Jan 22, 2020 · 47m · mad
Fireside Chat with Olivier Pomel (Founder & CEO of Datadog) and Matt Turck (Partner at FirstMark)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this fireside chat from Data Driven NYC, Datadog Founder and CEO Olivier Pomel joins FirstMark General Partner Matt Turck to discuss the founding, architectural vision, go-to-market strategy, and scaling of Datadog into a leading public observability company.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 15.7% of the talking time here. How this is scored →
Matt as informed peer 2.6 Guest teaching 3.0 Guest disagreement 0.4 Matt pushing back 0.1
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 25:01 Guest reframing freemium model
Pomel gently corrects the host's premise regarding freemium offerings, explaining why freemium is fundamentally incompatible with full-coverage infrastructure monitoring.
Hardest push from Matt ▶ 8:53 Host probing open-source usageThe host directly challenges the guest on an industry rumor that Datadog avoids open source or third-party tools, forcing the guest to clarify their development lifecycle.
Biggest teaching moment ▶ 14:22 Reality of AI and false positivesPomel educates the host on why generic AI fails in technical observability, explaining the dichotomy between what customers say they want and how they actually react to false positives.
Matt holds his own ▶ 5:25 Defining pillars of observabilityThe host showcases domain expertise by proactively naming and categorizing the three pillars of observability (metrics, traces, and logs) to structure the technical narrative.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Defining DevOps and Observability Basics | 1 | 3 | 0 | 0 | The host asks basic level-setting questions asking the guest to explain DevOps like he's five. The guest provides a friendly, foundational overview of the space and company history without any friction. | |
| The Origin Story Behind the Name Datadog | 3 | 3 | 0 | 0 | The host demonstrates basic domain knowledge by citing the three pillars of observability (metrics, traces, and logs). The guest elaborates on each pillar in an agreeable and educational tone. | |
| Datadog's Technical Architecture and Real-Time Data Ingestion | 4 | 3 | 1 | 1 | The host probes into Datadog's architectural choices and asks if they avoid third-party or open-source software. The guest gently corrects this belief by explaining how open-source is used initially before scaling out. | |
| Horizontal Product Expansion Strategy and Synthetic Monitoring | 3 | 3 | 1 | 0 | The host asks about horizontal product expansion and synthetic monitoring. The guest reframes how product success is defined, emphasizing broad adoption over RFP feature lists. | |
| Machine Learning, Watchdog, and Practical AI Applications | 3 | 4 | 1 | 0 | The host brings up machine learning and Watchdog. The guest educates on the realities of AI in IT ops, dispelling popular assumptions about false positives and highlighting why general AI fails in technical observability. | |
| Early Startup History, Team Building, and Content Marketing | 2 | 3 | 0 | 0 | The host guides a discussion on early company history and content marketing. The guest explains why typical content strategies like engineer blogging or marketing-written content fail. | |
| Go-To-Market Strategy and Bottom-Up Sales Dynamics | 3 | 3 | 1 | 0 | The host asks about sales motions and freemium models. The guest clarifies that Datadog offers a free tier rather than a freemium model due to the nature of infrastructure monitoring. | |
| Iterative Pricing Strategy and Value Alignment | 3 | 3 | 0 | 0 | The host asks about pricing strategies and value alignment as data scales. The guest candidly shares arbitrary early pricing decisions and how they decoupled log ingestion from retention. | |
| Navigating the IPO Experience and CNBC Floor Interview | 2 | 2 | 0 | 0 | The host asks about the IPO experience. The guest shares the reality of long-term investor relationship building and an amusingly terrifying CNBC floor interview experience. | |
| Founder Leadership Evolution, CEO Scaling, and Early Doubts | 2 | 3 | 0 | 0 | The host asks about personal growth as CEO and overcoming early doubts. The guest explains how he avoids micromanagement by consuming high-volume passive data feeds. |
Statements from this episode (19)
Pomel: Datadog was created to unify development and operations around shared data
“The starting point for Datadog was, no, there must be a better way. Maybe those people can talk to each other. And we wanted to build a system basically that brought the two sides of the house together. All of the data was relevant to both. Make sure they spea…”
Pomel: Datadog was founded to unify metrics, traces, and logs
“And one of the main reasons why we started a company was to bridge the gap between the different teams and break down the silos. It didn't make sense for us to have those things being separate and, you know, it turns out the problems actually don't stop at the…”
Pomel: Datadog starts with off-the-shelf software before building proprietary systems
“Well, we usually, when we build a new system or add a new data type, we usually start with something off the shelf and something open source. And at a small scale it's going to work. And then as we grow, as we get more and more data into it, as we see the many…”
Pomel: Datadog processes about 10 trillion records per day
“So we're talking about, like, 10 trillion records a day, something like that?”
Pomel: Datadog retains the last 24 hours of incoming data in memory
“The way we run it under the hood is that we're going to keep all of the data that's from the, let's call it the last 24 hours in memory so it can be accessed very, very quickly, and over time we age this data to various forms of, you know, colder and cheaper s…”
Pomel: Commoditized monitoring tools succeed when bundled into broader platforms
“It's a category that is not necessarily super interesting on its own, you know, because it tends to be a bit commoditized and, you know, it's a little bit high churn but it actually makes a lot of sense as part of a broader platform, which is what we offer. So…”
Pomel: Datadog defines success by broad team adoption, not buyer purchase
“The way we define success for those products is we want to get deployment and adoption that is as broad as possible at our customers. You know, so again, going back to what we're trying to solve here, we want to bridge the gap between the teams and bring every…”
Pomel: Leading with AI in IT monitoring will inevitably disappoint
“It's very hard for what we do to lead with, you know, AI because you, you're bound to disappoint, because the generic case is very, very hard to solve”
Pomel: Customers say they want false positives but hate noisy software
“When you ask, when we first visited customers to figure out what they wanted from our product. When you ask them to pick between a false positive and a false negative they'll say, oh, give me the false positive, I'll decide if it's right, but I want to know. I…”
Pomel: West Coast VCs dismissed NYC-based infrastructure startups
“Whenever we pitched West Coast investors, like, it was sort of seen as a form of mental deficiency to be to be based in New York and, you know, doing it, doing infrastructure.”
Datadog hired its first salesperson at under 30 employees
“We were, I think, a bit less than 30 when we hired the first salesperson.”
Pomel: Sales teams cannot figure out product positioning for startups
“When you hire a sales team, you need to have all that sort of in place. Otherwise the sales team is not going to figure it out for you. You still need to do that yourself.”
Pomel: Effective technical content requires hiring engineers reporting to engineering
“In the end we ended up hiring full-time engineers who are also interested in journalism, and those people exist and have them basically focus on research and writing and we have a team basically that's built that way. It reports into engineering, it doesn't re…”
Pomel: Datadog raised launch price to $15 then $18 without metric impact
“When we launched a product we priced the product initially, we were going to price it at 12 dollars per, you know, instance, per month and then the night before we said, let's put 15. And. It was, I cannot disclose that. It, We put 15, it worked fine and, you …”
Pomel: IPO preparation is mostly about the following 18 months
“Most of the work you do there is really to prepare for the 18 months that, that follow the IPO.”
Pomel personally monitors customer tickets, closed deals, and employee feedback
“I see a large volume of the tickets our customers are filing, so I want to stay in touch with what's happening with the product. And I'm going to see all the deals we close, and I'm also going to read all the comments we get from employees in employee service.”
Pomel: Real competition comes from novel approaches, not late copycats
“If their plan is to do whatever you're doing, or a fraction of whatever you're doing, ah, four years later, ah, that's not where your competition is going to come from. Your competition is going to come from people who are trying to do it differently, and Are …”
Pomel: Datadog never pivoted and built its original vision
“No, so we didn't pivot. So we were, it's a great question because when we didn't manage to fundraise initially we sort of gobbled the pitch quite a bit. We tried to turn it in different ways, and we thought maybe we'd, we'd do something a bit different, and we…”