A/B testing pages
Available on the Plus plan. Compare plans
Creating an experiment
Open any published page and click Create experiment. ConvertCanvas duplicates the page into a variant (B) — your original stays as the control (A). Edit the variant however you like: a different headline, a shorter page, an added guarantee box, a moved buy box. Experiments work best when you change one thing at a time.
How visitor assignment works
Once the experiment is live, incoming visitors are split between A and B automatically. Assignment is:
- Random — each new visitor has an equal chance of seeing either version.
- Sticky — a returning visitor always sees the same version they saw first, so the experience stays consistent and results stay clean.
- Server-friendly — both versions are compiled, native theme pages; there's no runtime redirect flicker.
Reading results
The experiment view in your dashboard shows, per variant:
- Views — how many visitors saw each version.
- Conversions — how many completed the page's goal action (e.g. add to cart).
- Conversion rate & lift — the variant's performance relative to the control.
Let the experiment run until each variant has a meaningful number of conversions before calling it — ending a test after a handful of views is the most common way to fool yourself.
Ending an experiment
When you have a result, click Promote winner to make the winning version the live page, or End experiment to keep the control. The losing variant is archived — nothing is deleted, so you can revisit it later.