We recently published a short article about increasing spam in google analytics. The spammy traffic is not only in the “Referrals” category as many think, but even in the “Direct“. We advised to create a new view within your analytics account and filter the spammy referrals. The hard part with this approach is that it does not work for 100% as the spammers come up with new domains.
Google analytics now provides data that cannot be trusted. If you have been happy with the increased traffic, this is unfortunately a very bad news for you. The only categories that might be trusted is “Organic search” and “Socials“.
Experiment with Google Analytics
We used a new domain that has never had any web site associated with it. The domain points to a single page containing only one <h1> tag and the google analytics code. (I don’t want to share the domain name yet to avoid real traffic from the readers 🙂 ). The file robots.txt even contains a command “Disallow: /”.
What are the results after a few days?
This is a standard graph showing the number of sessions on the web page. The day when we started, the traffic appeared. We were careful not to visit the page after the GA code was placed there.
What is the traffic structure?
Direct traffic is even more significant than the referrals. I cannot explain yet why this happens. However, it is an example of very bad reliability of this data.
Fake referral traffic is coming from these domains
It is obvious that the spammers can fake any kind of traffic. Bounce rate 0%, even 3 pages per visit on a single page site. This data contains only zeros in the column for Avg. session duration. However, for a more busy web site we experience non-zero time.
Where does the fake traffic come from?
It is not easy to filter the number of domains that spoil our traffic data. Referral category cannot be trusted at all, and the Direct category seems to be suspicious as well.
Filters within google analytics view are not a very elegant solution as the number of domains grows.
Let us know about your experience. Join the discussion or send comments directly to our email. If needed, we will update this article with new data later.