Leave a comment
Get the GH Bookmarklet

Ask GH

I've been trying to analyze cohorts for my startup that organises recruitment events online where companies meet students from universities. We have around 60 such events around different themes each year. These events last around a week. Where would you start in analyzing cohorts for such a business and moreover seasonal and event-based businesses? Thanks!

  • AH

    Adam Haney

    over 3 years ago #

    What if instead of looking at cohorts based on start time you based cohorts relative the events themselves (time before/after event). That would allow you to ask questions like, "How do users who joined 5 days before an event was held behave?". Alternatively there may be some sort of normalizing function you could perform like, for users that started this day/week/month of the year what percentage of events that were available did they attend? If there were zero events, then this percentage would naturally tend to dampen the seasonal impact.