Estimating total lift from a change: Is there a way natively get estimations of the total expected lift from some change. e.g. if we have some experiment and say the experiment is showing a 13% improvement, is there a way to say that ok with

n

eligible users and a current rev per user of

k

we’d expect a lift of

n*k*.13

This is something we could do backend pretty easily but i was wondering if theres any features built in to do it or if alternatively someone else has a clean implementation ?

Yeah, we've talked about ways to surface overall business impact on the experiment results. We already have the percent of traffic included and the allocation between variations. So we could say the overall impact would be

impact = (1/(percent of traffic)*(1/(variation allocation)*uplift*(number of users in variation)*(365/(length of experiment in days))

So if your experiment is running for 40% of traffic over 14 days, with a 50/50 split, and your variation had 1000 users and a 10% uplift, the overall impact would be

(1/0.4)*(1/0.5)*(0.1)*(1000)*(365/14) = $13000/year

So that makes sense but im not sure what the implementation of that looks like in growthbook

Nothing for you to do right now. I added a GitHub issue for us to add overall impact to the results page. https://github.com/growthbook/growthbook/issues/210 We should be able to get that done in the next week or two.

Awesome thank you !

Any idea if this could be a feature soon? Wondering if we should build the calculation on our side for our product team