Champion/challenger tests
Champion/challenger tests pit a proven variant against new variants with a smaller share; a better challenger can become the new champion. They are included on the Enterprise plan.
Last reviewed: 2026-07-22
What champion/challenger tests are
Champion/challenger tests are a special form of controlled comparison. The currently best variant, the champion, handles the majority of calls. One or more new variants, the challengers, receive a smaller share and are measured against the champion. If a challenger proves better, it can become the new champion.
Champion/challenger tests are included on the Enterprise plan. They build on the principle of A/B tests but split the calls deliberately unevenly.
How champion and challenger work together
The approach combines two goals:
- The champion continues to secure reliable results because it handles the larger share of calls.
- The challenger is trialed without placing a heavy burden on the overall performance of the campaign.
The evaluation uses the same metrics as the other analyses, such as the call outcomes and the booking rate.
When the approach is suitable
Champion/challenger tests are suitable when you already have a well-performing variant and want to trial new ideas without endangering current performance. For a first comparison of two equally ranked variants, a classic A/B test is usually the better fit.
Hinweis: Champion/challenger tests are available on the Enterprise plan. A change of champion should only take place once the challenger performs better across enough calls.
Common questions
When should a challenger become the new champion?
Only once the challenger reliably performs better than the champion across enough calls. A lead across a few calls can arise by chance and does not yet justify a switch. Judge the result using the same metrics as in the other analyses.
When is an A/B test the better choice?
For a first comparison of two equally ranked variants, a classic A/B test is usually the better fit. Champion/challenger tests are suitable once you already have a proven variant and want to trial new ideas without endangering current performance.