Hi all,
here is the weekly look at our most important readership metrics. Last week’s report was skipped for various reasons (one of them being that there really wasn’t much news in the week-over-week data then), but for now we’re still trying to keep this regular schedule.
As laid out earlier, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events in the preceding week, and facilitate thinking about core metrics in general.
We are still iterating on the presentation (e.g. to better take seasonality into account, in particular including year-over-year comparisons) and eventually want to create dashboards for those which are not already available in that form already. Feedback and discussion welcome.
Some other recent items of interest:
There is a new dissertation containing a chapter/paper about forecasting the time series of Wikimedia pageviews. I wrote up a review/summary for the monthly research newsletter: “Predicting Wikimedia pageviews with 2% accuracy”. (NB: that low error rate refers to a timespan of one week.)
Like the whole of WMF, the Reading team held its quarterly review meeting last week; the presentation slide deck contains a lot of readership metrics too.
In the wake of the “Google kills Wikipedia” media stories some weeks ago (that Oliver tried to rebut based on our internal data), one SEO blogger argues that actually “Google Still Loves Wikipedia (More Than Its Own Properties)”, although “Wikipedia did slide a bit in the rankings” since April which according to him is “the reason for the Wikipedia traffic loss”. The latter seems quite speculative to me, but the ranking slide, while tiny, appears to be based on a sufficiently large dataset.
Now to the usual data. (All numbers below are averages for October 5-11, 2015 unless otherwise noted.)
Pageviews
Total: 524 million/day (+1.8% from the previous week)
week until Oct 4: 514 million/day (-0.9% from the previous week)
Context (April 2015-October 2015):
Total pageviews increased for the first time again this week, after having fallen for three weeks in a row. (See also the Vital Signs dashboard.)
Desktop: 57.6%
Mobile web: 41.2%
Apps: 1.2%
Global North ratio: 77.0% of total pageviews
Context (April 2015-October 2015):
Unique app users
Android: 1.160 million/day (+1% from the previous week)
week until Oct 4: 1.148 million /day (-0.9% from the previous week)
Context (January 2015-October 2015):
Note: Due to a typo in the calculation, the average daily unique app users numbers for both iOS and Android were too high in the report for the week until September 27. It should have said 1.158 million for Android and 280k for iOS. Also, due to a separate issue (data corruption in the underlying database, https://phabricator.wikimedia.org/T114406 ) the charts plotting these two metrics were off for some earlier days in August and September, affecting all the weekly reports so far. I’m now posting this report with guesstimated data while the Analytics team is rolling out a fix. Does not seem to change the big picture in any way, but we want to get this right ;)
iOS: 278k/day (+1.5% from the previous week)
week until Oct 4: 274k / day (-2.5% from the previous week)
Context (January 2015-October 2015):
In the aforementioned quarterly review (minutes still to be posted), Lila asked about an apparent worsening of the app store ratings in September (after bugs had cost us users earlier this year already). Josh and I looked a bit through reviews from that time. There were several users complaining about seeing the UI in the wrong language and some of them connect that bug with the iOS 9 upgrade (which rolled out from September 16). That might have something to do with the decrease in the above chart in recent weeks - we don’t know for sure though.
New app installations
Android: 39,461/day (-7.1% from the previous week)
(Daily installs per device, from Google Play)
week until Oct 4: 42,482/day (-0.07% from the previous week)
Context (July-October 2015):
Showing this in a different context this time, compared to the number of uninstalls (#devices where the app was removed, 28,742/day on average last week), and zooming in on the last three months instead of a full year.
We didn’t discuss it earlier because it happened before the start of these reports, but there was a notable rise in installs from August 20 to August 23, largely sustained afterwards, and not offset by an equally large rise in uninstalls. The reason is not clear - putting it out here in case anyone has further insights - but it’s tantalizingly close to the dates where the app was part of a “Back to School” promotion in the Play store (which the WMF Partnerships team worked on with Google): August 13-20 and August 27-September 2, in the US only. So the dates don’t quite match, but our contact at Google pointed out that there could still be a causal connection, e.g. because of a third-party social media post inspired by the campaign some days earlier, or due to residual effects from the “Back to School” collection’s overall success (e.g. if a user downloaded another similar app within the collection, our app would be more likely to be recommended to them in the future since the algorithm recognizes similar apps).
iOS: 4,523/day (-2.4% from the previous week)
(download numbers from App Annie)
week until Oct 4: 4,633/day (+0.7% from the previous week)
Context (July-October 2015):
Switching to a three-month view here too this time. No source of uninstall data for iOS that I’m aware of.
----
For reference, the queries and source links used are listed below (access is needed for each). Most of the above charts are available on Commons, too.
hive (wmf)> SELECT SUM(view_count)/7000000 AS avg_daily_views_millions FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2015-10-05" AND "2015-10-11";
hive (wmf)> SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, sum(IF(access_method <> 'desktop', view_count, null)) AS mobileviews, SUM(view_count) AS allviews FROM wmf.projectview_hourly WHERE year=2015 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
hive (wmf)> SELECT access_method, SUM(view_count)/7 FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2015-10-05" AND "2015-10-11" GROUP BY access_method;
hive (wmf)> SELECT SUM(IF (FIND_IN_SET(country_code, 'AD,AL,AT,AX,BA,BE,BG,CH,CY,CZ,DE,DK,EE,ES,FI,FO,FR,FX,GB,GG,GI,GL,GR,HR,HU,IE,IL,IM,IS,IT,JE,LI,LU,LV,MC,MD,ME,MK,MT,NL,NO,PL,PT,RO,RS,RU,SE,SI,SJ,SK,SM,TR,VA,AU,CA,HK,MO,NZ,JP,SG,KR,TW,US') > 0, view_count, 0))/SUM(view_count) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2015-10-05" AND "2015-10-11";
hive (wmf)> SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")), SUM(view_count) AS all, SUM(IF (FIND_IN_SET(country_code, 'AD,AL,AT,AX,BA,BE,BG,CH,CY,CZ,DE,DK,EE,ES,FI,FO,FR,FX,GB,GG,GI,GL,GR,HR,HU,IE,IL,IM,IS,IT,JE,LI,LU,LV,MC,MD,ME,MK,MT,NL,NO,PL,PT,RO,RS,RU,SE,SI,SJ,SK,SM,TR,VA,AU,CA,HK,MO,NZ,JP,SG,KR,TW,US') > 0, view_count, 0)) AS Global_North_views FROM wmf.projectview_hourly WHERE year = 2015 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
hive (wmf)> SELECT SUM(IF(platform = 'Android',unique_count,0))/7 AS avg_Android_DAU_last_week, SUM(IF(platform = 'iOS',unique_count,0))/7 AS avg_iOS_DAU_last_week FROM wmf.mobile_apps_uniques_daily WHERE CONCAT(year,LPAD(month,2,"0"),LPAD(day,2,"0")) BETWEEN 20151005 AND 20151011;
hive (wmf)> SELECT CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, unique_count AS Android_DAU FROM wmf.mobile_apps_uniques_daily WHERE platform = 'Android';
hive (wmf)> SELECT CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, unique_count AS iOS_DAU FROM wmf.mobile_apps_uniques_daily WHERE platform = 'iOS';
https://www.appannie.com/dashboard/252257/item/324715238/downloads/?breakdown=country&date=2015-07-12~2015-10-11&chart_type=downloads&countries=ALL (select “Total”)