Wikipedia talk:Wikipedia Signpost/2016-05-28/Traffic report

Discuss this story

Thanks, guys! The redundancy means so much to me. Without redundancy there can be no competition. EllenCT (talk) 01:47, 29 May 2016 (UTC)[reply]

Topviews

@Serendipodous and Milowent: Saw you are doing a lot of math to get the numbers for the traffic report. There is a caveat with toollabs:topviews, as you might know, which I will try to explain again. The WMF pageviews API gives us the top 1000 most viewed pages on a per-day basis. So given a date range, Topviews will sum everything up and sort it accordingly. So the only time the numbers are wrong for a given page is if that page was not in the top 1000 at some point during that date range. For the top 10, you're probably only going to have a few pages that are off, and should only be off by a few thousand at most.

BUT there is any easy way to get the exact numbers. For each entry, just click on the view count on the right which will open that page up in toollabs:pageviews within the same date range. The total view counts you see there are 100% accurate, as they sum the view counts for that article for each day. There are rare cases where the top 10 in Topviews differs from the top 10 in WP:TOP5000, in which case maybe go by TOP5000 and cross-reference with toollabs:pageviews to get the exact numbers.

Very annoying work, I know, but usually you'll just be able to do the first aforementioned workaround in only a minute or two, and have the data you need. Fixing Topviews to work around this caveat isn't really feasible, but the analytics team is (hopefully) going to add weekly stats, in which case you won't need to do any extra work at all, the numbers at toollabs:topviews will always be right. You can track that progress at phab:T133575.

Let me know if you have any questions. If you want, I can prepare the top 10 (or top 25, whatever you prefer), for the next Signpost, and show you how I did it. Hope this helps! — MusikAnimal ^talk 04:05, 3 June 2016 (UTC)[reply]