YC Root AccessHow Olivier Pomel Built Datadog By Refusing Every Shortcut
At a glance
WHAT IT’S REALLY ABOUT
Datadog’s CEO: scale by staying close to reality, always.
- Olivier Pomel recounts how early rejection and fundraising difficulty led Datadog to build a durable, customer-value-first culture designed to be sustainable even without easy capital.
- He attributes Datadog’s long-term co-founder stability with Alexis to years of pre-founding collaboration plus ongoing, intentional communication and explicit alignment on existential choices like acquisition offers.
- Pomel explains his operating system for running an 8,000-person company: resist “managing up” by sampling ground truth directly through support, sales, and product artifacts and by designing processes that keep leadership connected to reality.
- Datadog’s product strategy scales through decentralized shipping with limited approval gates, and it expands the platform by watching what customers attempt to build on top of it while preserving room for both strategic bets and bottom-up innovation.
- AI is portrayed as a re-acceleration vector that increases system complexity and shifts importance toward production validation and operations, while also enabling the long-sought goal of automated detection and remediation of incidents.
IDEAS WORTH REMEMBERING
5 ideasEarly rejection can harden culture—if you turn it into customer-obsessed execution.
Early rejection (including YC and many VCs) pushed Datadog to prioritize real customer value and a business that could survive without constant fundraising. Pomel frames that pressure as foundational to a culture of practicality, endurance, and proving value in the market.
Long co-founder longevity is engineered through deliberate communication and alignment on “sell vs. build” decisions.
Pomel and Alexis had over a decade of working together before founding Datadog, giving them a deep read on each other under stress. They protect the relationship with recurring, agenda-less time to talk and they explicitly align on high-stakes topics like acquisition offers to prevent founder drift.
At scale, leaders need systems that force contact with ground truth—not prettier dashboards.
As Datadog scaled, Pomel built mechanisms to counter “managing up” by sampling reality directly—reading support tickets, sales notes, product briefs, and releases. His occasional questions create a jolt that forces managers to look downward at customer and product truth, not just polished narratives.
Avoid executive bottlenecks by making feedback available but rarely required.
Datadog’s default is execution without waiting for executive sign-off, with approvals reserved for changes that are hard to reverse (e.g., pricing/packaging). For most product work, teams ship, test with subsets of customers, learn quickly, and expand rollout—keeping feedback possible but not blocking.
Product expansion works best when you formalize three pipelines: customer pull, strategic bets, and internal invention.
New products often emerge from observing what customers build around Datadog (extensions, scripts, DIY solutions), signaling where the platform should expand (e.g., infra monitoring → APM/tracing; security automation). Pomel describes three inputs: customer-driven (reliable but incremental), top-down “strategic” bets (riskier but necessary), and bottom-up internal innovation (keeps novelty alive).
WORDS WORTH SAVING
5 quotesAnd, uh, you know, our little revenge on life has been that we did make the platform successful in the end.
— Olivier Pomel
So everything, uh, I do there in terms of the systems we put in place is to force people to look down instead of up, so to look at what's actually happening, what's happening with the customers, what's happening with the product, what's happening with the technology, what's working and not working.
— Olivier Pomel
In the short term, the, uh, the markets, the st- the stock market is a voting machine. Uh, in the long term, uh, it's a weighing machine.
— Olivier Pomel
So a lot of the value ac- is actually shifting from the writing of it to the validation and the operations. Um, and that's where we are. We live in production. We live in the contact with the real world.
— Olivier Pomel
Whatever your instinct is, it's too slow.
— Olivier Pomel
High quality AI-generated summary created from speaker-labeled transcript.
