75

After being inspired to gather some stats for Salesforce.SE, we were discussing it internally and we're wondering if other sites would want to see something similar. If so, then we would look to automate this and post it to each meta sites.

As far as stats, we'd look to include the following:

  • Count of New Users - Year over Year and a breakdown of new users by month for the reporting year
  • Total Number of Question asked
  • How many of the question received answers?
  • Average score of the answers
  • Top 10 Questions / Answers by Score for the year
  • Top Badges Awarded and How many users received each badge
  • Total Edits for the year
  • Total Flags for the year

Much of this information can be gathered as-needed from the Data Explorer or public data dumps, but if enough people think a comprehensive, site-specific retrospective post would be helpful (or even just entertaining), then it might be worth implementing anyway. So, what do you think of the idea, and is there anything else you'd like to see as a stat in an annual review?

  • Maybe ways that the site has improved since last year? – Doorknob Dec 28 '15 at 20:56
  • @Doorknob冰 It has to be quantitative somehow, so please specify what you mean. :) – Taryn Dec 28 '15 at 20:57
  • 5
    I'm a sucker for collective data like this! – Ash Dec 28 '15 at 20:57
  • 4
    Assuming that you were doing it anyway, and it's just a matter of changing the ID on the query to run it for a bunch of other sites, and that it won't take long to prepare, that sounds really cool! – corsiKa Dec 28 '15 at 21:01
  • 6
    On P.SE I've put together a meta post of interesting queries for people to peruse at their behest. Feel free to repoint them at your sites and fork as you please meta.programmers.stackexchange.com/q/7700/35276 - the "Site Participation" answer would likely have some of the types of queries you'd be interested in – Jimmy Hoffa Dec 29 '15 at 0:13
  • 3
    May I draw your attention to work already done - meta.stackexchange.com/a/261608/204631 ? – Deer Hunter Dec 29 '15 at 8:05
  • May I suggest these meta posts be tagged with something unique for the sake of filtering? I'm putting together a few statistics that can be queried in the public data. – Jon Ericson Jan 6 '16 at 19:53
  • @JonEricson New tag possibly? Suggestions for a name because naming is hard? – Taryn Jan 6 '16 at 22:48
  • is there anything else YessssI wish i can get a so t-shirt ;'( – Charuka Jan 4 '17 at 16:10
29

I think the idea is great! As an active member (and moderator, in some cases) of young beta sites that are constantly growing, I think this would be a great way for the community to sit back and take a look at how the past year has gone and how the site has evolved, and then apply this in the future. There are always things we can improve, and perhaps these stats can help us highlight some of them.


Tag stats

One stat I'd really like to see in this has to do with tag usage. It's easy to glance at the tags page and see how many questions have been asked in a certain tag over the span of the site (and the past 30 days), but this doesn't give a lot of details. So I'd like to see a few tag stats:

  • The 10 or so most used tags over the year (as well as during each month)
  • How many questions have been asked in each of the current top 10 or so most used tags over the year (again, a month-by-month breakdown)

These can help some sites figure out which topics are being asked about more frequently, and which aren't. Over time, sites evolve. This might give a better picture of that evolution.

Monica Cellio suggested a list of new tags created over the past year. This would also be another helpful measure of site evolution.

Meta stats

It would also be cool to see some of the same stats as those for the main site for the per-site meta (tag stats or no tag stats!). I don't think this is super-important, but it might be interesting.


One thing I wouldn't want to see, though, would be a measure of site health, like Area 51 has. Those stats are often terrible indicators of how a site is doing, and a lot of folks (myself included) would rather not have them.

  • 1
    Also, a list of new tags created this year. – Monica Cellio Dec 28 '15 at 21:13
  • @MonicaCellio That would be nice. I added that in. – HDE 226868 Dec 28 '15 at 21:15
  • I'd love to have this for Coffee. Smaller sites that are doing ok, but growing very slowly (as most beta sites are) could do with a little chance for a look at how far they've come! – fredley Dec 28 '15 at 21:15
18

Hotspots. I want hotspots. Places where the community fares worse. Tags which questions normally end ignored (views?), downvoted, closed, deleted, unanswered, etc. top ~20. How many answers where deleted because the question was closed and deleted (abandonedclosed), with breakdowns by tag, reputation, account age. All of this relative and absolute numbers, stats guys know why.

  • 7
    "Here's where y'all absolutely sucked. Happy new year!" ;) – Adam Lear Dec 29 '15 at 23:03
  • 2
    @AdamLear "Here's where y'all absolutely sucked, we expect y'all to improve on those areas. Happy new year!" FTFY – Braiam Dec 29 '15 at 23:20
  • 1
    This is much more useful than the usual "hooray, everything is great, there are no glaring quality issues or anything". (see SO's 10M help vamps) – bjb568 Dec 30 '15 at 0:19
  • @Braiam True. The point of annual recaps, though, is usually to celebrate something. I think including some of these would be interesting/useful (esp. the "answers lost because the question was removed"), but from a position that accounts for the inevitable feels. – Adam Lear Dec 30 '15 at 19:12
16

I'd love to see systematic data like this for my sites. Whether we're talking about a small beta still finding its place, a mature beta thinking about graduation, or an established site, it's good to be able to get an overview of activity from time to time.

When The Workplace graduated we had a performance review -- CM-assisted, as some of this data couldn't be pulled out of SEDE. Here's what we got:

The Workplace's first year, by the numbers:

  • Questions: 4,435
  • Answers: 11,451
  • Edits (to body): 9,526
  • Comments: 47,750, of which 11,520 were deleted
  • Upvotes: 106,677
  • Downvotes: 16,968
  • Total reputation gained: 1,310,840
  • Moderators promoted to management: 1

Ok, that last stat was kind of localized. But the others would be interesting on any of the sites where I participate.

We didn't collect user-related data, and that's a shame. You mention a count of new users; I'd also be interested in some measure of "users who stuck around" versus ones who posted once and never returned. I don't know whether this should be by post count ("new users total", "new users with 5+ posts", etc) or rep-based ("new users total", "new users with 200+ rep", etc).

  • Ooh, yeah, this would be nice. Have other graduated sites had these reviews when they graduated? – HDE 226868 Dec 28 '15 at 21:13
  • We specifically asked for ours; we had some recurring meta themes and thought this would be a good way to try to feed some data into those discussions. Plus, just curiosity. :-) But it was moderator-initiated, not SE-initiated. – Monica Cellio Dec 28 '15 at 21:15
11

I'd also like to see information about closed questions.

E.g.

  • How many questions have been closed this year
  • What percentage of the total questions asked this year have been closed?
  • Which close reasons are used the most?
  • (Most importantly) How many questions get edited and reopened? What percentage?

These stats would be helpful for determining what a sites main problems are, and how they can prevent these questions from being asked in the first place.


On a side note it would also be awesome to know the average time it takes for posts to go through the review queue.

  • 2
    All of this data is also available through the 10k tools though... – Zizouz212 Dec 29 '15 at 0:20
  • 2
    @Zizouz212 Sure, but not to the folks who don't have that much rep (or 2k on beta sites, I think). It would be nice for the data angussidney mentions to be in that summary post, regardless. – HDE 226868 Dec 29 '15 at 0:52
7

I could see this being useful on most sites, but oddly I think this may have some scale issues on larger sites like Stack Overflow.

  • Count of New Users - Year over Year and a breakdown of new users by month for the reporting year

I would imagine that on the larger sites we see an awful lot more users that are "just passing through". Users who create an account and never use it or post a single question/answer and never return.

  • Average score of the answers

On larger sites I would guess that this would fall very close to zero, just due to the sheer volume of answers.

  • Top Badges Awarded and How many users received each badge

This one could be interesting if tweaked to ignore the entry level bronze badges like Informed.



Things that I would like to see added:

  • How many close votes were cast, how many aged away?
    • some breakdown on close votes in the review queues vs not would be interesting
  • How many user suspensions?
    • some breakdown by reason and length could be interesting here as well
  • How many review audits, how many passed/failed?
  • Of the edits suggested, how many were approved/improved/rejected?
  • View to vote ratio?
    • How many users looked at it vs how many thought it was interesting enough to act on.

I'm just thinking that doing a "year in review" could be fun and interesting, but it would be better if we could dial it in a bit in order to get more actionable data. As in areas that need improvement.

  • What's the benefit of the user suspension stat? I can see interest in the others but suspensions, not seeing it. – Taryn Dec 29 '15 at 16:44
  • @bluefeet Just a curiosity really, sort of a "how's our moderation team doing, what sort of things are they dealing with?" – apaul Dec 29 '15 at 16:49
  • @bluefeet there may be a better way to address that area, but that was the only thing I could think of that would produce hard numbers. – apaul Dec 29 '15 at 16:52
  • 3
    Flags handled by mods VS by non-mods would probably hit that a bit harder, @apaul34208. – Shog9 Dec 29 '15 at 23:17
5
+25

I would honestly like to see the stats of this year versus the stats of the previous year. Passively, we are trying to make the site of better quality by solving meta issues and creating more ways to improve the site in general (better posts, chat etiquette, new moderators, etc.). The goal at the end is clearly to be better as a whole, as a community and to improve upon our mistakes.

This would be possible with a breakdown of stats from this year and the previous year and comparing them to see how we did. The statistics should at least contain:

  • Number of closed questions

The less the better since it means that users are getting the hang of the scope and rules within the site.

  • Number of new users

More users means more traffic and that's always good. This means that the posts are reaching Google and any advertisement ideas (social media, SE ads) are currently working.

  • Flags approved, declined, disputed

More flags approved is bittersweet. At least you know that the users are figuring out what's allowed and what's not but the OP's of the flagged posts are probably less informed on the situation. Less declined flags are better since that means the appropriate usage of flags has risen. Less disputed flags mean that issues in meta are being cleared and important issues are being solved.

  • Basic Area 51 stats (top users, questions per day, answer ratio, questions answered, visits per day)

The amount of avid users is quite important since they are the current role models of the sites (excluding moderators who might also be an avid user). The more there are, the more users there are that actually understand the site and can help maintain the site. More questions per day is technically better (if they are well-received) since there is more activity on the site. A higher answer-to-question ration is clearly better since it can help future readers with multiple solutions to choose from. Of course, a higher amount of visits per day means more activity.

The statistics should also include those mentioned in other answers. Answer to be steadily expanded...

3

I think we should also include a whole network-wide set of stats. I want to see how the network has evolved in a year, and how it's growing. Featuring some sites from certain statistics could draw them in and expose them to other parts of the network. Statistics such as:

  • Which site(s) had the most HNQs?
  • Which site(s) had the most positively received questions?
  • Which site(s) have the most active users (possible non-SO)?

Featuring a site on this basis may be able to interest others into participating in them. Some other statistics I suggest:

  • Which site(s) had the most community engagement and activity (non-SO)?
  • Which site(s) had the most answered questions?
  • Which site(s) graduated in the current year?

Etc..

2

It would be interesting to see such a summary post.

In addition to the suggested data, I'd like to see statistics regarding the number of deleted questions :

  • How many questions were deleted by the OP?
  • How many questions were auto-deleted (due to low score/low activity)?
  • How many questions were deleted by moderators?
  • How many questions were deleted by 10k users?

I'm mostly interested in the last two items, since the first two items usually correspond with bad questions having no answers at all or no decent answers.

It would be interesting to see how these numbers have changed in the past several years.

It would also be interesting to see these stats broken down by some of the leading tags.

2

Gathering stats without context is just busy work. Context is king.

One form of context could come from SE's (and specifically SO's) new features introduced this year - the developer story, and more importantly, the documentation (beta). Is there a correlation between documentation contributions and top-scoring users in those respective tags? Is there a significant decrease in basic questions? In copied content? In a similar fashion, internal events also produce a lot of context. Did Winterbash significantly change users' habits? E.g., did offering the Lifesaver "hat" significantly reduce the number of tumbleweed badges awarded?

Another form of content could be tying in to global trends and events. E.g., back in November Google announced it's joining the C# foundation. Did this cause any shift in SO's C# questions?

You must log in to answer this question.

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