You can extend a test that seems close to resulting in a statistically significant
result, but hasn’t yet in its allotted time period.
You can extend an A/B test by changing its end date to a future date and time. If you
extend a test for which the end date has already arrived, it can immediately start taking on new
participants. If the end date has not already arrived, it continues to take on new participants,
but for a longer time period. If there are active customer sessions at the time a test is
extended (and the original end date has arrived), they are not evaluated for participation. Only
new sessions are evaluated.
You can extend a test on a Staging instance, but the effect only
occurs if you replicate the test to Production. For a more immediate effect, extend an A/B test
on the Production instance. However, if you extend the test on Production, also extend it on
Staging so that a subsequent replication doesn't override it.
Note: A/B test
statistics are not adjusted or changed if the start or end dates of a test changes. You can
schedule an A/B test to last up to 90 days.