A/B Testing Best Practices

Creator plan required

Test one big change at a time

Compare one hypothesis per test — new hero copy or block order or primary product — not all three. Mixed changes make results impossible to interpret.

Keep variants live long enough

Wait for enough page views per variant before Declare winner. A few dozen visits is rarely enough for a high-ticket offer.

Do not edit variants while live

While a test is Live (running), Flo locks variant blocks and page settings so results stay valid. Pause the test, make your edits, then Resume. You can still rename variant labels and adjust traffic split while live.

Align traffic with launches

Starting Start test the same day you announce a huge promo skews data. Either pause the test during spikes or note the date when reading results.

Use Creator limits intentionally

Creator allows 2 variants — perfect for A/B. Use Pro when you need three or four arms (for example headline × offer combinations spread across labeled variants).

Plan the winner workflow

Before Start test, know what Declare winner should do to your live funnel (pricing block, email capture placement, etc.). Promotion overwrites the live page snapshot.

Pair with analytics

Creator plan required

Check revenue analytics and A/B test analytics so wins show up in revenue, not just clicks.

Payments and tests

Checkout behavior follows the variant a visitor was assigned. Do not change product IDs or prices underneath a running test without documenting it.