Tailor AI · Help · How-to
Set or change the control variant
From the Tailor AI team · Reviewed
Answer
After a test ends, you can promote a winner or use it as the new baseline for a fresh test. Avoid changing the control mid-test because it makes results harder to interpret.
When an experiment finishes, the cleanest move is to promote the winning variant — that becomes the new baseline (effectively the new control) for the next test you run. Changing the control mid-test breaks measurement: the prior data is no longer comparable to what you record after the swap. If you genuinely need to redefine the baseline before a test ends, treat it as ending the current test and starting a new one.
Steps
- If your lift is only on CTR, hold off, validate downstream impact first.
- After promotion, monitor for regression (novelty decay) for a few days in the Tailor dashboard at app.tailorhq.ai.
Caveats
If the win was caused by a temporary traffic mix shift, promoting it can ‘bake in’ a false positive.
Related questions
Trusted by performance marketers at leading B2C and PLG companies
See Tailor on your site
Preview how Tailor adapts your pages to traffic intent.
Looking for setup guides instead? Browse the documentation.