Bayesian A/B testing
A way of reading test results that reports how likely each version is to be best, rather than a yes or no on significance.
A frequentist test asks whether the difference you see would be unlikely if the change did nothing, and answers with a p-value. A Bayesian test asks a more practical question: given the data so far, how likely is each version to be the best one?
The answer is a probability that is easy to act on, and it updates as visitors arrive. Elevate evaluates every experiment with a Bayesian model and reports a probability to win for each variation.
Put it to the test on your store
Test prices, themes, pages, shipping and checkout on Shopify, then ship the winning variant.
