Flo Docs
A/B Testing

Reading Results

Creator plan required

After you Start test, open View results → from the Lifecycle section in the Split test panel.

Metrics you will see

Results compare variants over the full test lifetime (from Start test through today):

  • Page views and unique visitors per variant
  • Conversions (purchases, signups, bookings — whatever your page tracks)
  • Conversion rate — conversions divided by unique visitors (shown as a percentage)
  • Revenue per visitor (RPV) — total revenue divided by unique visitors

Conversions are attributed to a variant when the visitor was assigned or viewed that variant within the last 30 days — the same attribution window used for checkout tagging.

Primary metric

Creator plan required

By default, Flo ranks variants by conversion rate. On Pro and above, you can switch the primary metric to revenue per visitor (RPV) from the results panel. The stats engine and sample-size guard use whichever metric you select.

Bayesian confidence

Flo runs a Bayesian analysis on your test data and shows:

StatusMeaning
Collecting dataNot enough visitors yet — keep the test running
InconclusiveEnough traffic, but no variant is clearly ahead
Leader emergingOne variant is ahead, but confidence is below 95%
SignificantA leader has ≥95% probability of being best

Each variant shows probability of being best and a credible interval for the primary metric. This is guidance — you still choose when to declare a winner.

Mixed currencies

If conversions in the same test use different currencies, RPV statistics are hidden until currencies are consistent.

Tie results to Analytics

Creator plan required

Revenue analytics and related dashboards on Creator+ help validate winners beyond raw conversion counts. See A/B test analytics.

When to declare a winner

SignalAction
Significant status or clear business winnerDeclare winner
Inconclusive after a long runEnd test without winner and try a bolder change
Major external event (sale, press)Pause test, resume later, or end

Flo may block promotion when sample size is too low. You can still declare anyway if you accept the risk.

Significance email (Pro+)

Pro plan required

When a running test first reaches Significant status, Flo emails you a summary of which variant is leading (Pro and above). This is a notification only — Flo does not auto-promote the winner. Open the builder, review results, and Declare winner when you are ready.

Creators on Creator tier still see Bayesian stats in the results panel; only the email alert requires Pro+.

Paused tests

Pause stops new assignments but keeps historical data. Resume continues the test with the same splits.

What's next?