This is all desired behavior, of course, because if you’ve been doing something super-secret on a secure website, you don’t want to suddenly pass info about what you’ve been doing to any old non-secure site when you click an off-site link (though shame on the web developer who places sensitive information in the URL of a site, even if the URL is encrypted).
At the time, the web analytics industry’s concerns were mitigated by the expectation that relatively few users would proactively choose to search on Google’s secure site, and that consequently the data impact would be minimal. But the impact will jump significantly once the choice becomes a default.
One curious quirk of Google’s announcement is this sentence (my highlighting):
When you search from https://www.google.com, websites you visit from our organic search listings will still know that you came from Google, but won't receive information about each individual query.
This sentence caused me to waste my morning running tests of exactly what referrer information is made available by different browsers in a secure-to-insecure referral situation. The answer (as I expected) is absolutely nothing – no domain data, and certainly no URL parameter (keyword) data is available. So I am left wondering whether the sentence above is just an inaccuracy on Google’s part – when you click through from Google Secure Search, sites will not know that you came from Google. Am I missing something here? [Update: Seems I am. See bottom of the post for more details]
I should say that I generally applaud Google’s commitment to protecting privacy online in this way – despite the fact that it has been demonstrated many times that an individual’s keyword history is a valuable asset for online identity thieves, most users would not bother to secure their searches when left to their own devices. On the other hand, this move does come with a fair amount of collateral damage for anyone engaged in SEO work. Google’s hope seems to be that over time more and more sites will adopt SSL as the default, which would enable sites to capture the referring information again – but that seems like a long way off.
It seems like Google Analytics is as affected by this change as any other web analytics tool. Interestingly, though, if Google chose to, it could make the click-through information available to GA, since it captures this information via the redirect it uses on the outbound links from the Search Results page. But if it were to do this, I think there would be something of an outcry, unless Google provided a way of making that same data to other tools, perhaps via an API.
So for the time being the industry is going to have to adjust to incomplete referrer information from Google, and indeed from other search engines (such as Bing) that follow suit. Always seems to be two steps forward, one step back for the web analytics industry. Ah well, plus ca change…
Update, 10/25: Thanks to commenter Anthony below for pointing me to this post on Google+ (of course). In the comments, Eric Wu nails what is actually happening that enables Google to say that it will still be passing its domain over when users click to non-secure sites. It seems that Google will be using a non-secure redirect that has the query parameter value removed from the redirect URL. Because the redirect is non-secure, its URL will appear in the referrer logs of the destination site, but without the actual keyword. As Eric points out, this has the further unfortunate side-effect of ensuring that destination sites will not receive query information, even if they themselves set SSL as their default (though it’s not clear to me how one can force Google to link to the SSL version of a site by default). The plot thickens…