Average Session Duration % wrong

  • Resolved cybeardjm

    (@didierjm)


    6 months, 1 week ago

    Hi, been using the plugin since end of December ’23 on 2 blogs.

    Have some % calculations problems right now, due to a probable “glitch” in March ’24. Below, some results for 1 blog (almost the same for the other one)

    Months
    jan 4:10
    feb 4:55 +28%
    mar 4:31 +6,379% ??? (should be negative)
    apr 1:10 -100%

    Weeks
    26/02-03/03 4:44
    4-10/03 4:14 -11%
    11-17/03 5:50 +38%
    18-24/03 15:44 +19,713% ???
    25-31/03 42:44 -86% ???
    1-7/04 1:32 -99%
    8-14/04 1:36 +4%
    15-21/04 0:47 -51%
    22-28/04 0:48 +2%
    29/04-5/05 1:55 +140%

    Which means these results are all wrong too…
    Last 3 months 30:50 +2,071%
    Last 30 days 1:44 -99%
    Last 60 days 28:06 +1,795%
    Last 90 days 11:35 +1,558%

    Is there a way to force a recalculation for these bad results?

    I remember you had a fix in update 2.3.2 (Apr. 1) on this issue, but previous erroneous calculations for most of March weren’t fixed…

    On the second blog, the problem seems to start for the 11-17/03 range (+14,631% !!!)

    Thx.

Viewing 4 replies - 1 through 4 (of 4 total)
  • Plugin Author Ben Sibley

    (@bensibley)

    Hi there,

    I’m very sorry for the miscalculations. We made a mistake in version 2.3 that allowed the session duration to get overcounted, and this was addressed in the 2.3.2 patch so it won’t affect any sessions recorded after updating. If someone spends more than 30 minutes on a page, we end the session, but this limit was lost in 2.3, which allowed a small minority of sessions to have extremely long lengths and drive up the average. For instance, someone returning to a browser tab from a prior day could be counted as one very long session and this skewed the average session duration much higher. I’m afraid there isn’t a way to recalculate because the true length of these sessions wasn’t recorded.

    Thread Starter cybeardjm

    (@didierjm)

    This means most of the stats will be wrong (depending on the choosen range) for the next year or so (if I use “previous year” before end of March 2025).

    Plugin Author Ben Sibley

    (@bensibley)

    I understand it’s not ideal. We are upset with ourselves for letting this bug slip through our testing protocols.

    I have a couple of workarounds that you may find acceptable. First, you could add a filter to exclude sessions with a duration greater than 600 seconds (View Duration in the Pages report). Here’s an example showing how it returns the the session duration to a more accurate value: https://d.pr/i/oQrkNU. This works because there are just a few sessions that are extremely long. You could probably increase the limit to 1,000+ seconds and it will still work well. This technique is very fast to apply, and you can save a report to revisit the same data quickly.

    Otherwise, we may be able to take some kind of manual intervention. It’s not possible to re-calculate the session duration, but we could potentially delete the small number of sessions that are skewing things. If you’d like to explore that route, please reach out to [email protected] as we’d need to discuss sensitive site details that aren’t appropriate to share here.

    Thread Starter cybeardjm

    (@didierjm)

    OK, THX, the filter will do.

Viewing 4 replies - 1 through 4 (of 4 total)
  • The topic ‘Average Session Duration % wrong’ is closed to new replies.