outtest
Get started
Product analytics and experiments

Amplitude Experiment review

Amplitude Experiment brings feature flags, code-based experiments and a visual web editor into Amplitude's analytics platform, with sequential testing by default and an AI agent, still in development, that drafts web variants. Free and Plus allow only one active experiment of each type, so a real testing program ends up on custom-priced Growth or Enterprise.
What it is
Amplitude Experiment is Amplitude's A/B testing product, split into Feature Experimentation for code-based tests and Web Experimentation with a visual editor, both built on Amplitude analytics.
Best for
Product and growth teams already on Amplitude analytics who want experiments that use the same events, cohorts and metrics.
Price
Free plan, paid plans vary
Free plan or trial
Free plan
Company founded
2012
Founders
Spenser Skates, Curtis Liu and Jeffrey Wang (Y Combinator W12)
Public listing
Nasdaq: AMPL
Customers
5,200+, including 30 of the Fortune 100, by its own count
Statsig
Amplitude took on Statsig's brand, customers and platform on May 5, 2026
Experiment customers named on its site
Calm, Brainly, Super.com, Rebuy

What it is

Amplitude is a public product analytics company founded in 2012 by Spenser Skates, Curtis Liu and Jeffrey Wang, and it says it has more than 5,200 customers. Experiment comes in two parts. Feature Experimentation runs tests through feature flags, deployed client-side or server-side. Web Experimentation lets marketers and product teams edit text, swap images and restyle elements with a point-and-click editor, or test full-page changes across matching URLs. Web experiments use the Amplitude Browser SDK to send exposure and interaction data to Amplitude.

On May 5, 2026 Amplitude also took on Statsig's brand, customers and platform, and said it will build a more integrated roadmap for the two. If you're choosing between Amplitude Experiment and Statsig, ask how that roadmap affects each.

Stats and metrics

Sequential testing is the default. It analyzes results as data comes in instead of only at a fixed sample size, with a 95% confidence level and Bonferroni correction for multiple comparisons. You can switch to a T-test, a Bayesian analysis or Thompson sampling bandits, and CUPED is optional.

Metrics come from Amplitude events. Custom metrics support the sum or average of a property value, which is how you'd measure revenue per user. Aggregated metrics, precomputed in a data warehouse, can be imported too. Amplitude's own guidance warns against defaulting to revenue when a variant changes something several steps away from it.

Revenue

Amplitude has a Stripe source that receives revenue events, including charges, refunds, invoices, payment intents and checkout sessions, by webhook. It only records events after you set it up, with no backfill, and you need to put amplitude_id or user_id in Stripe metadata so payments match the right user. Once those events arrive you can build a revenue metric on them. Amplitude's integrations page also lists Shopify, RevenueCat, Adapty, Apphud and Nami.

AI

The Website Conversion Agent helps pick high-impact pages, set goals, generate test strategies and create variants for the strategy you choose. You start each experiment yourself from the Experiment view. Amplitude warns that the agent is under active development and strategy quality may vary.

Plan limits

Every plan includes experimentation, but active experiments are capped. Free and Plus allow 1 active feature experiment and 1 active web experiment. Growth and Enterprise allow 2 of each, with an add-on for more. Mutual exclusion groups, holdouts and bandits are add-ons on Growth and Enterprise, and the custom code editor for web experiments needs Growth or Enterprise.

Who it suits

Amplitude Experiment suits teams that already run analytics in Amplitude and want tests on the same events and cohorts. Calm and Brainly are among the customers on its Experiment page. Small teams can start free, but the one-test limit on Free and Plus means a steady testing program usually needs a Growth contract or the startup scholarship.

Pros and cons

What Amplitude Experiment does well

  • Experiments use the same events, cohorts and identity resolution as Amplitude analytics.
  • It covers both kinds of test, with Feature Experimentation for code-based tests on client or server and Web Experimentation for point-and-click edits.
  • Sequential testing by default, with T-test, Bayesian, bandits and CUPED available.
  • The Free plan includes 2M events a month, unlimited feature flags and one active experiment of each type.
  • The Website Conversion Agent picks pages, proposes strategies and creates variants for web experiments.
  • Startups with under $10M in funding and fewer than 20 employees can get a free year of Growth.

Where it falls short

  • Tight experiment caps. Free and Plus allow 1 active feature and 1 active web experiment, and Growth and Enterprise allow 2 of each without an add-on.
  • Growth, Enterprise and add-on prices aren't published. Add-ons are priced as a percentage of your platform plan.
  • The Stripe source only records events after you set up its webhook, and needs your user ID in Stripe metadata to attribute revenue correctly.
  • The Website Conversion Agent is under active development, and you still launch each experiment yourself.
  • Amplitude now also owns Statsig and says it's building an integrated roadmap for both, so expect the experimentation products to change.

Questions people ask

Is Amplitude Experiment free?+

Yes, with limits. The Free plan includes 2M events a month, unlimited feature flags, and 1 active feature experiment plus 1 active web experiment, with no card and no time limit.

How many experiments can I run on Amplitude?+

Free and Plus allow 1 active feature experiment and 1 active web experiment. Growth and Enterprise allow 2 of each, and an add-on lifts the limit.

What statistics does Amplitude Experiment use?+

Sequential testing is the default, with a 95% confidence level and Bonferroni correction. You can switch to a T-test, Bayesian analysis or Thompson sampling, and turn on CUPED.

Does Amplitude Experiment have a visual editor?+

Yes. Web Experimentation has a point-and-click editor for text, images and styles, plus full-page tests across matching URLs. The custom code editor needs Growth or Enterprise.

What happened between Amplitude and Statsig?+

On May 5, 2026 Amplitude announced a strategic partnership under which it took on Statsig's brand, customers and platform, and said it will keep developing Statsig while building a more integrated roadmap. Statsig's original team stayed at OpenAI, which had agreed to acquire Statsig in September 2025.

Sources

Checked on 28 September 2026 from Amplitude Experiment's own pages. Prices and features change; check Amplitude Experiment's site before you buy.

Let Outtest run your split tests

AI agents read your analytics and payments, find where you lose the most money, build the fix and test it. Every test is judged on revenue, not clicks. Plans from $29 a month.