Features are cheap now. Knowing which ones matter is not.
A/B testing belongs in the dev loop the way unit tests do. A unit test tells you the code does what you meant. An A/B test tells you the feature does what the customer needed.
With ABMeter, you run a real, correctly analyzed A/B test in an afternoon, free.
It is built for individuals and small teams without a data scientist. The terms say what an account includes.
Build
Connect an AI assistant
Add the ABMeter MCP server to Claude, Cursor or another MCP client. It sets up the experiment, connects your code and reads the results with you.
Install an SDK
Ruby, Python, Node.js, Go, browser JavaScript or React Native, with the REST API behind them. No AI assistant needed.
The quick start walks through both.
Your code reads parameter values, never variant names. A page asks for
button-color and gets a color. It does not know
whether an experiment is running or which variant the user is in. So you integrate once, and
every next experiment is a change in ABMeter, not in your code. That is what turns testing
into a habit instead of a project.
Want to look first? Explore the demo: sample experiments with live statistics.
Go deeper
You do not need statistics to run an experiment. If you want to check what "correctly analyzed" means, the method and the math behind it are public. Before you build an experiment, check that your traffic can answer the question.
Can you test this yet?
How many users a test needs and how many weeks that takes at your traffic, and what to do when the answer is too long.
How ABMeter evaluates experiments
The method: a two-sample z-test on per-user metrics, confidence intervals and p-values, and how to read them.
The Statistics of A/B Tests: The ABMeter Book (PDF)
A short book with the math behind the method, from estimators to sample size.
Blog
Why ABMeter works the way it does, one design decision per article.