21

We're being asked to give feedback on what should be done with the Hot Network Questions list that's shown on the right-hand side of the screen. Ideas range from elaborate—relying on communities on each site to manage candidates for the list, & consuming (as far as I can guess) much development resource—to simple—e.g. just keeping it off the side-bar.

Taking it as read that something must be done, on the basis of what information can it be decided what the best option is?

  • Does SE measure click-throughs from HNQs, how many are "converted" to sign-up to the target site or post there, anything like that?

  • Are there survey data, say, on how many users care about having HNQs in the side-bar, or having them at all?

  • Will site Metas be canvassed for opinions from beyond Meta SE, or to gauge interest in taking on the jobs that might be passed to them?

I'm concerned that SE might be about to embark on a big-ish project to deliver an improvement that doesn't interest many users or deliver significant benefits, at the expense of work that does both.

Improve this question
10
  • 1
    SE appears to collect (sampled) statistics on clicks that go to particular site from HNQ: an example snippet of these stats has been posted a while ago in Tavern chat room
    – gnat
    Commented Oct 24, 2018 at 15:45
  • 4
    I'm curious about sites where the usefulness of the questions are highly dependent on knowing the relevant technologies which presumably aren't that popular within the wider community (like Sitecore and Drupal, perhaps). I imagine the vast majority of people don't find such questions in any way useful or interesting. No offence to such sites, it's just: how would it help me to know how to set up an error page in something I have never and probably will never use?
    – Bernhard Barker
    Commented Oct 24, 2018 at 19:09
  • 1
    Avoiding twitter seems to be the benchmark at this time.
    – user316129
    Commented Oct 24, 2018 at 19:50
  • @gnat: That's good to hear. Of course, clicks alone aren't terribly informative - e.g. I couldn't resist clicking on Is it possible to remove the smell from rice mistakenly stored in a detergent box?, but wasn't much edified, & I didn't go on to contribute anything to the site.
    – Scortchi - Reinstate Monica
    Commented Oct 25, 2018 at 9:08
  • @Dukeling: There are some good suggestions on how to better target what each user sees & how to give them a measure of control over it here. At the moment the HNQ sidebar is rather akin to a door-drop campaign.
    – Scortchi - Reinstate Monica
    Commented Oct 25, 2018 at 9:21
  • @DonThermidor_LobsterMobster: If we only wanted to avoid bad PR on Twitter we could simply get rid of the HNQ side-bar altogether. But the premise of Revisiting the “Hot Network Questions” feature is that it provides some returns to SE & its users: here I want to discuss how these can be measured.
    – Scortchi - Reinstate Monica
    Commented Oct 25, 2018 at 9:40
  • 2
    This answer by Shog9 to Please open the Hot Network Questions to auditing via the Data Explorer suggests that SE does very little (if any) logging on HNQ traffic in the current setup.
    – E.P.
    Commented Oct 25, 2018 at 11:30
  • @Dukeling Say you're a Wordpress user and your usual exposure to SE is casual visits to SO. If you see that Sitecore question, you might well go "huh, I wonder if they have a Wordpress section" and go on to find it and sign up. That's the SE's stated goal for HNQ achieved right there. Broader point: let's not be too hasty in dismissing those smaller sites.
    – E.P.
    Commented Oct 25, 2018 at 11:35
  • @EPS: Ah! Following the link in that answer, SE do track "how many people click through" & "what else do they do once they've clicked". So it would appear they do have at least high-level data on what they're getting from HNQs.
    – Scortchi - Reinstate Monica
    Commented Oct 25, 2018 at 12:12
  • 2
    @E.P. I'm not saying questions from such sites necessarily shouldn't be there, but I personally don't find them in any way useful, and I expect many other users feel the same, so I'm looking for some data that either confirms or contradicts this belief. It would make sense to filter sites based on the click-through and conversion rates (people who click and vote or post an upvoted answer) of previous questions for each combination of source and destination site. Although I'd actually just prefer to be able to manage what I see myself (like I'm currently doing with user scripts).
    – Bernhard Barker
    Commented Oct 25, 2018 at 12:25

1 Answer 1

Reset to default
5

(in the absence of official answer I'll share what I have learned from details shared by SE team in the past)

The only effectiveness measure that I saw mentioned by SE team are clicks, these were used to justify particular change in HNQ functionality here:

change... resulted in a 30% increase in engagement. All of this engagement came from repeated clicks...

If you are interested to learn more about how they track clicks, some details were exposed in an example shared a while ago in Tavern chat room:

hot network question clicks grouped by source and destination site over the past 90 days... 

Source site                               Destination site                                    HNQClicks 
----------------------------------------- --------------------------------------------------- --------- 
The Workplace Stack Exchange              Politics Stack Exchange                             133       
Unix & Linux Stack Exchange               RPG Stack Exchange                                  56        
Code Golf Stack Exchange                  Academia Stack Exchange                             69
...

it's sampled at 1%, so... multiply all of that by 100...

You see, they seem to track source and destination sites, that is one hosting the with the link to particular question and the hosting site of that question.

Another worthy detail one can see above is that clicks are sampled at 1%, meaning that they track only one click of 100. This is probably because tracking all clicks would be a difficult load.

Improve this answer
1
  • 1
    I don't want to be a bounty hunter, but I could publish the data I have on view counts of HNQs (sampled every 15 minutes). I know those count both 'organic' clicks and HNQ clicks, but I can probably combine it with data from non-HNQs and questions like this one which were closed while on the HNQ, so disappeared from the HNQ, and reopened later and reappeared on the HNQ.
    – Glorfindel Mod
    Commented Feb 6, 2019 at 7:39

You must log in to answer this question.

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