This website is not affiliated with or endorsed by the Wikimedia Foundation, the nonprofit that maintains Wikipedia and manages the Wikipedia trademark.
Views until June 2015 for desktop come from stats.grok.se and include bot pageviews, while those from July 2015 are from the Wikimedia REST API and exclude identifiable bots and spiders (but may still end up counting many views by bots and spiders as legitimate human views). Views for mobile web, mobile app, desktop spider, and mobile web spider are from the Wikimedia REST API and available starting July 2015. Mobile web and mobile app views from July 2015 onward exclude identifiable bots, and "desktop spider" and "mobile web spider" are basically all the bots/spiders (excluding Wikipedia's own, which get identified as bots).
Wikipedia Views is a minimalistic website intended for people who're looking for information on how frequently particular Wikipedia pages or collections of pages get viewed.
Wikipedia Views has been developed by Vipul Naik using PHP and MySQL. Special thanks go to Stack Overflow for help with overcoming some of the coding hurdles that needed to be cleared to get to functioning code.
For desktop data between December 2007 and June 2015 (inclusive) we rely on statistics from stats.grok.se, which in turn uses the pagecounts-raw dump provided hourly by the Wikimedia Foundation. This dump includes all pageview data for the main domain but excludes mobile and Wikipedia Zero pageviews.
Desktop, mobile web, mobile app data, desktop spider and mobile web spider data from July 2015 onward, are from the Wikimedia REST API. Data before July 2015 for drilldowns other than desktop is not available through any easily accessible method, and we therefore do not include it. However, data starting September 2014 can in principle be reconstructed from the pagecounts-all-sites dump and pageviews dump released by the Wikimedia Foundation.
Cumulative Facebook like+comment+share counts are available starting October 2016, but not comprehensively. Only those view counts are available that were captured in real time from the Facebook API at the time. For those counts that we did not capture in time, a "cannot retrieve this data" message will be shown.
Referrer-based drilldowns are based on the Wikipedia Clickstream dataset (Wikimedia Research page), which is available for the months of January and February 2015, and February, March, April, August, and September 2016. The names of the drilldowns are not the same across months, due to changes in the way the clickstream dataset was generated. Also, values of 10 or less are missing and are thus treated as zero. If you request a referrer drilldown for a month where it does not apply, you will get "cannot retrieve this data". If you request it for a month where it does apply but it does not appear in the dataset, we will return a value of 0; however, the actual value could be any number 10 or less.
The development of the first iteration was concentrated between April 30, 2014 and July 31, 2014. Subsequent changes were concentrated in late 2016.
Code is available at GitHub.
For more on how to interpret the pageview data, see here.
How we retrieve and cache data: Here's how it works. When you send your list of (page, month, language, drilldown) combinations for which you want statistics fetched, we check, for each combination, if we already have cached data for it in our internal MySQL database. If we do, we serve the cached data. Otherwise, we fetch the data by making a HTTP page request to stats.grok.se or the Wikimedia REST API and parsing the HTML or JSON output to extract the number of pageviews, and then cache it using a MySQL database.
Cache size: Our database currently has partial or complete data for over 400,000 pages, and the main table has over 70 million rows (where each row provides the view count for a combination of page, month, language, and drilldown). The number of rows in the table is growing at a rate of between 2 million and 15 million a month; the minimum of 2 million is for filling in data for a new month for all the pages that have already been queried, and the higher end arises if we add many new pages to our cache.
Current month: Values for the current month may be cached from earlier queries, but cached values may not be accurate. In your form, under "Show technical settings (for advanced users only)" at "Enter the number of days after which ..." you can specify the number of days after which you'd like to force a purge of cached values for the current month data.
Timeout restrictions: For any individual page load, our timeout restrictions limit us to 400 page requests to stats.grok.se (fetching the pageviews for a given page, month, language, and drilldown requires 1 page request) and the Wikimedia API. However, it is possible to display more than 400 pieces of data if some of them have already been cached. You can specify your own bound on the number of external queries in your form, under "Show technical settings (for advanced users only)" at "Enter an upper bound on the number of external queries ...".
You can enter the following:
You can also select the display format and some other display-related options. You will then be presented a table (in your desired format) giving the total numer of pageviews in the selected month for each of your pages.
This is a more sophisticated display intended for cases where you want to compare data for one or more pages across multiple months. You enter the following data:
In addition, it is possible to select display options. You have a choice between HTML and CSV display, and you can choose whether you use rows for pages and columns for months, or rows for months and columns for pages.
Keep in mind that the number of page requests made to stats.grok.se is the product of the number of pages and the number of months. If this product exceeds 50, then the operation may time out unless some of the data for the pages has already been cached.
Sometimes, you're interested in getting a more big-picture view than individual monthly data. In these situations, getting pageviews by year can be more helpful. The multiple years display allows you to enter the following data: