Skip to main content

A/B test types

When you create a test, the first choice is the test type. Pick the one that matches what you want to compare.

Published pages (required)​

For every type that uses Atomicat pages (page variant and split URL):

  • Control and each variant must be published on the same site before you activate.
  • Those pages must stay published while the test is running. Unpublishing is blocked until you archive the test.
  • If you archive a test, unpublish a page, then try to activate again later, Atomicat checks publish state again and will ask you to republish first.
  • After you edit a variant page’s content, use Redeploy on the test so visitors see the latest version (page variant) or so routing stays correct (split URL).

Page variant (most common)​

Compare two or more published Atomicat pages on the same public URL. The visitor’s URL never changes — Atomicat serves a different page behind the scenes.

Use it when you want to test headlines, layout, offers, VSLs, or pricing on the same funnel step without changing the address bar.

You can add a third (or more) variant in the same test — weights must still total 100%.

Split URL​

Visitors hit one entry URL, then Atomicat redirects (302) them to each variant’s own published page URL.

Use it when you want each version on its real slug (for example /offer-a vs /offer-b), not rewritten under a single path.

Goals still work: the redirect includes experiment context (at_exp / at_var), so checkout clicks, button clicks, selectors, and URL-visit goals are measured on the destination Atomicat pages.

External redirect​

Split traffic from your Atomicat entry URL to external destinations (any URL outside Atomicat). Visitors are redirected immediately.

Use it when you only need to divide traffic and will measure winners in Hotmart, Stripe, Google Analytics, etc.

note

Atomicat cannot track clicks or conversions on external hosts. For this type the only goal is Traffic split only.

Choosing traffic weights​

  • A 50/50 split reaches a decision fastest for two variants.
  • Use uneven splits (for example 80/20) when you want to limit the risk of a new idea.
  • Weights must always add up to 100%.

Frequently asked questions​

Which type should I start with?

Page variant. Publish two pages, change one meaningful thing, and test both on the same URL.

When should I use Split URL?

When each variant should keep its own public slug and visitors should be redirected to that URL instead of staying on one entry path.

Can I unpublish a variant while the test runs?

No. Archive the test first, then you can unpublish. Activate will require the pages to be published again.

How much traffic do I need?

As a rule of thumb, wait for at least a few hundred visitors per variant before trusting a result. The Analytics decision badge stays in 'collecting data' until there is enough volume.

Can I test more than one change at a time?

You can, but you will not know which change caused the difference. Test one meaningful change per variant when possible.