# ask-questions


02/08/2024, 2:02 AM
Another one - the situation isn't ideal, but not sure how best to handle within GB. we want to conduct an experiment of widget A, vs Widget B, both are CPA monetization. the experiment is a simple feature flag, 50% of users get A, 50% get B. The issue is that in terms of the feedback from both widgets we get a daily report of cumulative revenue generated; it is not tied to which user / session / cookie inspired the action, or how much each action generated, just a single figure. For the purpose of our experiment it doesn't really matter, because if Widget B makes more money than Widget A, then the experiment is a success (whilst it would be nice to have granular detail on conversion rate, avg revenue per user etc. it's not as important as running the experiment) - is it feasible to run this kind of experiment through GB? my understanding is we need to attribute on the user-level which we can't at this time (and will require convincing these companies to yield this detail so will be some time - too much time to wait before running such an experiment) but would still like to run this through GB, including experiment result analysis.


02/09/2024, 8:20 PM
You could do it as a switch back test
It will be a bit hard as you’ve described to figure out attribution