SimilarWeb published a blog post that has been getting a lot of attention, questioning whether Wikipedia has a Google penalty.  However, there are a few problems with the data from their blog posts, as well as some other significant factors that are also in play here, which SimilarWeb fails to mention.

Switching to HTTPS

First, Wikipedia made the switch to https in June.  This perfectly coincides with what SimilarWeb sees as the first month with a drop in search traffic.  SEOs are well aware of the fluctuations that occur in rankings and traffic when switching a site from HTTP to HTTPS as  a side effect while Google matches up the non-secure URLs to the new ones.

In Wikipedia’s case, the switch was on such a huge scale that it showed up as fluctuations on Mozcast data simply because the changes from the HTTP URLs to the new HTTPS ones was on such a huge scale.  When they flipped the switch on June 12th, not only were those changes evident, but the slight HTTPS ranking boost was likely evident as positions changed.  There was also a known algo change at this time, but Wikipedia didn’t seem to suffer a widespread demotion at the time, which also would have shown in Mozcast due to the sheer number of Wikipedia URLs in the SERPS Mozcast tracks.

But SimilarWeb fails to mention the switch to HTTPS, or they were unaware of this move while they looked at their own data.

“Insane” Traffic Loss only 12% overall

They also talk about a drop of 250 million desktop visits.

What I saw shocked me. Wikipedia lost an insane amount of traffic in the past 3 months. And by insane I mean that the free encyclopedia site lost more than 250 million desktop visits in just 3 months!

For most websites, this kind of a drop would be panic inducing.  But when you consider the massive size of Wikipedia, the drop was only from 2.71B to 2.38B, which only equates to 12% of overall desktop referrals.  Many sites see this kind of fluctuation month to month all the time.

Even the graph doesn’t show anything too terribly dramatic.  But when SimilarWeb touts it as being 250 million visits, it will sound pretty terrible.

University Effect

Also coinciding with this drop is the end of the typical college student’s university/college term, which generally runs through mid-May depending on the school, and through the end of May for high school students in the US.

Missing Year Over Year Data

SimilarWeb only allows users to go back far enough to compare year over year data – data that would be much more accurate to see the larger picture of what is really going on – if you pay a hefty price for a monthly subscription to access it.  So it does raise the possibility that this is merely a clever marketing ploy.

In fact, all of SimilarWeb’s data they include in the article is only going back three months – which is what users get with their free version – instead of giving readers a look at numbers that would be much more statistically relevant of at least a year, or better yet, two years.

Here is SEMRush’s data going back two years.

We do see dips, including a smaller on in June and July 2014. But we’d also have to consider the roll various updates play.

Penalties

Based on this limited data, it doesn’t appear it is penalty related, unless it is from a slight shift in just one of Google’s many algo changes they make every year.  The timing doesn’t match either Panda or Penguin.  The Quality Update would have been reflected in May traffic, since it was released on May 3rd.

There is always the possibility of a manual action from Google, but again, the drop in traffic of only 12% doesn’t seem to match that either.

Estimates & Sample Sizes

We also don’t know too much about SimilarWeb’s data gathering practices nor their sample size which also makes it more difficult to verify this with any kind of accuracy. From the website:

Our data comes from 4 main sources: 1) A panel of over 100 million monitored devices, currently the largest panel in the industry. 2) Local internet service providers (ISPs) located in many different countries. 3) Our web crawlers that scan every public website to create a highly accurate map of the digital world, and 4) Hundreds of thousands of direct measurement sources from websites and apps that are connected with us directly.

Wikipedia

Obviously the best source of data would be from Wikipedia themselves, but they were not the source for the story, SimilarWeb was.  And SimilarWeb doesn’t appear to have contacted Wikipedia for a statement or their take on the Google situation, which is unfortunate.

Wikipedia would be the ones to know with accuracy whether there is a Google traffic issue or not.

Bottom line, there could be a possible issue going on, but trying to tie in a Google penalty, with no evidence other than a 12% drop in desktop visits, seems a bit far fetched, particularly when you add a major shift of every single URL on the site from HTTP to HTTPS.

Obviously, time will tell if the trend continues, but just based on the limited SimilarWeb data they have made available, it does seem more of HTTPS, school calendar year and what is really a 12% traffic drop… and not a Google penalty.