Enter an account through one cheap product with instant time to value, then expand across many tools on a shared data backend so each new use compounds retention.
Datadog
Datadog is a cloud observability platform that lands cheap on a single monitoring product and then expands across an ever-growing menu of tools inside the same customer account. The through-line across its stories is compounding inside accounts: a low-friction wedge, one shared data backend, usage-based pricing, and perfect timing on the cloud shift combined to produce some of the highest net-retention numbers in software history, all built by two founders who had lived the exact problem they set out to solve.
Land and Expand: A per-host wedge that turned into a 146% expansion machine
The problem. Selling a broad monitoring suite top-down means long procurement cycles and a big upfront bet from a customer who has not yet seen value. Datadog needed a way for a single team to start using it without a committee, a budget line, or a rip-and-replace decision.
The approach. Datadog landed on one easy, fast-time-to-value product, infrastructure monitoring (launched 2012), priced per host and adopted by dropping in an agent so value was visible in minutes. It then expanded the product menu rather than just seat count, shipping APM, Logs, security, RUM, and synthetics onto the same platform, each a new expansion vector inside accounts it already owned.
How it solved it. The S-1 disclosed a dollar-based net retention rate of 146% for the twelve months ended June 30, 2019 (141% in 2017, 151% in 2018), meaning existing customers spent roughly 46% more year over year. About 60% of the revenue increase in the first half of 2019 came from existing customers, so Datadog could have stopped signing new logos entirely and still grown substantially on expansion alone.
Moats: One data backend that makes every added product raise switching costs
The problem. Monitoring had fragmented into separate tools for infrastructure, application performance, and logs, owned by different teams who could not see each other's data. A single-purpose tool is easy to rip out; Datadog needed adoption to become load-bearing and hard to unbundle.
The approach. Datadog built every product on one shared data backend and UI, so adopting a second or third product is nearly frictionless and grows more valuable as data correlates across them (a metric spike, the trace behind it, and the log line that explains it, all in one place). That cross-product correlation is a within-account advantage a standalone rival selling one box cannot replicate.
How it solved it. Multi-product adoption compounded from IPO onward: by 2025 roughly 85% of customers used two or more products and about half used four or more, with the 8-plus-product cohort the fastest growing. As customers layer on products, APM and Log Management each later crossed $1 billion in annual recurring revenue, and each added product deepens the unified-data advantage and the cost of leaving.
Recurring Revenue: Usage-based pricing that grows with the customer automatically
The problem. Classic seat-based SaaS pricing decouples revenue from how heavily a customer actually uses the product, so a customer scaling its workloads 10x might still pay for the same handful of seats. Datadog wanted revenue to track the value delivered as customers grew.
The approach. Datadog priced bottoms-up and usage-based, principally per host (and per volume for logs and other products), so the bill scales as customers add hosts, containers, and workloads. Growth becomes a function of the customer's own expansion rather than a separate upsell sales motion.
How it solved it. Because expansion was largely automatic, the same 146% net retention translated into revenue nearly doubling from $100.8 million in 2017 to $198.1 million in 2018 (about 97% growth), a rare combination of hyper-growth with a capital-efficient, near-profitable profile at IPO. The pricing model meant Datadog could grow roughly 50% from existing-customer expansion alone.
Why Now (Timing): Riding the cloud and microservices shift as complexity exploded
The problem. Traditional monitoring assumed relatively static, long-lived servers. As companies moved to cloud, containers, and ephemeral, auto-scaling microservices, that model broke, and no one had built monitoring for architectures where the infrastructure itself was constantly appearing and disappearing.
The approach. Datadog bet specifically on cloud-native monitoring, launching its Infrastructure Monitoring product in 2012 purpose-built for ephemeral, distributed systems at the exact moment enterprises were beginning that migration in earnest. It positioned itself where the pain was growing fastest instead of defending the old on-premise world.
How it solved it. The same architectural shift that created the pain also drove the growth: as customers migrated more workloads to the cloud and spun up more ephemeral hosts, per-host usage-based revenue expanded as a structural tailwind rather than a sales effort. This is visible in the retention: expansion came from customers "adding more applications to the cloud," which is why net retention held above 140% through the IPO.
Founder-Market Fit: A dev and an ops engineer building the fix for their own silo
The problem. The core problem Datadog attacked, developers and operations teams working at cross-purposes because they could not see the same data, is easy to misunderstand from the outside. Solving it required knowing viscerally how the dev-versus-ops divide actually feels day to day.
The approach. Olivier Pomel and Alexis Lê-Quôc, who met as undergraduates at École Centrale Paris and worked together for nine years at Wireless Generation, had stood on opposite sides of that divide, Pomel on the dev side and Lê-Quôc on ops. After Wireless Generation was acquired by News Corp, they founded Datadog in New York in 2010 explicitly to break the silo and get dev and ops onto the same dashboard.
How it solved it. Building for a problem they had lived produced a product whose central value proposition, unifying teams around shared real-time data, became the foundation of everything above: the low-friction agent, the shared backend, and the multi-product platform all flow from that original insight. Pomel remained CEO as Datadog grew from that single agent into a platform used across multiple products by the large majority of its customers.