Skip to main content
Experiments let you change your storefront on evidence rather than opinion. Instead of debating which headline, layout, or ranking algorithm works better, you serve different versions to real visitors and measure what happens. The module contains two experiment types. They share most of their configuration and reporting, but they answer different questions.

​
The Two Experiment Types

A/B Testing

Split traffic across a control group and up to three variants by fixed percentages, then measure which one performs best.Answers: which version wins?

Personalization

Give each variant its own audience, and serve every visitor the version that matches them.Answers: which version suits this visitor?
Both run through the same step-by-step configuration flow and share their reporting. Where they differ: An A/B test is a decision you make once and apply to everyone. Personalization is a decision you make repeatedly, per visitor.

​
What You Can Experiment On

Both types target the same four experience types, and one experiment can include more than one.

​
Use Cases

​
A/B Testing

Validate a conversion hypothesis. Your add to cart rate sits at 1.8% against an industry benchmark of 3.0%, and you suspect the CTA button. Test the current blue button against orange, with add to cart as your primary metric and revenue as a secondary metric to catch any downstream damage. Compare search ranking strategies. Send half your search traffic to a relevance-weighted algorithm and half to one weighted toward margin. Measure click-through and conversion before committing. Test merchandising rules safely. Try a new product ordering rule on 20% of traffic rather than the whole catalog, and watch conversion before rolling it out. Evaluate a layout change. Run a new product detail page against the current one for two weeks, or until 10,000 add to cart events arrive.

​
Personalization

Reward returning customers. Serve a loyalty-focused homepage to a returning customer segment while first-time visitors see acquisition messaging. Match the landing page to the campaign. Serve a page that continues the promise of a paid search ad to visitors arriving with those UTM parameters. Adapt to market. Show region-appropriate merchandising and promotions based on the visitor’s location. Tune for device. Give mobile visitors a shorter component layout while desktop visitors see the full version.

​
Who Uses Experiments

Marketing analysts target experiments to specific campaigns and segments, then track primary and secondary metrics to isolate marketing impact. Ecommerce managers allocate traffic across merchandising and search variants, then pause or clone experiments as priorities shift.

​
Key Capabilities

  • Run experiments per channel and language.
  • Target by audience segment, UTM parameter, location, or device.
  • Track one primary metric and multiple secondary metrics.
  • Stop on a date, or end a running experiment manually.
  • Pause, resume, clone, and reorder experiments as priorities change.
A visitor is assigned to one experiment at a time. When a visitor qualifies for several experiments, priority order decides which one applies.

​
What’s Next