A/B testing engine
Split traffic across UGC widget variants. Live KPI counters update every 60s. Declare a winner once you cross 1,000+ impressions per variant.
Pro + Enterprise. Free public significance calculator at /tools/ab-significance.
Testing
Product page CTA + social proof variant
Traffic split
50 / 50
Primary metric
Add to cart
Min impressions
1,000
What you get
- Welch z-test for binary outcomes (CR, AOV uplift), same engine that ships in /tools/ab-significance
- One running experiment per widget, enforced at the dashboard level so test scopes never collide
- Live polling: KPIs refresh every 60s for any running experiment, so you can watch a test in real time
- Control (A) is your current layout; add Variant B (layout, CTA copy, source filter, overlay style)
- "Declare winner" stops the experiment and promotes the chosen variant automatically
- Per-variant impression + click + conversion counters with revenue attribution
How it works
- Step 01
Pick the widget
Any widget (PDP, PLP, homepage, collection, lookbook) can host one running experiment at a time.
- Step 02
Define variants
Control (A) is your current layout. Add Variant B with a different layout, CTA text, source filter or overlay. Traffic splits automatically when you hit Start.
- Step 03
Measure + decide
Watch live impressions + clicks + conversions per variant. Once each variant has 1,000+ impressions, declare a winner, Idukki promotes it and stops the experiment.
What’s included
- Auto traffic splitting (50/50 by default, weighting configurable)
- Live impressions + clicks + conversions + revenue per variant
- Welch z-test significance calculation under the hood
- 60-second poll for running experiments
- One-click "Declare winner" + auto-promote flow
- Public significance calculator at /tools/ab-significance for sanity checks
- Slack + email + Klaviyo result notifications
See a/b testing engine live in 30 seconds.
Drop your email, we’ll spin up a sample workspace with this feature pre-loaded and send a 5-minute walkthrough video.
