Running A/B Experiments

Test two section versions against each other. Smart Pages splits traffic, tracks results, and declares a winner.

Setup

1. Click menu (three dots) on a section → "Run experiment"

2. Configure:

   • Control — Original section (pre-selected)

   • Challenger — Variant to test (AI-generated or existing section)

   • Traffic split — 10% to 90% slider

   • Scoring mode — CTA clicks, Conversion rate, or Revenue

3. Click "Start experiment"

Monitoring

The experiment panel shows for each version:

• Sessions, Conversions, CVR (conversion rate)

• Confidence level (0–100%), Lift (% improvement)

• Winner badge when confidence is high enough

Ending

• Promote winner — Replace losing version with the winner

• Stop experiment — End test, no changes made

Status	Meaning
Running	Active, collecting data
Concluded	Winner declared
Cancelled	Stopped manually

Tips

• Run for at least 7 days for reliable results

• Start with 50/50 split

• Test one thing at a time — headline OR image, not both

• Results under 90% confidence may be random variation