Stories
Slash Boxes
Comments

SoylentNews is people

posted by NCommander on Friday June 27 2014, @06:15PM   Printer-friendly
from the finally-an-NCommander-post-with-<1k-words! dept.
So, given we've recently managed to find a good middle ground on the subscription front, I thought it would be nice to end the week with a lighthearted subject most of you care about.

STATISTICS!

As most of you are likely aware, we've been running piWik on SN for the last few weeks to get a better idea of what our general traffic looks like. We've learned quite a few interesting things from it, and I figured the community would love to take a look at the report itself. So, without further ado: piWik Statistics Report for mid-June. (warning: PDF file, the HTML output had far too many images to easily copy it off piWik and rehost it somewhere else, and there was no ASCII report that I could find).So, hopefully now most of you have taken a good look through the PDF. Its been fascinating watching the various trends of traffic, as well as getting an idea of the breakdown of mobile/desktop, various OS usage, and so-forth. First, a couple of notes:
  • For whatever reason, piWik groups Ubuntu separately from Linux, and it shows up as "Unknown" in the OS families list. I think this is a side-effect from one of the plugins we're using, but I'm not 100% sure.
  • Second, there are no referrals listed in the report; I specifically excluded them. There are two simple reasons: we get almost no inbound traffic from third-party sources (our top referral was Google searches at 114), and secondly, a fair number of users here have custom launch portals for themselves (basically an HTML page with links), who, due to our virtually non-existent referrals, show up in the top 25.
  • HTTPS traffic is excluded due to certificate issues, as there are users who have DNT set. If you use NoScript, it sets DNT automatically, even if you enable scripts globally. Furthermore, my original post about non-JS users being excluded was in error, there is a hidden image that cause those users to show up.
  • A few pages on the site don't include the footer template, causing hits to be slightly lower than they should be because of that.

So, with any bunch of data, we can start to draw some reasonable conclusions from it. Here's what I got, I'd love to hear your opinions below:

  • Slashcode's internal statistics monitor is badly out of touch of reality; it reports our traffic is 80% higher than piWik does. This might be an artificat of bitrot, varnish, or something else entirely.
  • I was kinda shocked to see how much of our traffic is based in the United States, I honestly though we had a more even distribution.
  • The lack of referrals is concerning; right now, we're roughly stable with the users in our community, but if you don't already know we exist, you aren't going to find us easily.

There's probably a lot more I could draw from this data, but I don't want to color the communities' impressions beside stating the obvious, so I'd like to hear what you think about it, and then go from there.

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 0) by Anonymous Coward on Friday June 27 2014, @07:11PM

    by Anonymous Coward on Friday June 27 2014, @07:11PM (#61005)

    If you collect this every month it would be shiny if you made it publicly available and preserve archives. As a more nerdy folk tracking things like OS and browser adoption could prove interesting in the future.

  • (Score: 2) by NCommander on Friday June 27 2014, @07:20PM

    by NCommander (2) Subscriber Badge <michael@casadevall.pro> on Friday June 27 2014, @07:20PM (#61012) Homepage Journal

    I was actually planning on running a post about doing something like this, perhaps keep an archive of how traffic and OS usage changes over time, with a monthly stats update post, though I'd need to reask the community to do so, we only asked permission to run piwik for a few weeks.

    --
    Still always moving
    • (Score: 0) by Anonymous Coward on Friday June 27 2014, @07:39PM

      by Anonymous Coward on Friday June 27 2014, @07:39PM (#61026)

      On the face of it I see no problem with gathering aggregate statistics . Tracking individual users is another question entirely.

      • (Score: 3, Informative) by NCommander on Friday June 27 2014, @07:42PM

        by NCommander (2) Subscriber Badge <michael@casadevall.pro> on Friday June 27 2014, @07:42PM (#61027) Homepage Journal

        We have to do the later to generate the former :-/ (there's a lot of information in piwik which isn't in the report; we purge out IPs and most other information after 7 days to limit how much we collect)

        --
        Still always moving
        • (Score: 0) by Anonymous Coward on Saturday June 28 2014, @03:01AM

          by Anonymous Coward on Saturday June 28 2014, @03:01AM (#61233)

          That's a decent compromise. It's not your job to write piwik, but I wonder if there is a way of filtering the data that goes into it, and dropping the ip addresses and other unique identifiers right at the source. It would be harder if not impossible to measure things like session time or unique visitors, but a number of indicators (eg country of visitor, OS, browser, hit-count, time) could still be available.

    • (Score: 1) by goodie on Friday June 27 2014, @07:43PM

      by goodie (1877) on Friday June 27 2014, @07:43PM (#61029) Journal

      I think it's a fantastic idea. Heck, if several sites did this we could aggregate the data and actually have an idea about these things rather than a guess. That's why when you were talking about the subscription and access to the SN db for querying I saw a use for this for the same type of purpose (although I don't know what is tracked actually in the SN db...).