When a deeplink or website has an A/B test running, its Analytics page gains an A/B Test Results card and per-variation filtering across the whole dashboard. This guide explains every number you will see and how to pick your winner.
The A/B Test Results Card
The card appears near the top of the Analytics page for any link or website with a test configured. For each variation it shows:
Views: how many visitors were assigned to that variation in the selected time range
Actual split versus target split: for example "48.7% actual / 50% target"
For websites: clicks and CTR (click-through rate) per variation, with the winner highlighted
The actual split naturally drifts a little from your target percentages, especially at low traffic volumes. Each visitor is permanently assigned to one variation, so the numbers converge on your targets as more unique visitors arrive.
Filtering the Whole Dashboard by Variation
Use the View stats for chips on the card (All traffic, Variant A, Variant B, and so on) to scope the entire Analytics page to a single variation:
The performance chart shows only that variation's views over time
Devices, platforms, countries, and referrers show only that variation's visitors
For websites, Click Insights shows only that variation's links with their own titles and counts
A notice bar reminds you a filter is active, with a one-click return to all traffic.
Which Number Picks the Winner?
For websites, use CTR per variation. Views mirror your split percentages by design, so views alone cannot crown a winner. Clicks per view is the number that tells you which design converts.
For deeplinks, Bouncy measures traffic delivered to each destination. What happens after the redirect lives on the destination site, so pair your test with the destination's own analytics (or your GA4 and Meta Pixel settings) to judge downstream results.
Things Worth Knowing
Visitors are consistent. The same visitor always counts toward the same variation. Refreshing a page never flips a real visitor between designs.
Preview visits do not count. Opening your link with
?abv=and a variation letter serves that variation without recording anything.Geo-filtered visitors are not part of the test. Anyone redirected by a geo or language rule is excluded from variation counts.
Combined Click Insights merge shared links. If your variations contain the same link (variations start as copies), the All traffic view combines its clicks into one row. Use the per-variation filter for the accurate per-design breakdown.
Results start when the test starts. Variations only accumulate data while the test is running.
Ending a Test
When you have a winner, edit the link or website, keep the winning version as your main design or destination, and switch the test off. Your historical per-variation data stays in Analytics for the time ranges where the test was live.