Documentation / Understand results
Two overlapping experiments
Inspect four combinations of a pair, or register a joint plan before launching both tests.
Step by step
- Open Analytics → Joint analysis and launch. An exploratory comparison of existing tests is diagnostic; it does not retrospectively register a confirmatory experiment.
- For a registered pair prepare compatible drafts on one site with the same goal identity/version, suitable two-variant allocation and independent salts. Review every eligibility blocker shown in the form.
- Set the joint enrollment end and conversion window, then register and launch together. Use the joint controls to stop or finalize the pair.
A storefront changes both the heading and the CTA. The four combinations distinguish the two changes and their interaction; the best heading alone may not form the best combination.
How to verify
Registered results require adequate data in every cell and completed observation/delivery windows. Finalization can wait until observation end plus 14 days and 5 minutes. Stopping both tests early yields no confirmatory winner.