Reading Results
What the numbers on a rule's analytics card mean
Where to Look
Open the rule you're testing from your dashboard and check its analytics card. If it has a split running, you'll see two different things: a raw impressions chip, and the experiment results themselves.
Two Different Numbers
Impressions (all traffic)
A quick pulse check — how many times Variant A and Variant B were shown, counting bots along with people. This is a raw count, not the comparison.
Experiment results
The real comparison: human visitors only, one card per rule (a redirect, link, content rule, bar, or popup can carry several geo rules, each with its own split).
Reading the Table
Each experiment card has a row for Variant A and a row for Variant B, with six columns:
Exposed
How many human visitors saw that variant.
Bounce
The share of exposed visitors GeoSwap never saw come back for another page view.
Pages/session
The average number of pages an exposed visitor viewed.
Avg time
The average time visitors spent on the page after being exposed to that variant.
Bounce vs A
How much higher or lower Variant B's bounce rate is than Variant A's, as a percentage of Variant A's own rate — not a percentage-point difference. Example: A at 50% and B at 40% shows −20%. Blank on Variant A's own row.
The Label
Below the table, GeoSwap shows one of four labels, based on comparing how many visitors bounced on each variant:
Not enough data yet
Fewer than 100 exposed visitors on one or both variants. GeoSwap won't call a winner on a small sample.
B keeps more visitors (≈95% confidence)
Variant B has a meaningfully lower bounce rate than Variant A.
B loses more visitors (≈95% confidence)
Variant B has a meaningfully higher bounce rate than Variant A.
No clear difference yet
Both variants have at least 100 exposures, but the difference in bounce rate isn't large enough to call.
The 100-exposure minimum applies to both variants before GeoSwap will call a winner or a loser — below that, you'll always see "Not enough data yet," however lopsided the numbers look.
Notes You Might See
Rules saved on <date>
Appears when you last saved this rule during the date range you're viewing. Results from before that date may mix the old and new variant, since the rule changed partway through.
Showing <start> to <end>
The date range the results cover. If it reads "(your plan's analytics window)," the range you asked for was longer than your plan allows, so GeoSwap shortened it.
Available on Every Plan
Experiment results aren't locked behind a paid plan — Free, Pro, and Business all see the full table and label. What changes by plan is how far back the date range can go, same as the rest of your analytics:
Free
Pro
Business
7 days
90 days
365 days
Things to Know
The two numbers won't match
"Impressions (all traffic)" counts bots along with people; the experiment results table is human visitors only. Don't expect them to add up to the same total.
Redirects and links need the destination to run GeoSwap
A split that sends visitors to a page without the GeoSwap script still counts an impression, but won't produce an Exposed count, bounce, pages, or time data for that variant. A rule split between a scripted and an unscripted destination can never reach a confidence label. See Tracking Across Sites.
It measures behavior, not conversions
GeoSwap tracks whether visitors bounced, how many pages they viewed, and how long they stayed — not purchases or sign-ups. Tag each variant with a different UTM or URL and check the result in your own analytics tool if you need to know conversions.
Give it time
With a 100-exposure floor on both variants, a low-traffic rule can take a while to earn a confident label. Let it run rather than judging it early.
Next up
See your site's overall traffic and engagement.