[akchin]

I found the article very insightful. I spent the weekend reviewing Dropbox's S1 and had the same issue - they are not being completely honest about their cohort retention numbers. They have created some new metrics/charts that show increasing value but doesn't show the real metrics - what is specific cohort churn etc. This article mentions that very clearly.

Don't get me wrong - Box's initial S1 also had some very abstract cohort metrics. They took some very early cohorts and extrapolated from that. I think they are more open now as there is more stability in those metrics and they are moving in the right direction.