So, I'm very excited to welcome Olivier and Amish. Olivier is the co-founder and CEO of Datadog, the iconic cloud monitoring and security platform. He started in 2010, took public in 2019, and has now got a market cap of around $80 billion on the back of really strong AI tailwinds. They've done an amazing job capitalizing on and building towards. Moderating is co-founder of FirstMark, Amish Ghani. So, please join me in giving them a warm welcome to the stage.
All right. Hopefully, you guys aren't tired of hearing my voice. So, super excited to be doing this with Olivier. You know, we were reflecting sort of backstage that this is, I don't know, probably the third or fourth different conversation we've had with him over the years as a New York-based firm, New York-based entrepreneur. You know, you'd been part of Wireless Generation, which was one of the great early exits in New York, and obviously have built an iconic company that has stood the test of time thus far and will continue to do so. So, it's a real privilege to have this conversation at multiple points along the journey, and it's amazing to see what you've done with the company and what you've built. So, thank you for doing this with us.
Well, thank you. Great to be here.
So, I want to just talk about the changes that happened. And let's start with, you know, probably a decade ago you would have been at a couple hundred million dollars in revenue. You're past 3 billion, approaching four. What changed?
4 billion, approaching five.
Well, so, we're going to get to the re-acceleration.
Who's counting, but you know.
There we go. Just dropped something.
It's just a mic. It's okay.
Just a mic. All right, we'll keep that tucked nicely under here. So, we're approaching five now, so the business is accelerating. I want to come through some of the transitions in the journey that you've had along the way. But, what changes as scale moves at that level for you running a business from the early days to today? And a couple hundred million to now many billions.
You know, actually not a ton changes. So, I actually spend most of my time on the same thing I did when we were a tenth of the scale. So, I spend my time mostly on product and what we build and what's working and what's not working. And the rest of my time mostly on either debugging some of the systems we have inside the company. So, you know, how are we doing in various parts of go to market, like these are all systems. And who do we have in charge of what? And that's been the same thing.
There's a new thing, you know, since the IPO that's been the speaking to the outside investors. But, that just replaced speaking to VC investors.
It's a little bit different like VC investors, you know, as you know, it's all the time.
Like, you know, my job as a CEO of a private company was to make sure that every single VC thought we were the future of sliced bread and you know, we were jumping at the bit to invest. And so, you still talk to people pretty much all the time. Make sure people know about you. Make sure that they understand the business. So, it's very time-consuming all the time. Whereas speaking to public investors is something that happens, you know, once or twice a quarter. It's very choreographed. There's a whole team that handles it. And then, you're not supposed to say anything new anyway when you meet new investors. You say the same thing over and over again.
Maybe at scale you would think that information flow is a little more challenging. And you've got potentially more layers. You know, Anjali Sud, the CEO of Tubi, I spoke with her this morning. And one of the things she talked about is going from her marketing position to her CEO position was all of a sudden it took a lot longer for news to travel, bad news in particular to travel.
Oh, good news they find you. Yeah.
Good news they'll find you anywhere. Do you find that to be true or do you have ways in which you can make the company feel smaller than the scale that it is?
So, it doesn't really look that way, but I'm a bit of a control freak. And the one thing I've done, I think, pretty early on is that I like to sample the reality of the business. Which means I'm on pretty much all the mailing lists and everything, you know, it all goes to filters, of course, but I'm going to read some support tickets. I'm going to read the summaries of the deals we're making with customers. I'm going to read the product release briefs. I'm going to read all of that stuff. Of course, I can't read everything. I mean, product I read everything, but the rest I can't read everything.
And what I'll do is I'll just answer with a question, like to whoever is sending that. So, there's a few positive effects to that. One is I get to really sample the fabric of reality and I can try and make sure that what I see with my own eyes on the ground agrees with the beautiful story that made its way to me, you know, so when I hear, 'Oh, that product is doing amazing.' And then I see customers saying, 'Well, we're going to churn, you know, it's not doing what we need.' I have to ask, you know, am I just unlucky with who I'm looking at or is there a problem there? But, that's the first effect, so I get a better idea of what's going on.
The second effect is whenever you fire a question like that, people immediately wonder why the hell is he looking at that, you know?
How did that get on his radar?
Yes. And but a positive effect is that it sort of focuses everybody in that hierarchy down instead of up. You know, so people start to realize I better understand what's going on there, so I can explain it as opposed to let me figure out what the boss wants to hear next.
But you know, as we grow, I think there are still more things that need to be put in place. So, for example, what had started happening is that I would ask a question and then a focus group would be assembled to answer the question, like a document would be put together, people would review it, you know. And so, I think now the next step is to have an SLA for a response, like, I need to respond within 10 minutes.
Yeah. So, you know, this way, you can't workshop it, like, it has to go back, you know.
Yeah. Makes sense. In one of our earlier conversations, I think you said the biggest fear you had early on was building the wrong thing. What's your biggest fear now? Is it still the same? Something else?
Still mostly missing the boat on what we need to build and what problems actually matter and what sounds good to us versus what solves real problems.
And you know, this fear, I mean, it came from two things. One is, so you mentioned I worked at Wireless Generation before, which sold to teachers. And it was very clear that what made sense to us, you know, engineers building it, was very different from what was valuable to teachers.
And so, that drilled into us the fact that just because you think it's going to be cool doesn't mean it's going to be a good product or useful product or valuable product. That was first. Second is when we first started raising money, I think it was around 2010, 2011. It was very difficult. And so, we were not under the illusion that we had it figured out and we thought, 'Oh my god, we're not going to make it, so we better focus on the fundamentals and make sure we build the right product.'
I would say we've done a very good job of that over time and we have a whole number of systems to make sure we build the right things and we have the right signals built into the product. So, we make sure they're not only cool products, but also valuable products. But I think it's becoming more difficult now because we have, especially with everything that's happening with AI, the markets are changing so much faster that you don't have as much time and it's harder to understand where things are going. So, I think it puts more pressure on that part of the business, you know, which is how do you make sure you build the right things and you go where you need to be 6, 9, 12 months from now.
So, do you fear more being too slow or wrong?
Both. They're both completely related.
But one helps maybe like if you're really fast, you might have more variance. If you're really slow, you might get it right, but too late to not matter.
Well, I think being wrong is okay as long as you're also right sometimes. And I think part of the equation here is that we're going from a world where you could assess a problem for a year or two before you'd solve it and then you'd be right 70% of the time to a market where you only have two or three months to get a sense of what you need to do and maybe you'll be right 30% of the time. But you have to explore more possibilities at once so you have enough of it that's right in the end.
And I think everybody, probably everybody feels that in the room. Like it's things are changing so fast that you have to be willing to be wrong and to reassess.
Yeah. Datadog is very famous for being a company that has just outputted products at a very, very rapid scale. You have 20s, 30s number of products. There are five that are over $100 million in scale from the research we did, three that are between 50 and 100. You have 18 others that are smaller, earlier, some other dimension. How do you think about what moves the needle for you and how do you manage sort of the smaller products that you have? How do you think about that?
So, the smaller products for us, they're mostly larger products in waiting, you know, so when we work on products, we think they all have a significant market. And typically, we have some idea of that because of the size of the problem we solve or the size of the, maybe sometimes there's already an existing category and we got a sense of how big that category is and what people already spend on that, you know. So we always have high expectations.
Some of them are on a slow burn. Some products have a fairly large amount of table stakes functionality you need to bring so that you can get them to be immediately useful and sell them out. Some of them, we make mistakes and it takes longer. Some others, they blow up very quickly. You know, our top three products are all at or above a billion in revenue. We have a few others in the hundreds of millions. And then we have, I forgot we have that in our disclosure. I don't want to give you the wrong number, but we have a dozen that are in the tens of millions, you know.
And do you think about an active pruning process? Do you view them as experiments and if they don't or do you have the desire to kind of keep those going and compounding with a small team to see if they meet the moment at some point?
Not really a pruning process. It's more of a, we adjust the increase of resources we give to those things. So, what we never do is we don't do it the Apple way. So, we don't assemble a whole team, build in secret for 3 years, unveil it and then that's it. And then we kill it if it doesn't work. Instead, we start with a small team. We typically build with customers. We have a number of milestones along the way where we try to assess that it's doing the right thing and it's valuable.
And for that, there's a few things. We start taking monetary commitments from early design partners. Then we start onboarding private betas, charging the private beta users, onboarding the broader user base, charging them, auto charging them over time for usage. And every time along those lines we get some further validation that the product is valuable already. And then we can scale the investment we make in them.
What happens in a world of now this world of AI where you can build a lot more stuff and you can ship a lot more quickly. Maybe talk about can you increase the velocity of what gets shipped? How do you get an organization to absorb and then move that out into market? Does distribution become the bottleneck, not product creation? What are you seeing happening in this world of AI build and product velocity?
Yeah, so far, I mean, we are pretty straightforward that we're bottlenecked on the distribution side by the sales capacity we have. And even though we have a bottom-up model, the model is very much built on sales capacity in the end. Like you need the sales capacity to convert the usage.
And on the product side we're limited by the engineering bandwidth to build enough of the right products and add them to the platform. So, our general inclination right now is we keep scaling the go-to-market side. And we don't see a big change in that in the future because, at least for now, it's humans buying from us. So, the humans are useful when it comes to covering the market and being in enough conversations.
On the product side we're also still hiring. But we're also changing the way we work so we can build more and build faster. We think that we'll just absorb the increase in productivity and produce more and have more products to ship. But we still have to cover like I think nobody's been doing that for more than 6 months. So we still have to see what comes out on the other end of that.
But to your point, I think the hard part is still figuring out what's worth building. And what's valuable and what's not. And I think the one thing we build in the company is a very strong muscle around that. Sometimes I get asked, for example, you have so many SKUs, it's hard to deal with, I have all these different things on my bill. And there's a reason for that. The reason is that's the signal we get. Like, without that, we don't know what's valuable and what's not.
Which part of the product is most valuable?
Yes. And we've lived that so many times. Like, we build a product for engineers. Engineers love the idea of cool stuff. So, if you talk about these new amazing features that you're going to work on, you propose them to customers, and they say, 'Yeah, of course I want that.' And so, you start building a product, you create a design partnership with a few customers, and they're your best friends, they show up every week for the calls when you show the demos of the product, they try them, etc. And then, 2 months in, you tell them, 'Oh, by the way, it's going to be 10 grand.' And then, they ghost you. They never reply, you know, they were your best friends, but you know. And that is the difference between a cool product and a valuable product. And we've learned that a number of times.
