Test Designer

Computed

Tell us what you want to test and how, and we'll turn your numbers into a properly powered plan — with an honest fit gate that tells you if you're big enough to run a clean test.

1 · What are you testing?

2 · Test inputs

Baseline conversion rate

That's a 3.00% conversion rate.

GREEN — good to run
  • Each arm should see ~1595 conversions over the test — healthy volume for a clean read.
  • Estimated duration ~14 days — a reasonable window.

Your test plan

to detect a 10% lift at 95% conf / 80% power
Sample per arm
53,182
106,364 total across both arms
Estimated duration
14 days
~2.0 weeks at 8,000/day
Incremental conversions needed
319
Net-new conversions the test must produce
Baseline → treatment rate
3.0% → 3.3%
The shift you're trying to detect

What fits in your time budget?

In 4 weeks you'd gather about 112,000 per arm.

Smallest detectable lift
6.8%
The minimum effect you could catch in that window
Your target lift
10.0%
Within reach in your budget ✓

Export this plan

Copy this and paste it into a doc, or share it with your team or agency.

REALITY CHECK — TEST PLAN
=========================

WHAT YOU'RE TESTING
  Test unit:   A whole channel
  Test type:   Conversion lift (platform-managed)
  Channel:     Meta

DESIGN
  Baseline conversion rate:  3.00%
  Target lift to detect:     10% (relative)
  Confidence / power:        95% / 80%

THE PLAN
  Sample per arm:            53,182
  Total sample (both arms):  106,364
  Estimated duration:        14 days (~2.0 weeks)
  Incremental conversions needed: 319

FIT GATE
  GREEN — good to run
  - Each arm should see ~1595 conversions over the test — healthy volume for a clean read.
  - Estimated duration ~14 days — a reasonable window.

Note: figures are computed from the validated closed-form engine. This is a
design, not a result — run the test to measure your real incremental lift.

Setup guide

Step-by-step instructions to set this exact test up in Meta.

Click “Generate setup guide” for instructions tailored to your numbers.

Pre-launch checklist

Tick these off before you launch — they prevent a silently broken test that reads as “no effect.”

Work through every item above before launching.