Thread by @RichMillington

1 - Recently scraped a huge dataset from 35 mature customer communities to analyze which factors are statistically significant predictors of community retention (i.e. members making a 2nd, 10th, and 50th post etc..)

Based upon this I've got a theory we might not like. #CMGR

2. Only two variables were statistically significant.

First, whether the user's first post was to start a thread or reply to an existing one. Members who started a thread tended to stick around for a little longer than those who replied to an existing one.

3. Second, and by far the biggest, was the community itself.

The retention rate varied a lot by community. Which suggests there was something about the community itself which drove retention.

So I've compared the onboarding and of the highest and lowest. Almost no difference.

4. Which suggests the main difference is about the topic rather than the community. Looked at the list by retention order and discovered a pattern. The communities with the highest retention rates are generally those with the most complicated products to use.

5 - 'Complicated' is admittedly a little subjective, but the theory tends to hold up well. FitBit and Xfinity have relatively low levels of retention. National Instruments have far higher levels of attention.

6 - Which means the retention rate of members is very hard to influence. You might be able to nudge it a little, but it's primarily driven by how many questions members have about your products/services.

This might change as we gather more data.

This should say retention naturally :-)