I'm following the Experiments documentation.

Do we have control over the percentage of which experiement variation displays? For example, if I have two variations, are they shown 50% each? Can I reduce the percentages?

Edit: for background, I'm researching how experiments can help with A/B type testing scenarios.

For example, we may want to test if certain content or functionality performs better, without impacting all users (e.g. test for only 5% of visitors).

This is not necessarily a SmartTarget example, but for 5% of visitors in a given location, we could set the default language to match the location. We could then measure if these users change the location back to the default or other language.

1 Answer 1


Each variant gets exactly the same 'air time', as it is the fairest way to judge the conversion rate of each relative to the others.

You could potentially work around it in code, but I wouldn't recommend that - certainly not without knowing more about the reasons behind the request.

  • Make sense. This is on the client's project backlog so we'll see how it goes. Jul 9, 2015 at 11:13
  • Alvin, I can definitely see a use for uneven splits. Hopefully you'll share how this goes if you end up pursuing it. Jul 9, 2015 at 16:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.