Ship what
actually works.
Know whether a test really won, how long to run the next one, and whether a change is safe to roll out. Significance, confidence intervals, chance to beat control, traffic-split checks and feature flag hygiene, in one free tool that never leaves your browser.
No more false wins.
Most A/B tests are stopped too early, read with the wrong math or shipped with a broken traffic split. ExperimentLab catches all three before you roll anything out.
Every launch, checked six ways.
Results entered, traffic split healthy, sample size reached, a full week of data, a significant win and a healthy guardrail metric. Only then does a test say "Ship B".
Results you can trust
Two-proportion z-test with confidence intervals on the lift, plus the Bayesian chance that B beats A, side by side.
Plan before you launch
Sample size per variant and run time from your baseline rate, minimum effect, power and daily traffic.
Flag hygiene
Track feature flags with owners and rollout. Stale, ownerless and fully-shipped flags are flagged for cleanup.
Your experiment workspace.
Sample data is loaded so you can explore. Replace it with your own numbers; everything saves automatically in this browser only.
Conversion test
Average value test (revenue per visitor, order value, time on task)
| Flag key | Owner | Rollout | Last changed | Health |
|---|
Rules: fully on or fully off for 30+ days means remove the flag; a partial rollout untouched for 14+ days is stalled; every flag needs an owner.