01
Wizard of Oz
Fake the system end to end if the workflow needs validation.
20–100 users you can serve manually. A simple landing/UI.
Buyer psychology// FOUNDER EXPERIMENT PICKER
Pick the experiment that fits your traffic.
Most founders default to A/B tests because they’re loudest. At your scale they almost never finish. Tell us your stage, your goal, and your traffic — get the experiment shape that actually concludes.
Your situation
Traffic to the page or funnel you'd be testing. Use 0 if you don't have one yet.
Also worth considering
02
Smoke / fake-door landing
A fake-door landing for the next feature is cheap signal.
A budget for paid traffic, or one warm channel. ~200 visits is enough.
GTM engineHow this picks
Three inputs decide the shape of every founder experiment.
- Stage. Pre-launch problems are demand and willingness-to-pay problems. Post-PMF problems are conversion, activation, and retention. The right experiment shape changes accordingly.
- Goal. Validating demand or pricing wants money or click-through; lifting an existing metric wants statistical comparison. Different evidence, different instrument.
- Traffic.A/B tests, behavior tests, and cohort retention all need a sample size to read a delta. Below the floor, qualitative methods are not a downgrade — they’re the right tool.
If the recommendation is an A/B test, your next stop is the sample-size reality check. It tells you whether the math actually works at your numbers — most of the time it doesn’t, which is exactly why this picker exists.
What this tool is not: a replacement for a hypothesis. The picker tells you what kind of experiment to run; you still have to write down the bet, the kill threshold, and the success bar before you start. The CRO field guide is the long-form version of that discipline.
Run this experiment in Xi.
A picker tells you what shape to run. Xi runs it: lock the hypothesis and kill threshold up front, track the metric automatically, and let agents call the verdict so you don’t drift into endless “just one more week.”