Torque Lift

Sample readout

What a week with Torque Lift looks like

One readout, every week. It says what moved and why, reports every test, including the ones that didn't win or failed a check, and names the next one. Each test's metric and stopping rule are fixed before it starts.

The company and the numbers here are made up. Your readout uses your data.

Weekly readout · Pinecart (a made-up marketplace) · Week of Oct 12

Sample data · synthetic company

Weekly actives

21,480

+780 (+3.8%)

New

3,100

first week here

Resurrected

1,530

back after a gap

Churned

3,850

active last week, not this

Tests

  • A shorter signup formFinished · ship it

    Signup completion 31.2% → 33.9%, +2.7 points (95% CI +1.3 to +4.1). Pre-registered: +2.0 points minimum, 14 days. Sample ratio check passed.

  • Annual plan nudge at checkoutRunning · day 9 of 21

    No result yet, on purpose. It's a fixed-horizon test, so the readout reports it once, on Oct 30.

  • Onboarding checklistNot reported · failed a check

    Sample ratio mismatch: 52.8% / 47.2% instead of 50 / 50 (p < 0.001), so the result can't be trusted and isn't shown. Cause: a flag targeting rule. A rerun is pre-registered.

Next test

Show recent bookings on listing pages, aimed at second-week churn. Pre-registered: listing-to-checkout rate, +1.5 points minimum, about 7,400 visitors per arm, 3 weeks. The pull request is open and waiting for your review.