Eppo

Eppo is a software platform for running and analyzing experiments — chiefly A/B tests and feature-flag experiments — that companies use to measure whether product and marketing changes actually work. In an A/B test, some users see one version of a feature or page and others see an alternative, and the platform measures which performs better on the metrics that matter. Eppo's distinguishing approach is that it is warehouse-native: rather than collecting and storing experiment data in its own separate system, it runs its analysis directly on top of the company's existing cloud data warehouse, where the business already keeps its metrics. It also handles feature flagging — switching features on or off for chosen groups of users — so teams can roll out a change gradually and experiment on it at the same time. Eppo is, in short, a tool for disciplined experimentation built close to a company's own data.

In 2025, Eppo was acquired by Datadog, the observability and monitoring company, and is now offered as Eppo by Datadog. This is a factual status point worth stating clearly, because a glossary entry should reflect current ownership: Eppo continues to serve existing and new customers as part of Datadog's product suite, which folds experimentation and feature flagging alongside Datadog's product analytics and monitoring tools. For a reader, the significance is that Eppo is no longer an independent startup but part of a larger platform, which can affect its roadmap, pricing, and how it fits with other tools. The underlying purpose is unchanged: help teams run trustworthy experiments and measure impact using the company's own data. Where possible, confirm the latest product details against the vendor's current documentation, since ownership changes are often followed by rebranding and repositioning that a static definition cannot fully track.

Eppo is best understood against generic A/B testing tools, because experimentation platforms are not all alike. Many classic testing tools, especially in marketing, are client-side and self-contained: they inject variations into a web page, collect their own clicks and conversions, and report results within their own system. That is convenient for quick page tests but tends to keep experiment data in a silo, use the vendor's own metric definitions, and struggle with deep or company-specific measures. Eppo takes a different path by being warehouse-native: it analyzes experiments on the data already sitting in the company's cloud warehouse, using the business's own metric definitions rather than a separate, shallower set. That makes it better suited to rigorous, data-team-led experimentation across product and growth, where the metrics that matter are complex and must match the numbers the rest of the company already trusts.

The trade-offs follow from that design, and they shape who each tool suits. A generic, client-side A/B tool is fast to deploy, friendly to marketers, and fine for testing headlines, layouts, or button colors, but it can drift from the company's real metrics and is harder to trust for subtle effects. A warehouse-native platform like Eppo demands a data warehouse and some data-team involvement to set up, but it rewards that effort with statistical rigor, consistent metrics, and the ability to measure deep outcomes like retention or revenue rather than just immediate clicks. It also pairs experimentation with feature flagging, so engineering rollouts and measurement live together. The distinction is not that one is A/B testing and the other is not — both run A/B tests — but that Eppo is built for warehouse-grounded, rigorous experimentation, while a generic tool optimizes for quick, self-contained page tests.

Place Eppo accurately as a warehouse-native experimentation and feature-flagging platform, now part of Datadog, chosen mainly by product and data teams that want rigorous tests tied to their own metrics. It fits organizations that already run a cloud data warehouse and care about measuring real, deep outcomes — retention, revenue, lifetime value — not just surface clicks, and that want experimentation and gradual feature rollout handled together. When comparing tools, weigh the setup cost of a warehouse-native approach against the trust and depth it buys, and consider a lighter client-side tool if the need is only quick page tests. Because ownership changed in 2025, check current documentation for the latest on how Eppo integrates within Datadog, its pricing, and its naming, rather than relying on older descriptions.

The traps, for a glossary reader, are describing Eppo as a simple page-testing widget when it is a warehouse-native experimentation platform aimed at rigorous, data-team-led testing; ignoring its 2025 acquisition by Datadog and presenting it as an independent startup; and assuming any experimentation tool measures the same things — client-side and warehouse-native approaches differ in rigor and in which metrics they can reach. People also treat experimentation platforms as interchangeable with the statistics behind them, when the tool is the plumbing and the discipline of a valid A/B test still matters. The discipline is to describe Eppo factually — its warehouse-native approach, its feature-flagging, and its current place inside Datadog — and to verify fast-moving product and ownership details against the vendor's own up-to-date sources.

Eppo is the brand name of the experimentation platform founded in 2020 and acquired by Datadog in 2025.

Etymology: source.

Search interest for this term over the last five years:

View interest-over-time on Google Trends →

Areas of marketing where eppo is a core concern: