Knowing which variant won is only half the story. The other half is understanding why.
Today we're launching Experiment Insights: a new feature that breaks down exactly what Squoosh synthetic shoppers experienced, what influenced their decisions, and how those experiences differed between your original and variant site versions.
More than a winner badge
Squoosh experiments already tell you whether site changes are likely to increase conversions. But a single metric never tells the full picture. Did shoppers on the variant struggle to find the product? Did the original's customer reviews tip the balance? Was there a trust gap?
Traditional A/B tests can't fully answer these questions. Real shoppers don't explain themselves; Synthetic shoppers do.
Experiment Insights answers those questions directly. After an experiment completes, Squoosh analyzes every shopper's full journey, including their decision, the friction they hit, and the product pages they browsed, surfacing what actually drove the outcome right on the results page.
Built around what actually happened, not just what shoppers said
Early versions of this feature produced summaries of shopper decision rationale, without grounding those explanations in the actual journey each synthetic shopper took. The result was surface-level findings that summarized sentiment without connecting it to concrete browsing behavior.
The breakthrough was pairing each shopper's stated reasoning with the full picture of what they actually did. The analysis now sees both together. For example, Experiment Insights can identify if displaying reviews on the product list page was effective by examining an increase of clicks to product details pages and references to reviews in decision rationales.
What you'll see
After a new experiment completes, an Experiment Insights section appears on the results page with three layers of insight:
- Behavioral drivers by variant. A focused analysis for each site version which contrasts how shoppers behaved differently.
- Site issues. Problems that appeared across both site versions are grouped by category: Trust & Security, Navigation, Product, UX & Design, and more. These are baseline improvements worth fixing regardless of which variant you ship.
- Expandable shopper quotes. Every finding is backed by direct quotes from your synthetic shoppers. Click the quote count to read exact reasoning in each shopper's own words, and for any quote, you can watch the full synthetic shopper session to understand the complete context behind it.
We're looking forward to getting your feedback!
Experiment Insights is available now. Questions? .