We all have heard the news that Google Plus is not developing as some people think it should. The Daily Mail even reported “Traffic plunges for Google+ as 60% of users log off“. The newspaper based it’s piece on a report by analytics firm Chitika, who thinks that the launch of Google Plus into public beta has been a failure.
First of all, I wouldn’t really see it as a failure if traffic first surges to 1200% of the value before the launch of the public beta and then drops by 60% resulting in a 480% increase all in all. Not really that bad, I think.
Ok, now to something completely different though not unrelated: version 14 of Google’s Chrome browser has been released on September 16th, which happens to be one day before Chitika’s graph starts. Unfortunately I don’t know how quickly the automatic updates of Google Chrome spread, but if it takes several days as I could imagine then this puts Chitika’s stats into some doubt.
I believe there might be a correlation between the rise in usage share of Chrome 14 and the decline of Chitika’s Google+ traffic index. As SEOMoz readers you all know that correlation is not causation so I had a deeper look into it. At first I recognized that if I visit my blog from my Google+ About page, Piwik logs a “Direct Entry”. A look into the access log files confirms that – the referrer field just contains “-”.
So I decided to take a deeper look and fired up OWASP’s Zed Attack Proxy to analyse the traffic between my browsers and the Internet.
First I recorded a session with Firefox 7. For the test, I simply clicked on the link to my blog which is on the About me page on Google+. Here’s the result in ZAP:
The sequence of requests here is:
The resulting request to the target server looks like this:
A pretty normal request including a ‘Referer’ which indicates to the target server and its analytics software that the visitor has been referred by Google+.
This same session using Google Chrome 15 looks a bit different in ZAP:
This time there are only two relevant requests:
The request to /url on Google Plus is basically the same as with Firefox, but with a notable difference: This one uses the https protocol. It’s an encrypted request and therefore the browser rightly does not transfer a Referer header. Just as requested by RFC2616:
Frankly, I have no idea why Google would do that. Why should they treat users differently depending on the browser simply for linking to another web site. Originally I thought I had found a bug in Chromium until I noticed that in one case the redirect page is transferred encrypted while in the other it is not.
Concerning Chitika’s statistics which I cited at the beginning of this article, this means that the start of the public beta of Google Plus might have overlapped with the spreading of Google Chrome 14 and thus decreasing referral data during that launch. So in reality the traffic index of Google+ after the rollout of Chrome 14 should be about 25% better as Chitika uses referral data to determine the traffic index.
Furthermore it means that you should not rely on your referrer data if you want to really measure the visitor numbers coming from Google+. And especially it means you can’t compare Referrer data from Google Plus and other Social Media sites as you would compare apples with pears. If you really want to know how many visits you get from Google+ you have to seed links which you can later identify. But be aware! Don’t create duplicate content along the way and bear in mind that others will just distribute any link they like, so you’ll never get the full picture anyway.