Is Wikipedia traffic dropping?

Some news outlets report that Wikipedia traffic is dropping.

• Is Wikipedia Being Hit By a Google Penalty?, SimilarWeb, July 2015, an article and concept which was picked up by more mainstream media
• Wikipedia suddenly lost a massive amount of traffic from Google, Business Insider, August 2015
• Google steals over 550 million clicks from Wikipedia in 6 months SimilarWeb, August 2015
• Part 3 – The Evolving Data Story Around Wikipedia, SimilarWeb, August 2015

In the links above, SimilarWeb started the rumor. Other media outlets, including the one I provide above, picked up the story from SimilarWeb. The SimilarWeb writer continued to speculate in a second post, and in the third post they report some information from the Wikimedia Foundation. The overall narrative is that Wikipedia traffic is dropping suddenly and drastically, and SimilarWeb and other media outlets question why this might be. Some of them speculate on the demise of Wikipedia, which is a popular story for any media outlet to report.

One of the oddities about this discussion is that insight and comment are not available on demand from “Wikipedia”. Most media would expect the Wikimedia Foundation to have its public relations department comment on this, but in fact, the comments which have come from the Wikimedia Foundation have gone to electronic mailing lists where developers and researchers have discussed the issue. This is not a traditional channel for popular communication, so I understand why the public would be confused. The Wikimedia community likes this system and of course the current way things are done is the consequence of how things used to be done. What is different now, and why it seems odd that there is no coordinated communications response to the traffic decline, is that increasingly Wikimedia projects are competing for audience attention on par with the world’s largest communication and media companies. Wikipedia is the only nonprofit in this space and by far the least funded. Wikipedia cannot afford polished metrics at this time, or at least, it is less obvious that public relations of the sort that say Facebook has should be something that Wikimedia projects should emulate.

I do not have the informed perspective of an inside developer but as a Wikipedia communications professional my daily routine puts me into conversations about Wikipedia’s metrics. Some things I understand and most things I do not. My perspective is that I do not think Wikipedia traffic is declining. The evidence is unclear. Wikipedia traffic reports themselves have always been unclear, and for some years now I have wished for better metrics about Wikipedia viewership. I am going to provide a little information which outsiders to Wikipedia might not have. What I have to share will not clear confusion about Wikipedia’s traffic trends but also are points which could be researched as factors to consider when discussing Wikipedia traffic.

Here are some possible causes for the seeming traffic drop:

Google changing Wikipedia prioritization in SERP

Most media outlets which consider the Wikipedia traffic drop say that Google changed PageRank or some other feature of Google Search with intent to lower Wikipedia’s prominence in the search engine results page (SERP). There is speculation that this is done to benefit Knowledge Graph, which is a Google product which completes with some uses of Wikipedia and a lot of the imagined uses for Wikidata in development especially now that Wikidata has consumed Freebase.

It is true that most people enter Wikipedia through Google. Wikipedia historically has been popular because it is the website which Google served as the answer to any query that matched the title of a Wikipedia article. If Google deprioritized Wikipedia in its rankings then Wikipedia would become less popular. Wikipedia’s claim to fame is that it is popular, and it is popular because the concept of SEO is sort of defined around the existence of Wikipedia. Wikipedia should be considered to have perfect SEO, or SEO as a standard and off the scale with competitors.

I cannot give any informed opinion about what Google has planned for the future of the Internet or for the future of Wikipedia. I am not sure who could give a good opinion on that matter. I hope that Google will always support Wikipedia by sending people there for information. There are perspectives that Wikipedia is to the benefit of Google and indeed all other websites, because part of the mythology of the Internet is that it is a marketplace of humanity and that human interests are all represented online. This mythology is hardly true – there is only one popular nonprofit website, and it is Wikipedia. There is only one popular website which is not supported by advertising, and again that is Wikipedia. Sue Gardner promoted this sentiment she got from Wikipedia community philosophy when she said, “Wikipedia is like a park. It’s a public space, accessible and used by everybody.” Wikipedia is in a weird enough marginalized unattractive space that for-profit interests have not seen fit to capture it as they have with the content service of all other websites. It is the last nonprofit information source and if it did not exist, the public might increasingly question who controls the Internet. While Wikipedia exists people get the idea that nonprofit presence in the media exists. There must be agreement among the major players (Google, Apple, Amazon, Facebook) that Wikipedia is allowed to exist as the only nonprofit peer. Maybe in the future Wikipedia will not need to be so self-deprecating as it stands with the others. Right now, I think Wikipedia’s influence is hidden, not known, and not clearly supported with data.

HTTP to HTTPS

The Wikimedia Foundation introduced a high-profile shift from serving websites using HTTP to HTTPS. Websites choose to deliver pages as either encrypted or unencrypted information. In his information disclosure Snowden revealed that the NSA monitors what people read on Wikipedia. Because of various privacy social movements, and because in the United States there is long-standing pre-Internet consensus among libraries that people should be able to check out books without having government or police monitor them, the Wikipedia community also takes the position that people should be able to read Wikipedia articles without having governments monitor them. The Wikimedia Foundation made hasty changes on a limited budget to move website functions to an encrypted service. For technical reasons related to that, Wikipedia traffic monitoring began to have failures in detecting when articles were being read. The recent major stories are at “Banning data encryption is ‘moronic’” and Wikimedia v. NSA. These do not talk about traffic because only organizations tracking metrics would care, but the encryption changes causing difficulty in detecting pageviews are the major reason for the seeming traffic drop. The Wikimedia community supported these changes, and few contributors to Wikipedia care about metrics, so a drop in reported metrics and introducing inaccuracy into metrics collecting processes was a serious concern to either the WMF staff or the Wikimedia community. I noticed, because I am one of a handful of people who regularly reports Wikipedia metrics to stakeholders who care only because I directed them to care about Wikipedia metrics.

Mobile traffic has never been counted

Wikipedia traffic seems to be dropping because more people are reading Wikipedia from mobile devices, and pageview and traffic counts in this Consumer Reports project have never included counts of pageviews from mobile devices including cell phones and tablets. Those counts were not included because the dataserver provided by the Wikimedia Foundation never included those counts, because when the system was set up years ago mobile traffic was insignificant to count and difficult to measure. In 2010 mobile traffic to Wikipedia was perhaps 2%, and as of 2015 it is about 50% – see “Traffic to Wikipedia’s mobile site is growing fast“. In November 2015 the Wikimedia Foundation analytics team posted that they were changing their traffic measurement protocols to do, among other things, better count mobile traffic. In those changes there is no acknowledgement that what they are doing should matter. I am linking to their changes because they are remarkable for not caring so much about talking about lack of mobile data collection. A drop in detected Wikipedia traffic could be because more readers are using mobile devices instead of desktop computers and mobile traffic is not yet countable.

Even with major traffic drops the dissemination through Wikipedia is a significant audience size. However, because a drop in the traffic count can be caused by things other than a decreasing audience size, and because there are reasons to believe there are technical problems in counting lately and no reason to believe that fewer people are reading Wikipedia, it is a fair assumption to make that Wikipedia traffic is staying the same if not growing as has been usual over the past approximately 15 years until earlier this year during the technical changes.