How Many of Your Visitors Are Logged In to Google?

With Google’s announcement that logged in users will now be, by default, passed through SSL causing you to lose valuable keyword tracking data, you may be wondering how to find out how many users actually are logged in. While the change was in response to legitimate informational leakage concerns, it has left many in the webmaster community concerned with the impact it may have on our ability to track simple, non-private information such as the keyword which generated the search (more info here from our Analytics team). We have been looking at this number on one particular site for quite a while. In this case, we looked at 135,000 unique visitors from Google to a site over a period of 3 months. As you can see in the graph below, we are able to chart the percentage that are and are not logged in to Google. From our tests, the numbers do bear out that right around 10% are logged into Google at any given time.

If you would like to perform this test, you will need to follow the instructions on on the next link provided. I will go ahead and warn you, the method of acquiring this information isn’t exactly the nicest thing in the world. The image used in the example is no longer available, so I would imagine that Google frowns upon this kind of methodology of getting data. We have subsequently stopped running the tests ourselves, but knowledge is power. The method for tracking this was presented by Mike Cardwell, who exposed this method over at his blog Grepular.com a while ago. It involves loading images that you can only get access to if you are logged in, then using Javascript to check whether they loaded successfully or not. Be careful.

The Big Data Concern
The problem with losing 10% of your tracking data is not that the number represents a significant impact on your data at large. The problem is that it creates a real, unavoidable selection bias in your Google Analytics. There are, more likely than not, substantive differences between individuals who tend to be logged in and those who do not. Subsequently, conversion statistics generated from organic data in Google may misrepresent the actual behavior of your general user base because it neglects to include that 10%. This is especially important if that 10% proves to be a valuable, higher converting audience (which could be related to their technological savvy, for example).

The Legitimate Frustration
Virante has been a strong proponent of search privacy and have actually run our own advocacy campaigns to encourage Google, Yahoo and Bing to up their efforts. However, this particular change is frustrating for myriad reasons.

1. It is hard to enunciate a reason why hiding the immediate search history of a user from their destination site is a pressing concern
2. The user gives him/herself away the second they leave the HTTPS environment – perhaps not revealing the keyword, but at least revealing the destination
3. Google clearly has a method of tracking and displaying keyword activity for Adwords users, but has chosen not to include this in Google Analytics.
4. If Google believes there is a strong reason to use HTTPS, there is no clear reason to restrict it only to logged in users, seeing as Google personalizes all search results to some degree.

In my opinion, Google should simply redirect users through a page that would carry referrers, unloading them to HTTP at least whenever they are visiting a site this is not secure. The entire search session would remain private until the user chose to exit the service by visiting a page. If that page was not secure, nothing substantive is given up in privacy to insert a simple redirect that would carry the keyword referrer onto the webmaster. You could image how much better the response from webmasters would be if Google’s announcement said “and, we are protecting your referrers!”