It didn’t take long to wrap my head around the collateral damage that would occur when Google suppresses their outbound referring URL information when patrons are using the search engine while they’re logged in. An article forwarded to me by a friend in the business laid out some of the damage that was expected. My friend went on to point out that some of his competition might be impacted by Google’s actions, he was not entirely displeased.
I’m trying to think of the impact so I looked around for some back of the napkin numbers, using the Google search engine of course. Be mindful, these are Internet numbers so they are obviously without reproach. I started with the assumption that most folks that have a Google account are probably using Gmail, so that’d be the most impactful user base. According to Panda algorithm chosen web page at the top of my search results it seems that Gmail had about 193 million users at the end of 2010. If Google trusts itself then this is probably accurate information.
That’s not everybody. I asked Amazon and its sorta saying that there are 1.7 billion unique Internet users. So Google’s got 11% that could potentially be logged in while user the search engine. I don’t log in at work, so let’s assume that I’m half of all people and I’m at work for half of my searches. So 11% drops to 8.25% of Internet searches impacted by this. Maybe we’ll give Google+ a little love and just say 10%. Sound good? . . .Yeah, I think so too. You know what 10% looks like to me, it looks like an experiment.
Google’s Double Standard
So what is Google up to? If you buy the spin you’ll believe that they’re trying to protect user privacy. The thinking is that when a user searches on Google they are sharing those search keywords with Google only. A Google user is, apparently, not aware that a query string contains the keywords and that it follows them to the web serving hosting the site anything they click on. It can show up as a referring URL at the destination. This information can be captured and tracked by the owner of the web site. It can be used to do some harmless statistical analysis – information the site of the keywords that are driving traffic. But it can also be used to build a user profile, which is probably what Google says it is protecting the users from.
But can they say that with a straight face, probably not. Their own documentation on SSL search does set the expectation that your keywords are not passed to the organic search result site owners.
Under most circumstances, when you use https://www.google.com your search terms are encrypted and are excluded from the referrer headers that are part of the request sent to the result site you visit. The landing site will still receive information that you are coming from Google, but not the query that was issued — namely, the host is still part of the referrer being passed.
This sounds pretty fair, until you get to the part where the text advertisers get the full referrer.
If you click on an ad on the results page, your browser will send a referrer that includes your query to the advertiser’s site. This provides a mechanism to the advertiser so that the advertiser can improve the relevancy of the ads that are presented to you.
So an advertiser can pay for the privilege of getting your referring keywords, but the organic search results don’t organically get that privilege. If user privacy was really the primary concern for the move wouldn’t they limit the keyword visibility to all destination sites? Seems that they would. In fact, they actually offer a service that does just that – encrypted.google.com. So why not adopt the same policy here? Alas, there’s no “privacy concern” based explanation for this.
The Secondary Keyword Market
Let’s fast forward, or rewind, or something. Let’s step away, yeah, step away. Let’s say there’s a secondary keyword marketplace out there, because there is one. It’s not huge, but it’s out there.
What’s this marketplace doing? It’s leveraging these referring URL keywords to sell advertising. It’s like the search advertising you see on your search results page, but it’s at one of the destination pages that you clicked on. These placements can be sold by the site owners in a variety of ways, you can buy them like adword placements, or you can pay a CPM or whatever.
Suffice it to say I know this market exists and I’m pretty sure it dilutes the dollars that are primarily going to Google. But how would we know for sure. What if we did an experiment with, oh . . . I don’t know . . . 10% of the traffic.
So what’s wrong here? Why is it anybody’s business if Google suppresses that referring URL information or not. It belongs to them, doesn’t it? They have the technology to control it. They’re not breaking the Internet are they? No one would complain about this activity if Google were not one of the dominant search engines.
I’m compelled to turn this around for a moment. If I had an opportunity to extend my dominance in the keyword marketplace and message it as a protection for the searchers visiting my search engine, would I do it? It sounds like a win-win to me. Would Tim Berners-Lee come bust me in the face for monkeying with so much web traffic? What about Bruce Schneier? Would he like this idea? He probably would. Maybe he’ll comment on it at some point with a much more nuanced argument.
I’m not sure that this half-measure to protect privacy and a full-measure to protect earnings can justify the potential destruction of that secondary keyword marketplace. Maybe this is Internet advertising marketplace defragmentation at its purest. But it’s Google do the defragging, once again.