Experimentation

The flag you shipped is the experiment you measure.

There’s no separate experiment to set up. Attach metrics to a flag and Foredeck computes lift and confidence intervals from the exposures it already emits — with SRM checks and sequential testing when you need statistical depth.

FreeIncluded from the Free tier.

new-checkoutactive · 14 days
conversion_rate lift+4.2%
95% CI1.8% – 6.6%
SRM checkhealthy

One object. Ship and measure are the same action.

What you get

Why Foredeck for experiments.

Flag = experiment

Metrics live on the flag, so measurement is free the moment you ramp — no second resource to wire up or keep in sync.

Trustworthy readouts

Lift with confidence intervals, exposure counts, and a built-in sample-ratio-mismatch check so you can trust the number.

Depth when you need it

Sequential testing and variance reduction (CUPED) plug into the same exposure⋈metric join behind entitlement gates — never a paywall on experiments themselves.

Any metric type

Counters, conversion rates, gauges, and distributions — including latency percentiles and warehouse-native metrics — all usable as an experiment metric.

How it works

Three steps, one pipeline.

01Attach a metric

Pick one of your metrics as the primary readout for the flag.

02Ramp

Exposures stream in automatically as units are bucketed into variations.

03Read the lift

See effect size, CI, and diagnostics on the flag page — and promote or roll back.

Get in

Try Experiments free.