3

Comment here expressed concern on whether site is attractive for new users. I would like some statistics to learn more about that.

Per user ids it looks like site has got 5,000 new users this month, I would like to find more about their activity, how many of them are registered, how many asked, answered, commented, suggested an edit, voted etc.

For the purpose of comparison, I would like to also have similar stats on how it was half year ago, one, two and three years ago.


This is how I would want to measure trends in user "stickiness": by comparing how many of users registered in 2010, 2011, 2012, 2013, 2014 visited site for 5, 10, 30, 100 days. SEDE query for this would be a piece of cake if visited days were in schema... but these aren't.

  • You could probably get most of that data from Data.SE, minus the deleted posts of course. I have links to some queries checking user activity here if you want, and they could be easily modified to limit results to a specific date range, and to "new" users during that time frame. Some other possibly worthwhile answers to read through are Shog's answers here and here too. – Rachel Jul 9 '14 at 15:46
  • @Rachel I thought of Data.SE prior to asking. It won't tell votes from new users which I suppose could be most of their activity (for those with association bonuses) – gnat Jul 9 '14 at 15:47
  • Yeah that's true, Data.SE doesn't show which posts a user voted for. I figured that Questions/Answers and their score would probably be more relevant than votes though for showing how a new user uses the site. I think it's very skewed though since it doesn't show Deleted posts though, especially since low-scoring closed posts are auto-deleted. – Rachel Jul 9 '14 at 15:51
5

I'm going to work up a more complex query, but here's some basic data about retention that I'd like to look at more in order to scope said query. We get a lot of users unlikely to stick around, in particular below the 200 reputation level.

To put this in context:

  • There are 3,282 users between 150 and 200 rep. Of them, 1,180 have not been seen in 180 days or more (the point at which we consider them lost). Roughly 35%.
  • There are 4,434 users between 200 and 500 rep. Of them, only 1,308 have not been seen in 180 days or more. Roughly 29%
  • There are 1,291 users between 500 and 1k rep. However, only 295 of them are gone, or 22%.

The percentages get even smaller as you work up the scales. On this site, if you get past 500 rep, you're very likely to become a long time user - we've got relatively decent retention if you consider that probably half of the folks in the first bracket probably found the site by accident in search of something else.

I think what we want to look at here is folks in the middle bracket that have actually voted and posted at least once. I'm sure there is some meaningful data out of the first bracket, but its surrounded by a whole lot of random noise.

Can't promise I'll get to this in the next 7 days, but I am going to see what I can get in a form that can be posted. You've got me curious as well.

  • thanks Tim, this seems to be the kind of data I am looking for. Wonder if it would be possible to estimate trends, that is to compare current state against how things were half year ago, 1-2-3 years ago – gnat Jul 21 '14 at 1:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .