama, uh, but also who are doers and fixers, you know? So you want people who are productive, are going to get things done and are good [01:07:00] fixing issues. Like the best people in organizations, you send them problems and order comes up.
Um, and in general, you can get a sense of who these are because, um, you know, it's like fluid, that fluid dynamics, you know, so, uh, Uh, work finds amazing people and avoids the others. Uh, so when you see people lining up to get to work with, with some people on the team that, you know, like they just make problems disappear.
Logan: I know. Seeking bad news is an important thing. We touched on it earlier, but how do you make sure that you're a, uh, that you maintain a culture that seeks bad news?
Olivier: Um, well, I mean, again, you focus on the details. So anything that's aggregated is suspect, right? Because when you aggregate, you lose the color, or maybe you tone down some things. Maybe somebody doesn't want to look bad and, you know, it's completely human, but they don't want to look bad. And they don't include something that reflects poorly on them or the teams, things like that.
So that's why it's important to go back to the source, go back to the data and see what's actually happening [01:08:00] in there. So that's number one. We discussed that already. The second one is having feedback looks that are actually painful to the organization. So they make it hard to ignore, you know, so when you, uh, attach revenue to a new product, which is something we do, uh, you get that, um, very hard feedback when the product was not ready, when it's not good enough.
Uh, when you, when there's a product producing a value for the customer, you get these, the revenues going down, churning, things like that. hard. It's impossible to ignore that. And as a result, you know, you, you sort of have to face reality and, and go for the bad news.
Logan: Are there other, uh, things that force, uh, that, that force the, uh, the, the bad news process that you guys have have found? I mean, revenue is a great one. Are there other things you guys have done?
Olivier: Uh, I mean, look, we, every single step on the way, we'll try was to figure out what's, uh, what's broken. You know, I don't have to look for it under, so I'm the CEO. So usually when something [01:09:00] arrives in front of me, it's bad, it's bad
Logan: Yeah.
Olivier: uh, but, you know, we try to make sure that every single layer of the, of the ignition people understand, uh, when naturally will always make things look better.
Um, so. So when you're a product manager, when you, when you go and get feedback from a customer, if they tell you two compliments and one bad thing, just ignore the two compliments, uh, the one thing they want to tell you is the bad thing. And so that's the only thing you should pay attention to. So again, we try to apply that to pretty much every single part of your, there are some exceptions, of course.
I mean, you, you don't want to be, uh, to be just negative about everything, like, you know, and. Depending on the function and the culture, like people can complain more or less, but, you know, in general, whenever it comes to the business and customers, it's very, very important to, uh, focus on the, on the bad news.
The good news, you're going to, you don't worry, you'll, you'll get them, um, you'll see customers ramp up, you'll see, you'll see them use more, you'll see that's going to be fine. The bad news, you really need some, [01:10:00] need some, some special attention.
Logan: you learned about hiring executives that you would, you know, Tell your younger self for someone that's early in their company journey.
Olivier: Well, so first of all, it's important as I, as I think we touched on that already, but. Hiring two partners is important. I think it gets you so much further when you can lean on an executive team, uh, and have people just run the show. And, you know, uh, I keep joking that my, my, my goal is to grow the team so that, you know, I can be at the bar downstairs all day and, you know, nobody will notice.
Uh, but look at the day, these are the people who need to run the company. So it's, it's. It's important to overinvest in having the right team around the table.
Logan: What, um, I I've heard that, um, I mean, I think this is intuitive, but the most talented people typically are looking for jobs. How, how have you gone about attracting those people now? I'm sure it's easier than it was once upon a time, but, uh, how, how do you think about, uh, getting people to want to join in the early days
Olivier: In the early days, [01:11:00] I think, I think it depends on the, on the functions. Uh, I think when you're small, like, you know, you can attract engineers very fairly easily, I would say, because engineers, I mean, and I'm an engineer, so I can completely relate, uh, love the ability to shape everything from scratch, leave their mark, you know, have a, have a lot of a say, you know, in the architecture.
I would say on the go to market side in sales, it's harder, um, because the folks who are very good at it, uh, will likely go where, you know, the chance of them making a lot of money is higher. Um, and that's typically not with an early stage startup with a lot of different, you know, hypothetical and things like that.
So I think as you grow, you know, you should expect to keep hiring and keep, keep building as you keep attracting better and better people for the, for your sales organizations. The two are a little bit different in that way. Look, at the end of the day, as we, as you scale the organization, um, the, um, the, the thing that matters is to give [01:12:00] people, uh, impact that both on the sales side and on the engineering side is to give people impact, um, make sure that they can build things.
They see the results. They understand how it impacts the business and they understand how it impacts their own career. Um, I think that's completely similar. Even though the, the motivation structures between, uh, engineers and sales might differ a little bit always.
Logan: you say that you sucked at fundraising, uh, in the, in the early days. Do you, uh, do you think you actually, uh, demonstrably better at it over time. Did the business just get, uh, so much more obvious and the numbers just so much more inevitable that, uh, that people gravitated to
Olivier: Well, we'll never know whether I got better at it or, you know, the business itself,
Logan: certainly got easier.
Olivier: it, it, it definitely got easier, but, but also look, like everything else, it's a question of understanding who you're talking to and, um, who the other side of that relationship works, like what are investors, um, looking for, what, uh, [01:13:00] what world do they live in, what, what's the, what data are they looking at, like who do they understand the various categories, like some of the things.
Um, some of the words I was using initially actually didn't work at all. Like I was, I was very naive about what, what the category we're setting into. It was very naive about the, uh, the state of the art, you know, what the other, uh, companies out there were doing. Um, I think you have to understand all that and understand through which prism the investors are looking at you to, um, to, um, to make that case.
And that's something I've had to learn with VC investors. And now something as a company, also, we've had to learn with public investors that we're a public company, we need to understand what they're looking at, how they understand the company, why they would invest in Datadog versus, you know, the other many, many stocks they can invest in, um, and what their frame of reference is to understand how, where we're the same and where we're different.
Logan: Is there, uh, it sounds like each of those, or those two things are fairly distinct and each individual investor might be distinct in their approach and how they're, [01:14:00] they're viewing things from an industry standpoint or whatever it is, were there, uh, was there one or two insights other than just putting yourself in that investor's shoes or trying to understand the language that they were using that you would tell yourself 10 years ago?
Olivier: Well, I think that's, that one's definitely important. The other one is to build relationships. Uh, and that's true. That was true with the, uh, VC investors before, as true with public investors now. Um, Investors like build trust over time. Um, so the biggest advice I give, for example, when I invest in a, in, in new companies and people have, are going through a hot deed or a hot series there or something like that, and this, they say, Hey, I've got what I need.
I don't need to talk to anybody else anymore. Uh, because I've got two term sheets already. And also I'm, I'm, I won't, I won't, I won't spend more time. Um, what I usually tell them is, no, actually talk to a few more people now. You probably will not take money from anybody else right now. That's fine. The people you talk to [01:15:00] right now are going to start doing the work.
Um, they'll ramp up on your company. They'll get to know you. Uh, and then they won't stop doing that work. And then by the time you need to raise the ladder of the next round, uh, they'll have built conviction about your company and they'll be the ones who actually will make your next one happen. So it is worth it to build these relationships over time.
Now that we're a public company, we do the same. Like, you know, the biggest misconception about an IPO is that it's an exit. It's actually not like, it's the starting line, not the finish line. Uh, after that you're on the treadmill actually. And you need to deliver every quarter. Uh, when you go on a roadshow and you talk to all those public investors, uh, what you're not doing is, uh, getting them to invest in the IPO.
Uh, what you're doing is starting this relationship. Um, so that maybe they will invest a bit now, maybe not. Uh, usually they will, but they might sell right away. Uh, but really what you want is you want them to do all the work and track your company and build this confidence in, uh, what the management team is doing, what the company is doing so that over time they will [01:16:00] build and build and build a position and they'll be long term investors in your company.
Logan: I realize it's a deeply personal, uh, decision, but along the way, every successful company seemingly has, um, people beating down their door, trying to acquire business, um, not speaking about any specific things, but along the way, like, As those situations potentially presented themselves, how did you think about the different constituencies, the decision to going versus not just like at a high level, what would you recommend for founders?
Olivier: so, uh, so high level. So first of all, it's against a personal, um, decision. Usually, I think there's usually reasons to, to sell and not sell. Uh, and everybody's in a different personal situation. Uh, for us, the first time we got those, um, um, opportunities, we, first of all, it was an occasion for us to understand that we've had, had come along pretty far, you know, from the company, nobody thought it was going to be successful.
Nobody wanted to [01:17:00] invest in. So that was, it was good in a way to take stock of that. Um, so that was positive. Uh, second, it was very important for me and my co founder that we actually agreed on what needed to happen next. So we spent quite a bit of time. Talking to each other about what do you think of these, uh, just to make sure that we don't end up in a situation where, you know, one is really deciding for the other and the other, the other one resents, you know, and then you end up with a, you know, poor situation a few years down the road.
So we were very careful about doing that. And the last thing is we tried to have a. Um, a framework for deciding, you know, is, does it make sense to sell or not? Um, and it came down to two things for us. One is when you don't sell, um, it means you think you have another 10 X. Can I, can I grow that another 10 X?
Maybe, probably. Okay. So let's try that. The same thing is understanding that when you pass on one of those opportunities, [01:18:00] uh, it's at least a five year commitment. And the question there is, okay, so I believe it's a great business. I believe I can scale it. Uh, and in addition to that, do I actually want to do that day to day?
Like, do I want to do that for at least another five years? I think those are the two ways. I mean, again, maybe those numbers are completely wrong. I don't know, but these are the two ways that we, we try to, uh, to frame things and, and help, decide because otherwise it's just nerve wracking to try and decide.
Logan: Um, artificial intelligence. We touched on it a little bit. Machine learning, uh, there's a lot being said today about how it's going to change software development, uh, if you were an engineer 22, 23, 24, uh, looking at this and you, you hear different, um, considerations about if it's come for certain jobs or not, like, how would you work with the tooling around AI and try to future proof where you're headed in your career?
Olivier: Yeah. I mean, the first thing I would say is [01:19:00] right now, if you're going to graduate in the next year or two, um, you really shouldn't skimp on, you know, matrix multiplications because apparently If you can multiply matrices, you can raise, raise a hundred million dollars, you know, so that's, that's good. Um, but more seriously, I think, look, it's super important for everyone to lean into the tooling there.
I think, uh, software development in general, lend itself very, very well to being amplified, augmented, accelerated by AI. We're seeing that today, and I think we'll see a lot more of it in the future. So it's important to lean into the tooling and, and, uh, and learn, you know, all of that. the way I think about it, by the way, is that.
It's a yet another step in the, um, uh, curve that we've been riding of, uh, a continuous, uh, improvement of developer productivity. Like if you go back 40 or 50 years, uh, you had people programming on the, uh, uh, in, in machine language [01:20:00] and punch cards. You know, and then you had machine language and, uh, keyboards and screen.
Then you had a more advanced language and keyboards and screen. Uh, then you had, but you still had, when you wanted to code, you still had to go by book at the library and, and then, you know, you, you know what there is in the book and that's it. Um, then you had the internet, so you could learn about anything, anytime.
Then you have all of the libraries, uh, in the open source. Then you had SAS and cloud with the APIs you can use to simple things. Uh, so now when you look at the, uh, Uh, the, uh, augmented, um, development, I think that's another step on that curve. And I think every single step of the way, like these major steps, we improve productivity anywhere from no 50 percent to 10X.
Uh, and I think we'll see something similar with the, uh, with AI. Well,
Logan: there are there particular things that you're, uh, paying attention to outside of the core benefits to data dog within artificial intelligence?
Olivier: think to us, the, um, the opportunity we have at Datadog, by the way, is that it's [01:21:00] a, on one hand, drives a lot of cloud migration and consumption and, you know, Um, yes, and, you know, move to storing your data and using more of your data. Like all of that is, uh, like I would say plays into some of the existing ways, um, that we've seen, but also, you know, if you follow the, the, the curve of productivity, I just mentioned, uh, a lot of the value actually is moving from writing the code to understanding it, to, uh, uh, adapting it to the real world, to securing it, to understanding what happens with other people make changes to it.
Um, so actually it falls more into all our part of the world. Um, so it creates a, I would say a very big opportunity for us to go after. Uh, it's also happens, so happens that the, uh, um, AI written code or AI in general makes the code less predictable. Uh, like it used to be everything was very, um, um, mathematically predictable.
Now, now you end up with probabilistic executions and things like that. So I think there's going to be a [01:22:00] lot of, uh, of new areas to cover there.
Logan: So on that, on that point on where you all fit into it, just to make sure I understand. So, uh, the, the, the dev side of the house is increasingly going to spend just less time on the low levels of code writing, and it's going to become more, uh, Dev and ops together in that way and actually understanding and executing.
And so then a system that can understand what's being written and the different bugs and all that is going to be even more important in the hour.
Olivier: Exactly. Like it takes, right now it's, it still takes a lot of time to draft the code. Like you, you have to figure out what to type, you you go on Stack Overflow, you, you know, you spend more time, then you fight with the compiler, like all that stuff. Uh, I think a lot of that is going away.
I think now to get from, okay, maybe I should do this to have a first version of it that I can plop in my, in my code, I think can go pretty fast. Uh, but then the question is, does it actually do what I want? Uh, [01:23:00] oh wait, somebody changed some other module, what does that, what does that, how does that impact what I do?
Um, I see those errors in production, what do they mean? Um, or, you know, it looks like there might be a security vulnerability here, how do I address that? So all of those things still have a lot more to do with the behavior of the code and the world around it than the code itself. And I think that's where a lot of the value is going to be moving forward.
Logan: So I guess to wrap, um, New York. So you've been here now 25 years. Uh, you resisted moving out to the Bay area. Was that actually asked of you in the early days of data dog? Did investors,
Olivier: Yeah. Some, some investors, um, uh, hinted that it would be easier for us to raise money if we went to the Bay Area and also that there was more talent there for what we're doing, which is true. There was more talent there for what we're doing.
Logan: was data dog a different business in your mind than it maybe would have been? And I realize it's hard to totally know, but,
Olivier: Yeah. I mean, again, counterfactuals are always difficult, but I would say, look, [01:24:00] just looking at the, the companies we competed with, because when we started, there were like a, every single major VC fund had a champion. Like they had a, uh, A company led by super smart people that were way more fit to market than we were.
Like they had worked at, uh, hyperscalers before or work in the, in the systems management field, which my co founder and I hadn't. So these people had all the ingredients for being successful. And
Logan: had already backed people in this monitoring observability
Olivier: exactly. Or they did shortly after, like, well, you know, there was always that. And we were a lot more successful than they were.
And I think part of it is. Yeah. Um, we were closer, further from the tech world, but closer to the real world. Um, so less into the, uh, the echo chamber of what kind of makes sense to the tech community right now. Um, and more into the, but what are the real problems that the real companies have about this [01:25:00] and, you know, what can we solve that one thing.
The second thing is, look, we were not nearly as hot as the other companies. Um, so we could never, um, uh, think that we had it all figured out and we just needed to ship the product and the, and the customers would come, we were always super scared that we were not going to get it right, that we're going to be out executed, out funded by, you know, just about anybody out there.
And so we focus super hard on those early customers and, you know, all of the values we discussed earlier, you know, whether that's the, uh, the Efficiency of the business, but also the humility in front of the customer, all that, you know, really comes from that. And I think that's a, uh, that's the risk I see today, by the way.
When I see, um, young founders, um, getting super, super successful, very, very early in terms of the, the fundraising traction. I worry a lot that they're not going to be set up for success. I see that too with ai, for example, right now, where a lot of, uh, brand new companies [01:26:00] raise amazing amounts of money without much to show initially.
And the, the risk there is really, um, to, uh, to fail to understand that, yeah. But at the end of the day, you are just here to solve somebody else's problem. And your focus should be completely on that. Anything else, whatever anybody else is telling you about how great your business is, doesn't mean anything because you haven't solved that problem for the customer yet.
Logan: It seems like every, uh, successful company has at least one difficult fund raise and it's I think one of my views is that if it comes earlier, it's It's better than coming later. 'cause it puts all those principles, you don't need to culturally reset all of that stuff along the way. So I guess to some extent, the early days being more difficult, uh, maybe was a benefit, uh, to, to the ultimate success.
Olivier: And look, it's like being, um, being efficient and discipline is a good thing, like in the long run, it works really, really well, you know, so. If I look back at the markets like a few years ago, you know, valuations were crazy, [01:27:00] um, the, uh, you know, we're in no interest rates and the, uh, and everybody was spending, uh, even on public markets, everybody was spending, um, you know, a ton to, to grow after that, there's been quite a bit of contraction of the multiples, the, uh, uh, companies had to be profitable all of a sudden in the public markets and in the private markets.
Yeah. And I think it really, um, was very, very painful for many companies in our space. Uh, as you know, they had to, you know, turn on a dime and, you know, all of a sudden they were not able to invest anymore in our case, it didn't change all that much. So we tend to die a little bit, you know, so we, uh, we grew profitability a little bit further and faster than we would have.
Otherwise we slowed down a little bit of hiring just because we were not completely sure what the, uh, the future would look like. Uh, but you know, we were still able to keep investing, you know, 30 percent of our revenue into R and D. And that hasn't changed the posture of the company all that much. I would say there's a lot of long term benefits to building [01:28:00] a company that is, um, you know, disciplined and profitable.
Logan: your sales pitch to future founders, uh, that are thinking about picking a location, a headquarters. Uh, do you think all the things you said earlier about ability to focus on the customers and being removed from Silicon Valley? Does that still hold true in the world that there's more venture firms here?
And would you recommend New York for the next
Olivier: I think so too. I think, look, there's definitely some trade offs. I think if you, uh, for types of talent, there's still a lot more talent in the Bay Area than in New York. Like, you'll have more choice. That's a given. On the flip side, that talent typically comes with a more employee turn. So people change jobs.
More it's, there's more places to go to for one thing, but also it's more in the culture. Like it's a bit weird to stay a very long time, you know, in a place, you know, in the Bay area, whereas in New York, it's, uh, more common. Um, like if people are happy with their jobs and treated well [01:29:00] and working on interesting things that tend to stay longer.
And then if you go to Europe, which is the other side. It's even, even more so, like people always say, even longer. So for us, you know, we, we struck the right balance of, uh, um, uh, uh, employee tenure and, uh, choice of talent by being in New York. I think it's a great equation there too. And by the way, I'm a firm believer in long tenures.
Um, I think you learn from your mistakes, um, and you need these feedback loops. I think we, it's a, it's been a theme
Logan: Yeah, yeah.
Olivier: and the feedback loops take years in what you build. Um, so if you are in a job for a year and a half, uh, you ship something six months in, uh, you, when you leave, you still think it was a genius idea.
Uh, you haven't seen the many ways in which it was not great or not perfect. Um, and you haven't learned. Um, so I compare that to having a, uh, you know, a, uh, a 10 miles experience of running a marathon. Uh, you just have no idea. [01:30:00] Um, and I think when I, when, when we hire, by the way, one thing we look at is did people build longer successful tenure?
Like did people spend three, four, five, six years at a place? Uh, because typically when you do that and you grow at the same time, like that's when you actually have true experience. People being promoted just by jumping around actually doesn't work that well in the long run, might work a little bit in the short term, but in the long run, you don't actually experience the same way.
Logan: Super interesting. Olivia, thanks for doing this.
Olivier: Alright, well, thank you very much.
Logan: this is great.
Thank you for joining this episode of the Logan Bartlett Show with co founder and CEO of Datadog, Olivier Pommel. If you enjoyed this conversation, we'd love for you to share with anyone else that you think might find it interesting, as well as subscribe on whatever platform that you're listening to this on.
We look forward to having you back next week with another great guest on the Logan Bartlett Show. Have a good weekend, [01:31:00] everyone.