Hey, Corey from Custora here. We're huge fans of cohort analysis in general - it provides all kinds of insights into how customer behavior is changing. It's a powerful way of viewing historical data.
However, we need to be careful when using cohort analysis as a technique to predict lifetime value. There are many situations where cohort-based CLV predictions miss the mark by quite a bit.
I respect that you guys are not just trying to hype your tools but giving in these blog posts actually a very balanced portrait of them. Would love to see some postings on how you cope with the aspect that the data is somewhat always historic or how to overcome that problem.
