Link to PDF of report in Commons https://commons.wikimedia.org/wiki/File:Wikimedia_Readership_metrics_for_the_timespan_until_February_5,_2017.pdf
Hi all,
Here is the usual look https://www.mediawiki.org/wiki/Reading/Readership_metrics_reports at our most important readership metrics. This time we see an overall rise in pageviews following the seasonal winter slump, examine the recent year-over-year growth in pageviews more closely, and introduce a new day-7 retention metric for the Wikipedia iOS app.
As laid out earlier https://lists.wikimedia.org/pipermail/mobile-l/2015-September/009773.html, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events, and facilitate thinking about core metrics in general. As always; feedback and discussion welcome. Week-over-week and month-over-month changes are being recorded on the Product page https://www.mediawiki.org/wiki/Wikimedia_Product#Reading at MediaWiki.org. This edition of the report covers a timespan of five weeks.
You can also find lots of other traffic and usage data in the quarterly metrics presentation https://commons.wikimedia.org/wiki/File:Wikimedia_Foundation_Reading_metrics_Q2_2016-17_(Oct-Dec_2016).pdf for Q2 2016-2017 (October - December) that was just published by the WMF Reading team.
All numbers below are averages for January 2 - February 5, 2017 unless otherwise noted. Pageviews
Total: 582 million/day (+10.14% from the previous report)
Context (April 2015-February 2017):
See also the Vital Signs dashboard https://analytics.wikimedia.org/dashboards/vital-signs/#projects=all/metrics=Pageviews
After the seasonal winter slump, we see a rise in desktop pageviews, as expected. Mobile pageviews continue to remain at higher levels than before christmas. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating mobile pageviews by roughly 5 million views per day.
This chart looks at long-term traffic trends from May 2013 - January 2017. This shows that over this timespan, the annual change in overall pageviews was -2%, desktop has been down 15%, and mobile (web + apps) has been trending upwards at a rate of 23% per year. However, the past few months have seen total pageviews increasing year-over-year (chart further below).
To facilitate our understanding of which traffic movements are seasonal and which may indicate lasting changes, here is a chart overlaying the total pageview numbers back to May 2013 (the earliest time for which we have data according to the current pageview definition):
Total pageviews have continued rising and are now higher than before the winter holidays. The blue line shows that the increase in overall pageviews year-over-year remains (January 2017 is up 5% from January 2016). It is possible that a smaller part of this is due to unidentified bot traffic (e.g. we just updated https://phabricator.wikimedia.org/T157528 the pageview definition to exclude a recently discovered bot that had been causing up to 0.9% of total pageviews). But overall it is starting to look like a small but sustained rise in real human pageviews.
Here we see that the changed trend in recent months can be attributed to desktop pageviews, which have mostly stopped declining year-over-year. This chart also shows that the seasonal christmas dip in pageviews comes predominantly from desktop views. Most recently, January 2017 desktop pageviews are again down 1.8% from January 2016, but that is still much less than the aforementioned -15% yearly trend since 2013.
Year-over-year mobile pageviews are still increasing at a notable rate, but slower than in 2013-15. January 2017 mobile pageviews are up 13% from January 2016, less than the yearly mobile growth of 23% in the overall 2013-17 timespan. We can see how mobile pageviews always increase around the winter holidays and maintain higher levels afterwards.
Desktop: 50.9% (previous report: 51.5%)
Mobile web: 46.7% (previous report: 46.8%)
Apps: 2.4% (previous report: 1.7%)
Mobile percentage dropped a bit following the christmas spike, but remains high at an average of 49% mobile views. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating this metric by about 1%.
Global North ratio: 77.3% of total pageviews (previous report: 76.8%)
Context (January 2016-February 2017):
Although Global North pageview percentage fell following the winter spike, but overall this metric grew since the last report. Levels remain in the high 70 percentile range.
NB: We are currently rethinking this metric and might replace it with a different country selection constructed as part of the work on the New Readers project Unique devices
See the announcement blog post https://blog.wikimedia.org/2016/03/30/unique-devices-dataset/ from March 2016 for background and details on this metric. These estimated numbers are provided for all Wikimedia language projects (separately for the desktop and mobile web version). Because of the instrumentation method, there is no global metric for all projects and all languages, but it is currently being extended https://phabricator.wikimedia.org/T138027 to a cross-language global metric per project at least. For now, we track the daily numbers of English Wikipedia in this report.
Daily unique devices estimate for English Wikipedia:
-
Mobile web: 36.3 million -
Desktop: 24.3 million -
Total: 60.6 million -
Average mobile web ratio: 60.0%
Context (January 2016- February 2017):
Since the last report, both daily unique desktop devices and mobile devices grew. Mobile web daily uniques mirrors the daily pageviews chart - mobile web unique levels have remained higher than before the winter bump.
January 2017 was the first month we could compare year-over-year changes in unique devices. While pageviews on English Wikipedia are up from January 2016, the number of monthly unique devices is down 16%. For this report timespan (Jan 2 - Feb 5), daily mobile web uniques are up 2.6%, while daily desktop uniques are down 8.2%. This is currently being investigated.
While the number of daily unique desktop devices has grown overall, the ratio of pageviews to devices on desktop has steadily declined back to pre christmas levels. The small, but notable increase in daily views per device for mobile web which happened in November has held up.
New app installations
Android: 18.0k/day (-5.9% from the previous report)
Daily installs per device, from Google Play
Context (last nine months):
The number of daily installs dropped after the christmas spike, but remains higher than the number of uninstalls (except for January 26). Overall, the number of daily installs has dropped nearly 6% and daily uninstalls dropped about 8% since the last report.
iOS: 5.67k/day (+6.5% from the previous report)
Download numbers from App Annie
Context (last two months):
[image: Wikipedia iOS app daily downloads by country, Dec 6, 2016 - Feb 5, 2017 (App Annie).png]
Similar to other mobile metrics measured in this report, the iOS app download metric has remained higher than usual following the winter bump. During the timespan of this report, iOS app downloads remain fairly steady besides two large, unusual spikes at the end of January in the United Arab Emirates and the United States (which follow the spike from the Netherlands in December) which are likely anomalies. We are looking into filing a bug report with App Annie to investigate this. App user retention
Android: 15.6% (previous report: 16.8%)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 1:100 sample)
Context (last six months):
As remarked in earlier reports, this data is a bit too noisy for drawing conclusions about whether retention changed significantly between different releases. However, we can at least rule out the existence of major shifts during this timespan. There was a small spike in day 7 retention for installs on January 22nd.
iOS: 19.2% (previous report: N/A https://phabricator.wikimedia.org/T126693 )
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 100:100 sample since iOS app only sends data when users opt-in so there is a small user base)
Context (last two months):
After a long break where we refrained from reporting iOS retention here due to data quality issues with the number provided by Apple, we are happy to report iOS app 7 day retention again, based on a new EventLogging instrumentation that parallels the one on Android. There doesn’t seem to have been any significant changes in retention levels during the timespan of this report, even after new releases of the app. Looking back at December though, it appears that there was a period of lower retention which roughly corresponds with stability issues (crashes) the app was having at this time, and which have since been fixed. Unique app users
Android: 1.211 million / day (+2.7% from the previous report)
Context (last eight months):
Android daily average levels have remained higher following the christmas and new year’s bump. This matches the mobile pageviews and unique mobile devices metrics.
iOS: N/A https://phabricator.wikimedia.org/T130432
Zareen Farooqui, Data Analyst Intern, Wikimedia Foundation
Tilman Bayer, Senior Analyst, Wikimedia Foundation Data sources
For reference, the queries and source links used are listed below (access is needed for each). Unless otherwise noted, all content of this report is © Wikimedia Foundation and released under the CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ license. Most of the above charts are available on Commons, too. https://commons.wikimedia.org/wiki/Category:Wikimedia_readership_metrics_reports
SELECT year, month, CONCAT(year,"-",LPAD(month,2,"0")) as yearmonth, SUM(view_count) AS allhuman, sum(IF(access_method = 'desktop', view_count, null)) AS desktophuman FROM wmf.projectview_hourly WHERE year = 2017 AND agent_type = 'user' GROUP BY year, month ORDER BY year, month LIMIT 1000;
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 = 2017 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT access_method, SUM(view_count)/(7*5) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2017-01-02" AND "2017-02-05" GROUP BY access_method;
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,I S,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 = 2017 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.m.wikipedia%', uniques_estimate, 0)) AS enwiki_mobile_web FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.wikipedia%', uniques_estimate, 0)) AS enwiki_desktop FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) AS date, SUM(IF(access_method = 'mobile web', view_count, null)) AS mobilewebviews, SUM(IF(access_method = 'desktop', view_count, null)) AS desktopviews FROM wmf.projectview_hourly WHERE year=2017 AND agent_type = 'user' AND project = 'en.wikipedia' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
https://console.cloud.google.com/storage/browser/pubsite_prod_rev_0281252275...
https://www.appannie.com/dashboard/252257/item/324715238/downloads/
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE timestamp LIKE '2017%' AND userAgent LIKE '%-r-%' AND userAgent NOT LIKE '%Googlebot%' GROUP BY date ORDER BY DATE;
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE userAgent LIKE '%iPhone%' OR userAgent LIKE '%iOS%' GROUP BY date ORDER BY DATE;
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 year = 2017 AND platform = 'Android';
Zareen Farooqui
Thanks Zareen. I'm particularly interested in the mobile app retention percentages, which seem rather low. I wonder if it would make more sense to take all the money and employee hours that are currently being invested in mobile apps, and redirect those resources to mobile web.
Pine
On Tue, Feb 14, 2017 at 4:45 PM, Zareen Farooqui zareenf@gmail.com wrote:
Link to PDF of report in Commons https://commons.wikimedia.org/wiki/File:Wikimedia_Readership_metrics_for_the_timespan_until_February_5,_2017.pdf
Hi all,
Here is the usual look https://www.mediawiki.org/wiki/Reading/Readership_metrics_reports at our most important readership metrics. This time we see an overall rise in pageviews following the seasonal winter slump, examine the recent year-over-year growth in pageviews more closely, and introduce a new day-7 retention metric for the Wikipedia iOS app.
As laid out earlier https://lists.wikimedia.org/pipermail/mobile-l/2015-September/009773.html, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events, and facilitate thinking about core metrics in general. As always; feedback and discussion welcome. Week-over-week and month-over-month changes are being recorded on the Product page https://www.mediawiki.org/wiki/Wikimedia_Product#Reading at MediaWiki.org. This edition of the report covers a timespan of five weeks.
You can also find lots of other traffic and usage data in the quarterly metrics presentation https://commons.wikimedia.org/wiki/File:Wikimedia_Foundation_Reading_metrics_Q2_2016-17_(Oct-Dec_2016).pdf for Q2 2016-2017 (October - December) that was just published by the WMF Reading team.
All numbers below are averages for January 2 - February 5, 2017 unless otherwise noted. Pageviews
Total: 582 million/day (+10.14% from the previous report)
Context (April 2015-February 2017):
See also the Vital Signs dashboard https://analytics.wikimedia.org/dashboards/vital-signs/#projects=all/metrics=Pageviews
After the seasonal winter slump, we see a rise in desktop pageviews, as expected. Mobile pageviews continue to remain at higher levels than before christmas. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating mobile pageviews by roughly 5 million views per day.
This chart looks at long-term traffic trends from May 2013 - January 2017. This shows that over this timespan, the annual change in overall pageviews was -2%, desktop has been down 15%, and mobile (web + apps) has been trending upwards at a rate of 23% per year. However, the past few months have seen total pageviews increasing year-over-year (chart further below).
To facilitate our understanding of which traffic movements are seasonal and which may indicate lasting changes, here is a chart overlaying the total pageview numbers back to May 2013 (the earliest time for which we have data according to the current pageview definition):
Total pageviews have continued rising and are now higher than before the winter holidays. The blue line shows that the increase in overall pageviews year-over-year remains (January 2017 is up 5% from January 2016). It is possible that a smaller part of this is due to unidentified bot traffic (e.g. we just updated https://phabricator.wikimedia.org/T157528 the pageview definition to exclude a recently discovered bot that had been causing up to 0.9% of total pageviews). But overall it is starting to look like a small but sustained rise in real human pageviews.
Here we see that the changed trend in recent months can be attributed to desktop pageviews, which have mostly stopped declining year-over-year. This chart also shows that the seasonal christmas dip in pageviews comes predominantly from desktop views. Most recently, January 2017 desktop pageviews are again down 1.8% from January 2016, but that is still much less than the aforementioned -15% yearly trend since 2013.
Year-over-year mobile pageviews are still increasing at a notable rate, but slower than in 2013-15. January 2017 mobile pageviews are up 13% from January 2016, less than the yearly mobile growth of 23% in the overall 2013-17 timespan. We can see how mobile pageviews always increase around the winter holidays and maintain higher levels afterwards.
Desktop: 50.9% (previous report: 51.5%)
Mobile web: 46.7% (previous report: 46.8%)
Apps: 2.4% (previous report: 1.7%)
Mobile percentage dropped a bit following the christmas spike, but remains high at an average of 49% mobile views. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating this metric by about 1%.
Global North ratio: 77.3% of total pageviews (previous report: 76.8%)
Context (January 2016-February 2017):
Although Global North pageview percentage fell following the winter spike, but overall this metric grew since the last report. Levels remain in the high 70 percentile range.
NB: We are currently rethinking this metric and might replace it with a different country selection constructed as part of the work on the New Readers project Unique devices
See the announcement blog post https://blog.wikimedia.org/2016/03/30/unique-devices-dataset/ from March 2016 for background and details on this metric. These estimated numbers are provided for all Wikimedia language projects (separately for the desktop and mobile web version). Because of the instrumentation method, there is no global metric for all projects and all languages, but it is currently being extended https://phabricator.wikimedia.org/T138027 to a cross-language global metric per project at least. For now, we track the daily numbers of English Wikipedia in this report.
Daily unique devices estimate for English Wikipedia:
Mobile web: 36.3 million
Desktop: 24.3 million
Total: 60.6 million
Average mobile web ratio: 60.0%
Context (January 2016- February 2017):
Since the last report, both daily unique desktop devices and mobile devices grew. Mobile web daily uniques mirrors the daily pageviews chart - mobile web unique levels have remained higher than before the winter bump.
January 2017 was the first month we could compare year-over-year changes in unique devices. While pageviews on English Wikipedia are up from January 2016, the number of monthly unique devices is down 16%. For this report timespan (Jan 2 - Feb 5), daily mobile web uniques are up 2.6%, while daily desktop uniques are down 8.2%. This is currently being investigated.
While the number of daily unique desktop devices has grown overall, the ratio of pageviews to devices on desktop has steadily declined back to pre christmas levels. The small, but notable increase in daily views per device for mobile web which happened in November has held up.
New app installations
Android: 18.0k/day (-5.9% from the previous report)
Daily installs per device, from Google Play
Context (last nine months):
The number of daily installs dropped after the christmas spike, but remains higher than the number of uninstalls (except for January 26). Overall, the number of daily installs has dropped nearly 6% and daily uninstalls dropped about 8% since the last report.
iOS: 5.67k/day (+6.5% from the previous report)
Download numbers from App Annie
Context (last two months):
[image: Wikipedia iOS app daily downloads by country, Dec 6, 2016 - Feb 5, 2017 (App Annie).png]
Similar to other mobile metrics measured in this report, the iOS app download metric has remained higher than usual following the winter bump. During the timespan of this report, iOS app downloads remain fairly steady besides two large, unusual spikes at the end of January in the United Arab Emirates and the United States (which follow the spike from the Netherlands in December) which are likely anomalies. We are looking into filing a bug report with App Annie to investigate this. App user retention
Android: 15.6% (previous report: 16.8%)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 1:100 sample)
Context (last six months):
As remarked in earlier reports, this data is a bit too noisy for drawing conclusions about whether retention changed significantly between different releases. However, we can at least rule out the existence of major shifts during this timespan. There was a small spike in day 7 retention for installs on January 22nd.
iOS: 19.2% (previous report: N/A https://phabricator.wikimedia.org/T126693)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 100:100 sample since iOS app only sends data when users opt-in so there is a small user base)
Context (last two months):
After a long break where we refrained from reporting iOS retention here due to data quality issues with the number provided by Apple, we are happy to report iOS app 7 day retention again, based on a new EventLogging instrumentation that parallels the one on Android. There doesn’t seem to have been any significant changes in retention levels during the timespan of this report, even after new releases of the app. Looking back at December though, it appears that there was a period of lower retention which roughly corresponds with stability issues (crashes) the app was having at this time, and which have since been fixed. Unique app users
Android: 1.211 million / day (+2.7% from the previous report)
Context (last eight months):
Android daily average levels have remained higher following the christmas and new year’s bump. This matches the mobile pageviews and unique mobile devices metrics.
iOS: N/A https://phabricator.wikimedia.org/T130432
Zareen Farooqui, Data Analyst Intern, Wikimedia Foundation
Tilman Bayer, Senior Analyst, Wikimedia Foundation Data sources
For reference, the queries and source links used are listed below (access is needed for each). Unless otherwise noted, all content of this report is © Wikimedia Foundation and released under the CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ license. Most of the above charts are available on Commons, too. https://commons.wikimedia.org/wiki/Category:Wikimedia_readership_metrics_reports
SELECT year, month, CONCAT(year,"-",LPAD(month,2,"0")) as yearmonth, SUM(view_count) AS allhuman, sum(IF(access_method = 'desktop', view_count, null)) AS desktophuman FROM wmf.projectview_hourly WHERE year = 2017 AND agent_type = 'user' GROUP BY year, month ORDER BY year, month LIMIT 1000;
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 = 2017 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT access_method, SUM(view_count)/(7*5) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2017-01-02" AND "2017-02-05" GROUP BY access_method;
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,I S,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 = 2017 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.m.wikipedia%', uniques_estimate, 0)) AS enwiki_mobile_web FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.wikipedia%', uniques_estimate, 0)) AS enwiki_desktop FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) AS date, SUM(IF(access_method = 'mobile web', view_count, null)) AS mobilewebviews, SUM(IF(access_method = 'desktop', view_count, null)) AS desktopviews FROM wmf.projectview_hourly WHERE year=2017 AND agent_type = 'user' AND project = 'en.wikipedia' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
https://console.cloud.google.com/storage/browser/pubsite_ prod_rev_02812522755211381933/stats/installs/
https://www.appannie.com/dashboard/252257/item/324715238/downloads/
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE timestamp LIKE '2017%' AND userAgent LIKE '%-r-%' AND userAgent NOT LIKE '%Googlebot%' GROUP BY date ORDER BY DATE;
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE userAgent LIKE '%iPhone%' OR userAgent LIKE '%iOS%' GROUP BY date ORDER BY DATE;
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 year = 2017 AND platform = 'Android';
Zareen Farooqui
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Hi Pine,
out of curiosity, what is the "rather low" assessment based on? Does this refer to an industry standard (links welcome), or is it more a subjective, personal impression?
In any case, thanks for reading the report and sharing your thoughts - glad to see that it stimulates metrics-based thinking.
On Tue, Feb 14, 2017 at 8:33 PM, Pine W wiki.pine@gmail.com wrote:
Thanks Zareen. I'm particularly interested in the mobile app retention percentages, which seem rather low. I wonder if it would make more sense to take all the money and employee hours that are currently being invested in mobile apps, and redirect those resources to mobile web.
Pine
On Tue, Feb 14, 2017 at 4:45 PM, Zareen Farooqui zareenf@gmail.com wrote:
Link to PDF of report in Commons https://commons.wikimedia.org/wiki/File:Wikimedia_Readership_metrics_for_the_timespan_until_February_5,_2017.pdf
Hi all,
Here is the usual look https://www.mediawiki.org/wiki/Reading/Readership_metrics_reports at our most important readership metrics. This time we see an overall rise in pageviews following the seasonal winter slump, examine the recent year-over-year growth in pageviews more closely, and introduce a new day-7 retention metric for the Wikipedia iOS app.
As laid out earlier https://lists.wikimedia.org/pipermail/mobile-l/2015-September/009773.html, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events, and facilitate thinking about core metrics in general. As always; feedback and discussion welcome. Week-over-week and month-over-month changes are being recorded on the Product page https://www.mediawiki.org/wiki/Wikimedia_Product#Reading at MediaWiki.org. This edition of the report covers a timespan of five weeks.
You can also find lots of other traffic and usage data in the quarterly metrics presentation https://commons.wikimedia.org/wiki/File:Wikimedia_Foundation_Reading_metrics_Q2_2016-17_(Oct-Dec_2016).pdf for Q2 2016-2017 (October - December) that was just published by the WMF Reading team.
All numbers below are averages for January 2 - February 5, 2017 unless otherwise noted. Pageviews
Total: 582 million/day (+10.14% from the previous report)
Context (April 2015-February 2017):
See also the Vital Signs dashboard https://analytics.wikimedia.org/dashboards/vital-signs/#projects=all/metrics=Pageviews
After the seasonal winter slump, we see a rise in desktop pageviews, as expected. Mobile pageviews continue to remain at higher levels than before christmas. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating mobile pageviews by roughly 5 million views per day.
This chart looks at long-term traffic trends from May 2013 - January 2017. This shows that over this timespan, the annual change in overall pageviews was -2%, desktop has been down 15%, and mobile (web + apps) has been trending upwards at a rate of 23% per year. However, the past few months have seen total pageviews increasing year-over-year (chart further below).
To facilitate our understanding of which traffic movements are seasonal and which may indicate lasting changes, here is a chart overlaying the total pageview numbers back to May 2013 (the earliest time for which we have data according to the current pageview definition):
Total pageviews have continued rising and are now higher than before the winter holidays. The blue line shows that the increase in overall pageviews year-over-year remains (January 2017 is up 5% from January 2016). It is possible that a smaller part of this is due to unidentified bot traffic (e.g. we just updated https://phabricator.wikimedia.org/T157528 the pageview definition to exclude a recently discovered bot that had been causing up to 0.9% of total pageviews). But overall it is starting to look like a small but sustained rise in real human pageviews.
Here we see that the changed trend in recent months can be attributed to desktop pageviews, which have mostly stopped declining year-over-year. This chart also shows that the seasonal christmas dip in pageviews comes predominantly from desktop views. Most recently, January 2017 desktop pageviews are again down 1.8% from January 2016, but that is still much less than the aforementioned -15% yearly trend since 2013.
Year-over-year mobile pageviews are still increasing at a notable rate, but slower than in 2013-15. January 2017 mobile pageviews are up 13% from January 2016, less than the yearly mobile growth of 23% in the overall 2013-17 timespan. We can see how mobile pageviews always increase around the winter holidays and maintain higher levels afterwards.
Desktop: 50.9% (previous report: 51.5%)
Mobile web: 46.7% (previous report: 46.8%)
Apps: 2.4% (previous report: 1.7%)
Mobile percentage dropped a bit following the christmas spike, but remains high at an average of 49% mobile views. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating this metric by about 1%.
Global North ratio: 77.3% of total pageviews (previous report: 76.8%)
Context (January 2016-February 2017):
Although Global North pageview percentage fell following the winter spike, but overall this metric grew since the last report. Levels remain in the high 70 percentile range.
NB: We are currently rethinking this metric and might replace it with a different country selection constructed as part of the work on the New Readers project Unique devices
See the announcement blog post https://blog.wikimedia.org/2016/03/30/unique-devices-dataset/ from March 2016 for background and details on this metric. These estimated numbers are provided for all Wikimedia language projects (separately for the desktop and mobile web version). Because of the instrumentation method, there is no global metric for all projects and all languages, but it is currently being extended https://phabricator.wikimedia.org/T138027 to a cross-language global metric per project at least. For now, we track the daily numbers of English Wikipedia in this report.
Daily unique devices estimate for English Wikipedia:
Mobile web: 36.3 million
Desktop: 24.3 million
Total: 60.6 million
Average mobile web ratio: 60.0%
Context (January 2016- February 2017):
Since the last report, both daily unique desktop devices and mobile devices grew. Mobile web daily uniques mirrors the daily pageviews chart - mobile web unique levels have remained higher than before the winter bump.
January 2017 was the first month we could compare year-over-year changes in unique devices. While pageviews on English Wikipedia are up from January 2016, the number of monthly unique devices is down 16%. For this report timespan (Jan 2 - Feb 5), daily mobile web uniques are up 2.6%, while daily desktop uniques are down 8.2%. This is currently being investigated.
While the number of daily unique desktop devices has grown overall, the ratio of pageviews to devices on desktop has steadily declined back to pre christmas levels. The small, but notable increase in daily views per device for mobile web which happened in November has held up.
New app installations
Android: 18.0k/day (-5.9% from the previous report)
Daily installs per device, from Google Play
Context (last nine months):
The number of daily installs dropped after the christmas spike, but remains higher than the number of uninstalls (except for January 26). Overall, the number of daily installs has dropped nearly 6% and daily uninstalls dropped about 8% since the last report.
iOS: 5.67k/day (+6.5% from the previous report)
Download numbers from App Annie
Context (last two months):
[image: Wikipedia iOS app daily downloads by country, Dec 6, 2016 - Feb 5, 2017 (App Annie).png]
Similar to other mobile metrics measured in this report, the iOS app download metric has remained higher than usual following the winter bump. During the timespan of this report, iOS app downloads remain fairly steady besides two large, unusual spikes at the end of January in the United Arab Emirates and the United States (which follow the spike from the Netherlands in December) which are likely anomalies. We are looking into filing a bug report with App Annie to investigate this. App user retention
Android: 15.6% (previous report: 16.8%)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 1:100 sample)
Context (last six months):
As remarked in earlier reports, this data is a bit too noisy for drawing conclusions about whether retention changed significantly between different releases. However, we can at least rule out the existence of major shifts during this timespan. There was a small spike in day 7 retention for installs on January 22nd.
iOS: 19.2% (previous report: N/A https://phabricator.wikimedia.org/T126693)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 100:100 sample since iOS app only sends data when users opt-in so there is a small user base)
Context (last two months):
After a long break where we refrained from reporting iOS retention here due to data quality issues with the number provided by Apple, we are happy to report iOS app 7 day retention again, based on a new EventLogging instrumentation that parallels the one on Android. There doesn’t seem to have been any significant changes in retention levels during the timespan of this report, even after new releases of the app. Looking back at December though, it appears that there was a period of lower retention which roughly corresponds with stability issues (crashes) the app was having at this time, and which have since been fixed. Unique app users
Android: 1.211 million / day (+2.7% from the previous report)
Context (last eight months):
Android daily average levels have remained higher following the christmas and new year’s bump. This matches the mobile pageviews and unique mobile devices metrics.
iOS: N/A https://phabricator.wikimedia.org/T130432
Zareen Farooqui, Data Analyst Intern, Wikimedia Foundation
Tilman Bayer, Senior Analyst, Wikimedia Foundation Data sources
For reference, the queries and source links used are listed below (access is needed for each). Unless otherwise noted, all content of this report is © Wikimedia Foundation and released under the CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ license. Most of the above charts are available on Commons, too. https://commons.wikimedia.org/wiki/Category:Wikimedia_readership_metrics_reports
SELECT year, month, CONCAT(year,"-",LPAD(month,2,"0")) as yearmonth, SUM(view_count) AS allhuman, sum(IF(access_method = 'desktop', view_count, null)) AS desktophuman FROM wmf.projectview_hourly WHERE year = 2017 AND agent_type = 'user' GROUP BY year, month ORDER BY year, month LIMIT 1000;
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 = 2017 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT access_method, SUM(view_count)/(7*5) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2017-01-02" AND "2017-02-05" GROUP BY access_method;
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,I S,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 = 2017 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.m.wikipedia%', uniques_estimate, 0)) AS enwiki_mobile_web FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.wikipedia%', uniques_estimate, 0)) AS enwiki_desktop FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) AS date, SUM(IF(access_method = 'mobile web', view_count, null)) AS mobilewebviews, SUM(IF(access_method = 'desktop', view_count, null)) AS desktopviews FROM wmf.projectview_hourly WHERE year=2017 AND agent_type = 'user' AND project = 'en.wikipedia' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
https://console.cloud.google.com/storage/browser/pubsite_pro d_rev_02812522755211381933/stats/installs/
https://www.appannie.com/dashboard/252257/item/324715238/downloads/
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE timestamp LIKE '2017%' AND userAgent LIKE '%-r-%' AND userAgent NOT LIKE '%Googlebot%' GROUP BY date ORDER BY DATE;
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE userAgent LIKE '%iPhone%' OR userAgent LIKE '%iOS%' GROUP BY date ORDER BY DATE;
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 year = 2017 AND platform = 'Android';
Zareen Farooqui
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Hi Pine, I think for those who are not used to working with it, the day 7 retention rate can look low, but it's important to remember that this day 7 retention metric is not the number of users who continue to use the app after 7 days. It is the number of people who actually visit on the 7th (not the 6th, not the 8th, not the 100th) day after using it for the first time. It is a standard app metric and we fit well within the benchmark here. We are working on a similar metric for the web, and early results suggest the number for both apps is much higher than the web.
-J
On Tue, Feb 14, 2017 at 10:54 PM, Tilman Bayer tbayer@wikimedia.org wrote:
Hi Pine,
out of curiosity, what is the "rather low" assessment based on? Does this refer to an industry standard (links welcome), or is it more a subjective, personal impression?
In any case, thanks for reading the report and sharing your thoughts - glad to see that it stimulates metrics-based thinking.
On Tue, Feb 14, 2017 at 8:33 PM, Pine W wiki.pine@gmail.com wrote:
Thanks Zareen. I'm particularly interested in the mobile app retention percentages, which seem rather low. I wonder if it would make more sense to take all the money and employee hours that are currently being invested in mobile apps, and redirect those resources to mobile web.
Pine
On Tue, Feb 14, 2017 at 4:45 PM, Zareen Farooqui zareenf@gmail.com wrote:
Link to PDF of report in Commons https://commons.wikimedia.org/wiki/File:Wikimedia_Readership_metrics_for_the_timespan_until_February_5,_2017.pdf
Hi all,
Here is the usual look https://www.mediawiki.org/wiki/Reading/Readership_metrics_reports at our most important readership metrics. This time we see an overall rise in pageviews following the seasonal winter slump, examine the recent year-over-year growth in pageviews more closely, and introduce a new day-7 retention metric for the Wikipedia iOS app.
As laid out earlier https://lists.wikimedia.org/pipermail/mobile-l/2015-September/009773.html, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events, and facilitate thinking about core metrics in general. As always; feedback and discussion welcome. Week-over-week and month-over-month changes are being recorded on the Product page https://www.mediawiki.org/wiki/Wikimedia_Product#Reading at MediaWiki.org. This edition of the report covers a timespan of five weeks.
You can also find lots of other traffic and usage data in the quarterly metrics presentation https://commons.wikimedia.org/wiki/File:Wikimedia_Foundation_Reading_metrics_Q2_2016-17_(Oct-Dec_2016).pdf for Q2 2016-2017 (October - December) that was just published by the WMF Reading team.
All numbers below are averages for January 2 - February 5, 2017 unless otherwise noted. Pageviews
Total: 582 million/day (+10.14% from the previous report)
Context (April 2015-February 2017):
See also the Vital Signs dashboard https://analytics.wikimedia.org/dashboards/vital-signs/#projects=all/metrics=Pageviews
After the seasonal winter slump, we see a rise in desktop pageviews, as expected. Mobile pageviews continue to remain at higher levels than before christmas. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating mobile pageviews by roughly 5 million views per day.
This chart looks at long-term traffic trends from May 2013 - January 2017. This shows that over this timespan, the annual change in overall pageviews was -2%, desktop has been down 15%, and mobile (web + apps) has been trending upwards at a rate of 23% per year. However, the past few months have seen total pageviews increasing year-over-year (chart further below).
To facilitate our understanding of which traffic movements are seasonal and which may indicate lasting changes, here is a chart overlaying the total pageview numbers back to May 2013 (the earliest time for which we have data according to the current pageview definition):
Total pageviews have continued rising and are now higher than before the winter holidays. The blue line shows that the increase in overall pageviews year-over-year remains (January 2017 is up 5% from January 2016). It is possible that a smaller part of this is due to unidentified bot traffic (e.g. we just updated https://phabricator.wikimedia.org/T157528 the pageview definition to exclude a recently discovered bot that had been causing up to 0.9% of total pageviews). But overall it is starting to look like a small but sustained rise in real human pageviews.
Here we see that the changed trend in recent months can be attributed to desktop pageviews, which have mostly stopped declining year-over-year. This chart also shows that the seasonal christmas dip in pageviews comes predominantly from desktop views. Most recently, January 2017 desktop pageviews are again down 1.8% from January 2016, but that is still much less than the aforementioned -15% yearly trend since 2013.
Year-over-year mobile pageviews are still increasing at a notable rate, but slower than in 2013-15. January 2017 mobile pageviews are up 13% from January 2016, less than the yearly mobile growth of 23% in the overall 2013-17 timespan. We can see how mobile pageviews always increase around the winter holidays and maintain higher levels afterwards.
Desktop: 50.9% (previous report: 51.5%)
Mobile web: 46.7% (previous report: 46.8%)
Apps: 2.4% (previous report: 1.7%)
Mobile percentage dropped a bit following the christmas spike, but remains high at an average of 49% mobile views. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating this metric by about 1%.
Global North ratio: 77.3% of total pageviews (previous report: 76.8%)
Context (January 2016-February 2017):
Although Global North pageview percentage fell following the winter spike, but overall this metric grew since the last report. Levels remain in the high 70 percentile range.
NB: We are currently rethinking this metric and might replace it with a different country selection constructed as part of the work on the New Readers project Unique devices
See the announcement blog post https://blog.wikimedia.org/2016/03/30/unique-devices-dataset/ from March 2016 for background and details on this metric. These estimated numbers are provided for all Wikimedia language projects (separately for the desktop and mobile web version). Because of the instrumentation method, there is no global metric for all projects and all languages, but it is currently being extended https://phabricator.wikimedia.org/T138027 to a cross-language global metric per project at least. For now, we track the daily numbers of English Wikipedia in this report.
Daily unique devices estimate for English Wikipedia:
Mobile web: 36.3 million
Desktop: 24.3 million
Total: 60.6 million
Average mobile web ratio: 60.0%
Context (January 2016- February 2017):
Since the last report, both daily unique desktop devices and mobile devices grew. Mobile web daily uniques mirrors the daily pageviews chart - mobile web unique levels have remained higher than before the winter bump.
January 2017 was the first month we could compare year-over-year changes in unique devices. While pageviews on English Wikipedia are up from January 2016, the number of monthly unique devices is down 16%. For this report timespan (Jan 2 - Feb 5), daily mobile web uniques are up 2.6%, while daily desktop uniques are down 8.2%. This is currently being investigated.
While the number of daily unique desktop devices has grown overall, the ratio of pageviews to devices on desktop has steadily declined back to pre christmas levels. The small, but notable increase in daily views per device for mobile web which happened in November has held up.
New app installations
Android: 18.0k/day (-5.9% from the previous report)
Daily installs per device, from Google Play
Context (last nine months):
The number of daily installs dropped after the christmas spike, but remains higher than the number of uninstalls (except for January 26). Overall, the number of daily installs has dropped nearly 6% and daily uninstalls dropped about 8% since the last report.
iOS: 5.67k/day (+6.5% from the previous report)
Download numbers from App Annie
Context (last two months):
[image: Wikipedia iOS app daily downloads by country, Dec 6, 2016 - Feb 5, 2017 (App Annie).png]
Similar to other mobile metrics measured in this report, the iOS app download metric has remained higher than usual following the winter bump. During the timespan of this report, iOS app downloads remain fairly steady besides two large, unusual spikes at the end of January in the United Arab Emirates and the United States (which follow the spike from the Netherlands in December) which are likely anomalies. We are looking into filing a bug report with App Annie to investigate this. App user retention
Android: 15.6% (previous report: 16.8%)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 1:100 sample)
Context (last six months):
As remarked in earlier reports, this data is a bit too noisy for drawing conclusions about whether retention changed significantly between different releases. However, we can at least rule out the existence of major shifts during this timespan. There was a small spike in day 7 retention for installs on January 22nd.
iOS: 19.2% (previous report: N/A https://phabricator.wikimedia.org/T126693)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 100:100 sample since iOS app only sends data when users opt-in so there is a small user base)
Context (last two months):
After a long break where we refrained from reporting iOS retention here due to data quality issues with the number provided by Apple, we are happy to report iOS app 7 day retention again, based on a new EventLogging instrumentation that parallels the one on Android. There doesn’t seem to have been any significant changes in retention levels during the timespan of this report, even after new releases of the app. Looking back at December though, it appears that there was a period of lower retention which roughly corresponds with stability issues (crashes) the app was having at this time, and which have since been fixed. Unique app users
Android: 1.211 million / day (+2.7% from the previous report)
Context (last eight months):
Android daily average levels have remained higher following the christmas and new year’s bump. This matches the mobile pageviews and unique mobile devices metrics.
iOS: N/A https://phabricator.wikimedia.org/T130432
Zareen Farooqui, Data Analyst Intern, Wikimedia Foundation
Tilman Bayer, Senior Analyst, Wikimedia Foundation Data sources
For reference, the queries and source links used are listed below (access is needed for each). Unless otherwise noted, all content of this report is © Wikimedia Foundation and released under the CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ license. Most of the above charts are available on Commons, too. https://commons.wikimedia.org/wiki/Category:Wikimedia_readership_metrics_reports
SELECT year, month, CONCAT(year,"-",LPAD(month,2,"0")) as yearmonth, SUM(view_count) AS allhuman, sum(IF(access_method = 'desktop', view_count, null)) AS desktophuman FROM wmf.projectview_hourly WHERE year = 2017 AND agent_type = 'user' GROUP BY year, month ORDER BY year, month LIMIT 1000;
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 = 2017 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT access_method, SUM(view_count)/(7*5) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2017-01-02" AND "2017-02-05" GROUP BY access_method;
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,I S,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 = 2017 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.m.wikipedia%', uniques_estimate, 0)) AS enwiki_mobile_web FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.wikipedia%', uniques_estimate, 0)) AS enwiki_desktop FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) AS date, SUM(IF(access_method = 'mobile web', view_count, null)) AS mobilewebviews, SUM(IF(access_method = 'desktop', view_count, null)) AS desktopviews FROM wmf.projectview_hourly WHERE year=2017 AND agent_type = 'user' AND project = 'en.wikipedia' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
https://console.cloud.google.com/storage/browser/pubsite_pro d_rev_02812522755211381933/stats/installs/
https://www.appannie.com/dashboard/252257/item/324715238/downloads/
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE timestamp LIKE '2017%' AND userAgent LIKE '%-r-%' AND userAgent NOT LIKE '%Googlebot%' GROUP BY date ORDER BY DATE;
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE userAgent LIKE '%iPhone%' OR userAgent LIKE '%iOS%' GROUP BY date ORDER BY DATE;
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 year = 2017 AND platform = 'Android';
Zareen Farooqui
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
-- Tilman Bayer Senior Analyst Wikimedia Foundation IRC (Freenode): HaeB
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Jon, thanks for the explanation. Comparisons between mobile web and the mobile app retention rates would indeed be helpful.
Does Wikipedia Zero work only with the apps, or do users also get free access to Wikipedia mobile web?
As an editor, I previously found that the Android app was missing many features that I wanted, and I found that mobile web editing was difficult to accomplish. In the upcoming annual plan for the mobile experience, both for apps and mobile web, will there be a focus on improving usability for mobile contributors?
Thanks,
Pine
On Wed, Feb 15, 2017 at 9:54 AM, Jon Katz jkatz@wikimedia.org wrote:
Hi Pine, I think for those who are not used to working with it, the day 7 retention rate can look low, but it's important to remember that this day 7 retention metric is not the number of users who continue to use the app after 7 days. It is the number of people who actually visit on the 7th (not the 6th, not the 8th, not the 100th) day after using it for the first time. It is a standard app metric and we fit well within the benchmark here. We are working on a similar metric for the web, and early results suggest the number for both apps is much higher than the web.
-J
On Tue, Feb 14, 2017 at 10:54 PM, Tilman Bayer tbayer@wikimedia.org wrote:
Hi Pine,
out of curiosity, what is the "rather low" assessment based on? Does this refer to an industry standard (links welcome), or is it more a subjective, personal impression?
In any case, thanks for reading the report and sharing your thoughts - glad to see that it stimulates metrics-based thinking.
On Tue, Feb 14, 2017 at 8:33 PM, Pine W wiki.pine@gmail.com wrote:
Thanks Zareen. I'm particularly interested in the mobile app retention percentages, which seem rather low. I wonder if it would make more sense to take all the money and employee hours that are currently being invested in mobile apps, and redirect those resources to mobile web.
Pine
On Tue, Feb 14, 2017 at 4:45 PM, Zareen Farooqui zareenf@gmail.com wrote:
Link to PDF of report in Commons https://commons.wikimedia.org/wiki/File:Wikimedia_Readership_metrics_for_the_timespan_until_February_5,_2017.pdf
Hi all,
Here is the usual look https://www.mediawiki.org/wiki/Reading/Readership_metrics_reports at our most important readership metrics. This time we see an overall rise in pageviews following the seasonal winter slump, examine the recent year-over-year growth in pageviews more closely, and introduce a new day-7 retention metric for the Wikipedia iOS app.
As laid out earlier https://lists.wikimedia.org/pipermail/mobile-l/2015-September/009773.html, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events, and facilitate thinking about core metrics in general. As always; feedback and discussion welcome. Week-over-week and month-over-month changes are being recorded on the Product page https://www.mediawiki.org/wiki/Wikimedia_Product#Reading at MediaWiki.org. This edition of the report covers a timespan of five weeks.
You can also find lots of other traffic and usage data in the quarterly metrics presentation https://commons.wikimedia.org/wiki/File:Wikimedia_Foundation_Reading_metrics_Q2_2016-17_(Oct-Dec_2016).pdf for Q2 2016-2017 (October - December) that was just published by the WMF Reading team.
All numbers below are averages for January 2 - February 5, 2017 unless otherwise noted. Pageviews
Total: 582 million/day (+10.14% from the previous report)
Context (April 2015-February 2017):
See also the Vital Signs dashboard https://analytics.wikimedia.org/dashboards/vital-signs/#projects=all/metrics=Pageviews
After the seasonal winter slump, we see a rise in desktop pageviews, as expected. Mobile pageviews continue to remain at higher levels than before christmas. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating mobile pageviews by roughly 5 million views per day.
This chart looks at long-term traffic trends from May 2013 - January 2017. This shows that over this timespan, the annual change in overall pageviews was -2%, desktop has been down 15%, and mobile (web + apps) has been trending upwards at a rate of 23% per year. However, the past few months have seen total pageviews increasing year-over-year (chart further below).
To facilitate our understanding of which traffic movements are seasonal and which may indicate lasting changes, here is a chart overlaying the total pageview numbers back to May 2013 (the earliest time for which we have data according to the current pageview definition):
Total pageviews have continued rising and are now higher than before the winter holidays. The blue line shows that the increase in overall pageviews year-over-year remains (January 2017 is up 5% from January 2016). It is possible that a smaller part of this is due to unidentified bot traffic (e.g. we just updated https://phabricator.wikimedia.org/T157528 the pageview definition to exclude a recently discovered bot that had been causing up to 0.9% of total pageviews). But overall it is starting to look like a small but sustained rise in real human pageviews.
Here we see that the changed trend in recent months can be attributed to desktop pageviews, which have mostly stopped declining year-over-year. This chart also shows that the seasonal christmas dip in pageviews comes predominantly from desktop views. Most recently, January 2017 desktop pageviews are again down 1.8% from January 2016, but that is still much less than the aforementioned -15% yearly trend since 2013.
Year-over-year mobile pageviews are still increasing at a notable rate, but slower than in 2013-15. January 2017 mobile pageviews are up 13% from January 2016, less than the yearly mobile growth of 23% in the overall 2013-17 timespan. We can see how mobile pageviews always increase around the winter holidays and maintain higher levels afterwards.
Desktop: 50.9% (previous report: 51.5%)
Mobile web: 46.7% (previous report: 46.8%)
Apps: 2.4% (previous report: 1.7%)
Mobile percentage dropped a bit following the christmas spike, but remains high at an average of 49% mobile views. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating this metric by about 1%.
Global North ratio: 77.3% of total pageviews (previous report: 76.8%)
Context (January 2016-February 2017):
Although Global North pageview percentage fell following the winter spike, but overall this metric grew since the last report. Levels remain in the high 70 percentile range.
NB: We are currently rethinking this metric and might replace it with a different country selection constructed as part of the work on the New Readers project Unique devices
See the announcement blog post https://blog.wikimedia.org/2016/03/30/unique-devices-dataset/ from March 2016 for background and details on this metric. These estimated numbers are provided for all Wikimedia language projects (separately for the desktop and mobile web version). Because of the instrumentation method, there is no global metric for all projects and all languages, but it is currently being extended https://phabricator.wikimedia.org/T138027 to a cross-language global metric per project at least. For now, we track the daily numbers of English Wikipedia in this report.
Daily unique devices estimate for English Wikipedia:
Mobile web: 36.3 million
Desktop: 24.3 million
Total: 60.6 million
Average mobile web ratio: 60.0%
Context (January 2016- February 2017):
Since the last report, both daily unique desktop devices and mobile devices grew. Mobile web daily uniques mirrors the daily pageviews chart - mobile web unique levels have remained higher than before the winter bump.
January 2017 was the first month we could compare year-over-year changes in unique devices. While pageviews on English Wikipedia are up from January 2016, the number of monthly unique devices is down 16%. For this report timespan (Jan 2 - Feb 5), daily mobile web uniques are up 2.6%, while daily desktop uniques are down 8.2%. This is currently being investigated.
While the number of daily unique desktop devices has grown overall, the ratio of pageviews to devices on desktop has steadily declined back to pre christmas levels. The small, but notable increase in daily views per device for mobile web which happened in November has held up.
New app installations
Android: 18.0k/day (-5.9% from the previous report)
Daily installs per device, from Google Play
Context (last nine months):
The number of daily installs dropped after the christmas spike, but remains higher than the number of uninstalls (except for January 26). Overall, the number of daily installs has dropped nearly 6% and daily uninstalls dropped about 8% since the last report.
iOS: 5.67k/day (+6.5% from the previous report)
Download numbers from App Annie
Context (last two months):
[image: Wikipedia iOS app daily downloads by country, Dec 6, 2016 - Feb 5, 2017 (App Annie).png]
Similar to other mobile metrics measured in this report, the iOS app download metric has remained higher than usual following the winter bump. During the timespan of this report, iOS app downloads remain fairly steady besides two large, unusual spikes at the end of January in the United Arab Emirates and the United States (which follow the spike from the Netherlands in December) which are likely anomalies. We are looking into filing a bug report with App Annie to investigate this. App user retention
Android: 15.6% (previous report: 16.8%)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 1:100 sample)
Context (last six months):
As remarked in earlier reports, this data is a bit too noisy for drawing conclusions about whether retention changed significantly between different releases. However, we can at least rule out the existence of major shifts during this timespan. There was a small spike in day 7 retention for installs on January 22nd.
iOS: 19.2% (previous report: N/A https://phabricator.wikimedia.org/T126693)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 100:100 sample since iOS app only sends data when users opt-in so there is a small user base)
Context (last two months):
After a long break where we refrained from reporting iOS retention here due to data quality issues with the number provided by Apple, we are happy to report iOS app 7 day retention again, based on a new EventLogging instrumentation that parallels the one on Android. There doesn’t seem to have been any significant changes in retention levels during the timespan of this report, even after new releases of the app. Looking back at December though, it appears that there was a period of lower retention which roughly corresponds with stability issues (crashes) the app was having at this time, and which have since been fixed. Unique app users
Android: 1.211 million / day (+2.7% from the previous report)
Context (last eight months):
Android daily average levels have remained higher following the christmas and new year’s bump. This matches the mobile pageviews and unique mobile devices metrics.
iOS: N/A https://phabricator.wikimedia.org/T130432
Zareen Farooqui, Data Analyst Intern, Wikimedia Foundation
Tilman Bayer, Senior Analyst, Wikimedia Foundation Data sources
For reference, the queries and source links used are listed below (access is needed for each). Unless otherwise noted, all content of this report is © Wikimedia Foundation and released under the CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ license. Most of the above charts are available on Commons, too. https://commons.wikimedia.org/wiki/Category:Wikimedia_readership_metrics_reports
SELECT year, month, CONCAT(year,"-",LPAD(month,2,"0")) as yearmonth, SUM(view_count) AS allhuman, sum(IF(access_method = 'desktop', view_count, null)) AS desktophuman FROM wmf.projectview_hourly WHERE year = 2017 AND agent_type = 'user' GROUP BY year, month ORDER BY year, month LIMIT 1000;
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 = 2017 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT access_method, SUM(view_count)/(7*5) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2017-01-02" AND "2017-02-05" GROUP BY access_method;
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,I S,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 = 2017 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.m.wikipedia%', uniques_estimate, 0)) AS enwiki_mobile_web FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.wikipedia%', uniques_estimate, 0)) AS enwiki_desktop FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) AS date, SUM(IF(access_method = 'mobile web', view_count, null)) AS mobilewebviews, SUM(IF(access_method = 'desktop', view_count, null)) AS desktopviews FROM wmf.projectview_hourly WHERE year=2017 AND agent_type = 'user' AND project = 'en.wikipedia' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
https://console.cloud.google.com/storage/browser/pubsite_pro d_rev_02812522755211381933/stats/installs/
https://www.appannie.com/dashboard/252257/item/324715238/downloads/
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE timestamp LIKE '2017%' AND userAgent LIKE '%-r-%' AND userAgent NOT LIKE '%Googlebot%' GROUP BY date ORDER BY DATE;
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE userAgent LIKE '%iPhone%' OR userAgent LIKE '%iOS%' GROUP BY date ORDER BY DATE;
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 year = 2017 AND platform = 'Android';
Zareen Farooqui
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
-- Tilman Bayer Senior Analyst Wikimedia Foundation IRC (Freenode): HaeB
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
Hi Pine, I believe I answer your questions inline.
On Wednesday, February 15, 2017, Pine W wiki.pine@gmail.com wrote:
Jon, thanks for the explanation. Comparisons between mobile web and the mobile app retention rates would indeed be helpful.
Does Wikipedia Zero work only with the apps, or do users also get free access to Wikipedia mobile web?
Wikipedia Zero works on *both* apps and mobile web.
As an editor, I previously found that the Android app was missing many features that I wanted, and I found that mobile web editing was difficult to accomplish. In the upcoming annual plan for the mobile experience, both for apps and mobile web, will there be a focus on improving usability for mobile contributors?
So glad you asked as its something I am personally excited about! A consultation on one angle https://www.mediawiki.org/wiki/Reading/Readers_contributions_via_Android on this subject is wrapping up on this topic and I will be summarizing the results shortly--feel free to chime in as it hasn't ended yet. Our Android app is also exploring a specific approach- editing wikidata descriptions in a mobile-optimized way https://www.wikidata.org/wiki/Wikidata:Requests_for_comment/Editing_descriptions_from_Wikipedia_Android_app. As far as annual plan, we are still working on a draft for community review and I don't want to make any statements that might be proven wrong, but it's definitely under consideration.
Thanks,
Pine
On Wed, Feb 15, 2017 at 9:54 AM, Jon Katz <jkatz@wikimedia.org javascript:_e(%7B%7D,'cvml','jkatz@wikimedia.org');> wrote:
Hi Pine, I think for those who are not used to working with it, the day 7 retention rate can look low, but it's important to remember that this day 7 retention metric is not the number of users who continue to use the app after 7 days. It is the number of people who actually visit on the 7th (not the 6th, not the 8th, not the 100th) day after using it for the first time. It is a standard app metric and we fit well within the benchmark here. We are working on a similar metric for the web, and early results suggest the number for both apps is much higher than the web.
-J
On Tue, Feb 14, 2017 at 10:54 PM, Tilman Bayer <tbayer@wikimedia.org javascript:_e(%7B%7D,'cvml','tbayer@wikimedia.org');> wrote:
Hi Pine,
out of curiosity, what is the "rather low" assessment based on? Does this refer to an industry standard (links welcome), or is it more a subjective, personal impression?
In any case, thanks for reading the report and sharing your thoughts - glad to see that it stimulates metrics-based thinking.
On Tue, Feb 14, 2017 at 8:33 PM, Pine W <wiki.pine@gmail.com javascript:_e(%7B%7D,'cvml','wiki.pine@gmail.com');> wrote:
Thanks Zareen. I'm particularly interested in the mobile app retention percentages, which seem rather low. I wonder if it would make more sense to take all the money and employee hours that are currently being invested in mobile apps, and redirect those resources to mobile web.
Pine
On Tue, Feb 14, 2017 at 4:45 PM, Zareen Farooqui <zareenf@gmail.com javascript:_e(%7B%7D,'cvml','zareenf@gmail.com');> wrote:
Link to PDF of report in Commons https://commons.wikimedia.org/wiki/File:Wikimedia_Readership_metrics_for_the_timespan_until_February_5,_2017.pdf
Hi all,
Here is the usual look https://www.mediawiki.org/wiki/Reading/Readership_metrics_reports at our most important readership metrics. This time we see an overall rise in pageviews following the seasonal winter slump, examine the recent year-over-year growth in pageviews more closely, and introduce a new day-7 retention metric for the Wikipedia iOS app.
As laid out earlier https://lists.wikimedia.org/pipermail/mobile-l/2015-September/009773.html, the main purpose is to raise awareness about how these are developing, call out the impact of any unusual events, and facilitate thinking about core metrics in general. As always; feedback and discussion welcome. Week-over-week and month-over-month changes are being recorded on the Product page https://www.mediawiki.org/wiki/Wikimedia_Product#Reading at MediaWiki.org. This edition of the report covers a timespan of five weeks.
You can also find lots of other traffic and usage data in the quarterly metrics presentation https://commons.wikimedia.org/wiki/File:Wikimedia_Foundation_Reading_metrics_Q2_2016-17_(Oct-Dec_2016).pdf for Q2 2016-2017 (October - December) that was just published by the WMF Reading team.
All numbers below are averages for January 2 - February 5, 2017 unless otherwise noted. Pageviews
Total: 582 million/day (+10.14% from the previous report)
Context (April 2015-February 2017):
See also the Vital Signs dashboard https://analytics.wikimedia.org/dashboards/vital-signs/#projects=all/metrics=Pageviews
After the seasonal winter slump, we see a rise in desktop pageviews, as expected. Mobile pageviews continue to remain at higher levels than before christmas. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating mobile pageviews by roughly 5 million views per day.
This chart looks at long-term traffic trends from May 2013 - January 2017. This shows that over this timespan, the annual change in overall pageviews was -2%, desktop has been down 15%, and mobile (web + apps) has been trending upwards at a rate of 23% per year. However, the past few months have seen total pageviews increasing year-over-year (chart further below).
To facilitate our understanding of which traffic movements are seasonal and which may indicate lasting changes, here is a chart overlaying the total pageview numbers back to May 2013 (the earliest time for which we have data according to the current pageview definition):
Total pageviews have continued rising and are now higher than before the winter holidays. The blue line shows that the increase in overall pageviews year-over-year remains (January 2017 is up 5% from January 2016). It is possible that a smaller part of this is due to unidentified bot traffic (e.g. we just updated https://phabricator.wikimedia.org/T157528 the pageview definition to exclude a recently discovered bot that had been causing up to 0.9% of total pageviews). But overall it is starting to look like a small but sustained rise in real human pageviews.
Here we see that the changed trend in recent months can be attributed to desktop pageviews, which have mostly stopped declining year-over-year. This chart also shows that the seasonal christmas dip in pageviews comes predominantly from desktop views. Most recently, January 2017 desktop pageviews are again down 1.8% from January 2016, but that is still much less than the aforementioned -15% yearly trend since 2013.
Year-over-year mobile pageviews are still increasing at a notable rate, but slower than in 2013-15. January 2017 mobile pageviews are up 13% from January 2016, less than the yearly mobile growth of 23% in the overall 2013-17 timespan. We can see how mobile pageviews always increase around the winter holidays and maintain higher levels afterwards.
Desktop: 50.9% (previous report: 51.5%)
Mobile web: 46.7% (previous report: 46.8%)
Apps: 2.4% (previous report: 1.7%)
Mobile percentage dropped a bit following the christmas spike, but remains high at an average of 49% mobile views. The previously mentioned iOS app’s pageview increase is still under investigation, and may turn out to be an anomaly https://phabricator.wikimedia.org/T154735 inflating this metric by about 1%.
Global North ratio: 77.3% of total pageviews (previous report: 76.8%)
Context (January 2016-February 2017):
Although Global North pageview percentage fell following the winter spike, but overall this metric grew since the last report. Levels remain in the high 70 percentile range.
NB: We are currently rethinking this metric and might replace it with a different country selection constructed as part of the work on the New Readers project Unique devices
See the announcement blog post https://blog.wikimedia.org/2016/03/30/unique-devices-dataset/ from March 2016 for background and details on this metric. These estimated numbers are provided for all Wikimedia language projects (separately for the desktop and mobile web version). Because of the instrumentation method, there is no global metric for all projects and all languages, but it is currently being extended https://phabricator.wikimedia.org/T138027 to a cross-language global metric per project at least. For now, we track the daily numbers of English Wikipedia in this report.
Daily unique devices estimate for English Wikipedia:
Mobile web: 36.3 million
Desktop: 24.3 million
Total: 60.6 million
Average mobile web ratio: 60.0%
Context (January 2016- February 2017):
Since the last report, both daily unique desktop devices and mobile devices grew. Mobile web daily uniques mirrors the daily pageviews chart - mobile web unique levels have remained higher than before the winter bump.
January 2017 was the first month we could compare year-over-year changes in unique devices. While pageviews on English Wikipedia are up from January 2016, the number of monthly unique devices is down 16%. For this report timespan (Jan 2 - Feb 5), daily mobile web uniques are up 2.6%, while daily desktop uniques are down 8.2%. This is currently being investigated.
While the number of daily unique desktop devices has grown overall, the ratio of pageviews to devices on desktop has steadily declined back to pre christmas levels. The small, but notable increase in daily views per device for mobile web which happened in November has held up.
New app installations
Android: 18.0k/day (-5.9% from the previous report)
Daily installs per device, from Google Play
Context (last nine months):
The number of daily installs dropped after the christmas spike, but remains higher than the number of uninstalls (except for January 26). Overall, the number of daily installs has dropped nearly 6% and daily uninstalls dropped about 8% since the last report.
iOS: 5.67k/day (+6.5% from the previous report)
Download numbers from App Annie
Context (last two months):
[image: Wikipedia iOS app daily downloads by country, Dec 6, 2016 - Feb 5, 2017 (App Annie).png]
Similar to other mobile metrics measured in this report, the iOS app download metric has remained higher than usual following the winter bump. During the timespan of this report, iOS app downloads remain fairly steady besides two large, unusual spikes at the end of January in the United Arab Emirates and the United States (which follow the spike from the Netherlands in December) which are likely anomalies. We are looking into filing a bug report with App Annie to investigate this. App user retention
Android: 15.6% (previous report: 16.8%)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 1:100 sample)
Context (last six months):
As remarked in earlier reports, this data is a bit too noisy for drawing conclusions about whether retention changed significantly between different releases. However, we can at least rule out the existence of major shifts during this timespan. There was a small spike in day 7 retention for installs on January 22nd.
iOS: 19.2% (previous report: N/A https://phabricator.wikimedia.org/T126693)
(Ratio of app installs opened again 7 days after installation for all dates that fall within this report. 100:100 sample since iOS app only sends data when users opt-in so there is a small user base)
Context (last two months):
After a long break where we refrained from reporting iOS retention here due to data quality issues with the number provided by Apple, we are happy to report iOS app 7 day retention again, based on a new EventLogging instrumentation that parallels the one on Android. There doesn’t seem to have been any significant changes in retention levels during the timespan of this report, even after new releases of the app. Looking back at December though, it appears that there was a period of lower retention which roughly corresponds with stability issues (crashes) the app was having at this time, and which have since been fixed. Unique app users
Android: 1.211 million / day (+2.7% from the previous report)
Context (last eight months):
Android daily average levels have remained higher following the christmas and new year’s bump. This matches the mobile pageviews and unique mobile devices metrics.
iOS: N/A https://phabricator.wikimedia.org/T130432
Zareen Farooqui, Data Analyst Intern, Wikimedia Foundation
Tilman Bayer, Senior Analyst, Wikimedia Foundation Data sources
For reference, the queries and source links used are listed below (access is needed for each). Unless otherwise noted, all content of this report is © Wikimedia Foundation and released under the CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ license. Most of the above charts are available on Commons, too. https://commons.wikimedia.org/wiki/Category:Wikimedia_readership_metrics_reports
SELECT year, month, CONCAT(year,"-",LPAD(month,2,"0")) as yearmonth, SUM(view_count) AS allhuman, sum(IF(access_method = 'desktop', view_count, null)) AS desktophuman FROM wmf.projectview_hourly WHERE year = 2017 AND agent_type = 'user' GROUP BY year, month ORDER BY year, month LIMIT 1000;
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 = 2017 AND agent_type = 'user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT access_method, SUM(view_count)/(7*5) FROM wmf.projectview_hourly WHERE agent_type = 'user' AND CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) BETWEEN "2017-01-02" AND "2017-02-05" GROUP BY access_method;
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,I S,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 = 2017 AND agent_type='user' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.m.wikipedia%', uniques_estimate, 0)) AS enwiki_mobile_web FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) as date, SUM(IF(uri_host LIKE 'en.wikipedia%', uniques_estimate, 0)) AS enwiki_desktop FROM wmf.last_access_uniques_daily WHERE year=2017 GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
SELECT year, month, day, CONCAT(year,"-",LPAD(month,2,"0"),"-",LPAD(day,2,"0")) AS date, SUM(IF(access_method = 'mobile web', view_count, null)) AS mobilewebviews, SUM(IF(access_method = 'desktop', view_count, null)) AS desktopviews FROM wmf.projectview_hourly WHERE year=2017 AND agent_type = 'user' AND project = 'en.wikipedia' GROUP BY year, month, day ORDER BY year, month, day LIMIT 1000;
https://console.cloud.google.com/storage/browser/pubsite_pro d_rev_02812522755211381933/stats/installs/
https://www.appannie.com/dashboard/252257/item/324715238/downloads/
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE timestamp LIKE '2017%' AND userAgent LIKE '%-r-%' AND userAgent NOT LIKE '%Googlebot%' GROUP BY date ORDER BY DATE;
SELECT LEFT(timestamp, 8) AS date, SUM(IF(event_appInstallAgeDays = 0, 1, 0)) AS day0_active, SUM(IF(event_appInstallAgeDays = 7, 1, 0)) AS day7_active FROM log.MobileWikiAppDailyStats_12637385 WHERE userAgent LIKE '%iPhone%' OR userAgent LIKE '%iOS%' GROUP BY date ORDER BY DATE;
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 year = 2017 AND platform = 'Android';
Zareen Farooqui
Mobile-l mailing list Mobile-l@lists.wikimedia.org javascript:_e(%7B%7D,'cvml','Mobile-l@lists.wikimedia.org'); https://lists.wikimedia.org/mailman/listinfo/mobile-l
Mobile-l mailing list Mobile-l@lists.wikimedia.org javascript:_e(%7B%7D,'cvml','Mobile-l@lists.wikimedia.org'); https://lists.wikimedia.org/mailman/listinfo/mobile-l
-- Tilman Bayer Senior Analyst Wikimedia Foundation IRC (Freenode): HaeB
Mobile-l mailing list Mobile-l@lists.wikimedia.org javascript:_e(%7B%7D,'cvml','Mobile-l@lists.wikimedia.org'); https://lists.wikimedia.org/mailman/listinfo/mobile-l
On Wed, Feb 15, 2017 at 9:54 AM, Jon Katz jkatz@wikimedia.org wrote:
early results suggest the number for both apps is much higher than the web.
Hey Pine, I need to walk this portion of my email back. I'm not saying the opposite is true, just that it is just too early to say this with confidence. Thanks to Tilman for rightly pointing this out ;). I apologize for any confusion caused. Thanks, J
OK, thanks Tilman and Jon. I look forward to seeing the mobile web retention info.
Encouraging mobile users to contribute to Wikidata makes sense.
I think there could be gamified ways of encouraging contribution to Wikidata in particular, but perhaps also to infoboxes, maps, and Commons. However, getting quality contributions instead of spam, vandalism, nonsense, or low-quality contributions might be a challenge.
Editing full text Wikipedia articles by phone was painful when I last tried it on mobile web. I'm hoping that there have been improvements since that time, and that there will be such big improvements during the next FY that people will find editing full text articles only modestly more difficult on mobile than it is on desktop. That's ambitious, but given the trends toward mobile devices globally and that many people in less developed parts of the world may have access to Wikipedia Zero on their phones but no desktop access, I think that this direction is worth at least exploratory effort.
Looking forward to future developments,
Pine