What if AB testing is impossible to setup? I wrote a blog to measure impact using backdoor adjustment, a type of causal analysis

Follow the full discussion on Reddit.
To ensure that every feature has a measurable impact on the broader platform my team will set up and run A/B testing on each new feature or product change, but what happens when a new feature needs to be released quickly and there is not enough time for a traditional testing approach? To make sure that these quick changes could still be measured I found a way to perform accurate pre-post analysis using a back-door adjustment of causal analysis. I wanted to share my findings with the community as it was able to help my team at DoorDash make quick bug fixes and still be able to measure the impact. Please check out the article to get the technical details and provide any feedback on my approach. https://doordash.engineering/2022/06/02/using-back-door-adjustment-causal-analysis-to-measure-pre-post-effects/