I'm new to conversion rate optimisation, I'm currently prioritising my test ideas based on the effort and impact.

Am I right in thinking that the MDE is the minimum uplift from baseline to be confident in the data?


  • Current monthly conversions = 23
  • Monthly traffic = 1000
  • Duration of A/B Test = 4 weeks
  • MDE = 54.78% (CXL tool)

This would mean if control got 16 conversions during the test, Variation A must get a minimum of 24 conversions to be statistically significant?

If this is correct, how would I measure this with current avg conversions being at 0 due to lack of tracking?

1 Answer 1


It doesn’t really matter if you’ve got no historic data because of issues beforehand as whilst the test is running you should be getting conversion data from both A (the control) and B (the variant).

As long as your tracking is set up during the testing then you’ll be able to see which variant is the favourite

  • Thanks for the comment, my issue is calculating the MDE pre-test with no historic data. Am I misunderstanding MDE?
    – GrapeSoda
    Jan 7, 2020 at 16:19

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.