9

When visiting analytics for beta sites, one can view different guidelines on how each proposal is doing based on the following "healthy" guidelines:

  • 15 questions asked per day
  • 90% of questions answered
  • A user core of 150 users at 200+ rep, 10 users at 2000+ rep, 5 users at 3000+
  • 2.5 answers ratio / per question
  • 1500 visits per day

How were these guidelines calculated?

  • 3
    Questions about Area 51 should be asked on its meta site. – kiamlaluno Apr 4 '14 at 4:31
  • Oops, sorry about that I wasn't even aware there was a meta for Area51. – Emilie Apr 4 '14 at 4:32
  • 1
    @kiamlaluno This is perfectly on-topic here. Area 51 doesn't have an immediately obvious meta site. – Tim Post Apr 4 '14 at 4:33
  • 1
    @TimPost It is called in a different way, but it has tags a meta site should have; in fact, there are area51-meta, bug, support, and feature-request. – kiamlaluno Apr 4 '14 at 4:37
  • @TimPost Still don't know about discuss.area51.stackexchange.com? I'm not surprised, it's one of our best-kept secrets. – michaelb958 Apr 4 '14 at 4:41
  • 1
    @michaelb958 I know about it, but it's not immediately obvious to users new to the network (and Area 51). It's sort of like any feature request that affects all sites - welcome on child meta sites or the central one. – Tim Post Apr 4 '14 at 5:01
  • 1
    I often have an impression, that questions about Area51 on Area51-meta are simply ignored, so it makes perfectly sense to ask here. You have an answer here after a few hours, in Area51-meta you'd have to wait at least a few days, if not a few weeks. – Danubian Sailor Apr 4 '14 at 7:33
5

In a mostly arbitrary manner. Those guidelines describe a site on a broad topic where there's likely to be plenty of participation. Cooking is a lot different from particle physics in this respect, especially early on.

It's easier to explain the motivation behind each one, with the understanding that we evaluate sites mostly on a case-by-case basis:

  • Sites need enough new questions every day to keep people interested, there has to be something on the site to do or people get bored and go elsewhere.

  • Our implied 'contract' with users is that if they give us a question, they'll receive a quality, peer-reviewed answer within a reasonable amount of time.

  • Sites are moderated by the community, in order for that to be possible, users must be actively unlocking privileges by earning reputation. Additionally, we hold democratic moderator elections once a site graduates, the user base must be able to support this. The rep 'spread' is also a good indicator of how well a site retains new users.

  • Back to our implied 'contract' - most questions do warrant several peer-reviewed answers, even when there can only really be one 'correct' answer.

  • Sites need plenty of traffic in order to grow, most of which is going to come from search

The metrics you see there are just a general suggestion of what one of our typical healthy sites is going to look like. Some sites have a higher number of unanswered questions, others don't usually get more than 2 answers per question on average.

With really niche topics, this can vary quite a bit. The important thing is simply don't panic if those numbers look scary, we'll let you know if we see a problem.

Making that health dashboard a bit smarter is something that's on the road map for the next generation Area 51. Yes, the milestones are an artifact of our experience launching sites, but they're far from a one-size-fits-all sort of proposition.

  • I think the most useful way to create guidelines for these statistics would be to use the data from already graduated sites. I'd also emphasize the range of acceptable values, a single target can be very misleading. – Mad Scientist Apr 4 '14 at 14:00
  • 1
    @MadScientist "Mostly arbitrary" is an flippant exaggeration. Those numbers were derived by observing what "worked" among the thousands of sites that were created when SE 1.0 was a subscription service. But they're broad guidelines and not meant to be some sort of “report card” filled with pass/fail grades. We've since learned that sites can scale by the popularity of the subject, and I would have gotten rid of that Area 51 report if how to actually measure "healthy" quantitatively wasn't so damned elusive. – Robert Cartaino Apr 4 '14 at 14:32
  • 1
    @RobertCartaino They are very often understood as a threshold the sites have to meet to graduate, I know they aren't, but there is at least one meta post with this assumption on nearly every beta site. That is why I would really emphasize a range of values instead of a single one to make clear that there are large differences between sites. Seeing the current value compared to all beta sites in some form would be far more useful IMO than the one-size-fits-all threshold we have currently, even though the metrics themselves are somewhat flawed. – Mad Scientist Apr 4 '14 at 14:38
  • @MadScientist What range would you suggest? That's rhetorical — Amateur Radio is growing heartily with 1.8 questions/day. We'd be in utter panic if Stack Overflow fell below 5,000. – Robert Cartaino Apr 4 '14 at 14:50
  • 1
    @RobertCartaino I'd use the historical data from the graduation day of all sites that went through Area 51 and graduated sucessfully. – Mad Scientist Apr 4 '14 at 15:07
  • @MadScientist Questions/day (for example) historically ranged between 3.5 q/day and 35 q/day at graduation. Not very helpful. – Robert Cartaino Apr 4 '14 at 18:44
  • 1
    @RobertCartaino But the hard threshold of 15 currently that simply does not make sense for certain sites is even less useful. Showing the range (with some extremes excluded, maybe) is not all that useful, but far more honest and less misleading. – Mad Scientist Apr 4 '14 at 18:49
  • +1 for the answer as well as comments, nice discussion! – Emilie Apr 5 '14 at 18:38

You must log in to answer this question.

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