Great talk. I've been struggling with understanding the true meaning of a cohort analysis, and the explanation over multiple types of charting and your custom tool was exactly what I needed for clarification
Don't the time axis for both vertical and horizontal axis need to be the same? Ie. For "Week 5", those are the users that joined your product in week 5, so you have 5 weeks of data on them... which means you can't possibly have data on their retention going out 3 months.... or am I missing something?
Yeah that graph is especially confusing! This is part of why I built my own tool to visualize cohort data that I demo'ed later in the video: bslatkin.github.io/cohorts/
The very first triangle graph example at 06:30, the explanation is far from clear. Each row represents people starting in a different WEEK. Columns show MONTHLY retention rate. While this is fine data to plot, the lack of 'month 6' data for 'week 2' cohort makes no sense. Approx 4 cohorts should share the number of monthly data-points, since there are 4 weeks in a month.
![[500Distro] Retention and Cohort Analysis with Bei Lu](/img/1.gif)
the audience should've clapped harder, thank you for your generous sharing of information.
Great talk. I've been struggling with understanding the true meaning of a cohort analysis, and the explanation over multiple types of charting and your custom tool was exactly what I needed for clarification
Very nice tool, Bret. You've just solved some headaches for me trying to do this in Google docs.
+Brett Slatkin great resource
Brett Slatkin This is awesome, Thanks! :D
Fantastic work mate, very, very cool tool. I am going to try to extract and run this tool over my meetup.com signups / presentations..
Don't the time axis for both vertical and horizontal axis need to be the same? Ie. For "Week 5", those are the users that joined your product in week 5, so you have 5 weeks of data on them... which means you can't possibly have data on their retention going out 3 months.... or am I missing something?
Yeah that graph is especially confusing! This is part of why I built my own tool to visualize cohort data that I demo'ed later in the video: bslatkin.github.io/cohorts/
The very first triangle graph example at 06:30, the explanation is far from clear. Each row represents people starting in a different WEEK. Columns show MONTHLY retention rate. While this is fine data to plot, the lack of 'month 6' data for 'week 2' cohort makes no sense. Approx 4 cohorts should share the number of monthly data-points, since there are 4 weeks in a month.
What is wrong with video? all is good but this flickering is killing me
We use this tool for cohort analysis - cohort.ly