Hi guys, we tried to understand you explanation of changes in 1.7 version with conversion delay https://github.com/growthbook/growthbook/pull/537 Can you share more details with example about fix of conversion delay for latest growthbook version? With example

It mostly affected experiments that were using activation metrics and/or custom denominators. For those experiments, we need to join multiple metrics together in a sequence (e.g. viewed an item page, then viewed checkout, and then purchased). Each of those metrics have their own conversion windows and we weren't properly combining them before. For example, if they each have a conversion window of 12 hours, we should be waiting 36 hours total, but we were actually waiting less. Now that's fixed.

@future-teacher-7046 thanks, for details Can you explain with my case, please? We have activation_event = main_pageview and goal = purchase. I have users that can purchase goods more than one item per day. Also I have default setting in 1.7 version - conversion_window = 72 hours. If user purchase item in 80 hours he didn't count as conversion to our database in previous version of GrowthBook. Did you fix it in 1.7 version that window conversion affects on time when I purchase item?

In the latest version, the user would have 72 hours to fire the activation event. Once that event is fired, they would have 72 hours to purchase.

@future-teacher-7046 is that real to do conversion window setting optional? Like switch button Cause for our product 72 hours too little

You can change the conversion windows by editing the metric behavior settings

@future-teacher-7046 Can we change the value of the conversion windows by 0 for excluding his impact on our experiments and include all players for the exp result?

@lemon-librarian-69651 thanks for help

@future-teacher-7046 we don’t use all of these settings(in the step “Behavior”) and search the way how to get around them. So, if you explain how to do it, it’s getting really helpful for us

You can set a really high number for Conversion Window (like

999

) if you want. We have a project on our roadmap to let you ignore conversion windows and just include everything that happens while an experiment is running. https://github.com/growthbook/growthbook/issues/721

@future-teacher-7046 tnx you so much! If I understand correctly, the conversion window is only one param, than we must change to include all players/metrics/etc in our dataset, yea?