Since Google launched Instant, there have been numerous claims that Instant has a bias towards brands. In October, for instance, Siddharth Shah of Efficient Frontier Insights observed that "of the 26 letters in the alphabet, 21 have brands as the first suggestions."
Based on this, he suggested that Google Instant is "going to make brand key words more expensive, increase impression volumes by 30% - 40%."
According to Google's Amit Singhal, who heads up the search behemoth's search quality team, Google Instant has no brand bias. He told Fast Company:
We didn't want to introduce any bias into the mathematical modeling--our modeling is predicting, given a letter, what's the probability of completion.
Most people typing A are seeing Amazon, but that probability is predicting that most people typing A are going to complete to Amazon. If you type T, most people typing T will go to Target. That's the probability model. If you add R to it ("Tr"), most people are looking for a translation system. It's actually just pure mathematical modeling.
His explanation seems reasonable. Brands are popular, and there's no reason to doubt that a lot of searches are brand-related. But far more questionable is Singhal's comment that "What we do at Google and what we've done for years is to not inject any subjectivity into these algorithms."
It's a curious statement that doesn't appear to be 100% accurate. Google's Vince update, for instance, was widely seen as a boon to big brands. Although Google insists that this update had less to do with brands than trustworthiness, the results seem speak for themselves. When coupled with Google's other features and experiments that tend to promote brands and Eric Schmidt's past "cesspool" comments, those who believe Google is favoring brands, even if subtly, have a decent amount of ammunition.
What's far less subtle: Google's favoritism of its own properties. On that front, it's hard to reconsile Singhal's claim that "what we've done for years is to not inject any subjectivity into these algorithms [emphasis mine]" with Marissa Mayer's admission that Google promotes its own properties. Certainly, changing particular sets of SERPs to display Google links first, overriding what Google's algorithm produces naturally, is an inherently subjective act that is based on the assumption that Google's own properties are superior, even if consumers haven't yet voted with their mice.
At the end of the day, one thing is clear: Google is increasingly going to have to face scrutiny over how it ranks sites, particularly its own. While there is an argument to be made that "objectivity" and "neutrality" are overrated anyway, it would probably be wise for Google to stop implying that this is what it always delivers.



Reader comments (6)
10:49AM on 14th December 2010
Google instant will always have a brand bias - to the bigger brands I may be looking for Argos but typing in A will bring up Amazon simply because of the size of the brand.
How much rests with the searcher though? If they just see Amazon and immediately click then this will drive yet more traffic to the Amazon site and further improve their positions with Google instant. If on the other hand they are more savvy then we would hope to see this 'bias' slowly disappearing.
Director at Digital Sparkle
4:29PM on 14th December 2010
Um, typing in A brings up Argos first for most people.
5:39AM on 15th December 2010
At the end of the day people want to see brands, if Google can semantically match a brand to what it "thinks" a user is looking for then why not?
If you search Go as in (GOD) you'll actually see GoDaddy first.... how biased is that...?
Phil
6:03AM on 15th December 2010
well we all are niticing this biasing factor of Google in case of Amazon but at some time it seems like Amazon have a larger trust worthy group of people too. ,
11:14AM on 15th December 2010
Erm, I just tried all 26 letters, and got 26 brand names! It's different for each person, based on location, browser, history, etc
12:09AM on 16th December 2010
I still have a hard time believing that there is no brand bias with Google Instant. If its truly a probablistic prediction that Google is making then one is to believe that humans search in brands all the time. I dont have the data but that seems unlikely to me. From an advertising standpoint the brand-centricity of recommendations has the following effect: It forces advertisers to bid on other branded keywords to get "lost" traffic. Think about someone typing "travel". You type in tr and you see a travelocity ad. If I am priceline I will now have to bid on travelocity ad to get back some of my lost traffic on the travel keyword. The increased competition should make travelocity more expensive. I dont expect cpcs on brands to go up right away but there should be a long term effect. .If you do an analysis AFTER Google Instant, you will find that brand keywords get more impressions and you might conclude that brand searches are indeed more common. But that will be a flawed conclusion as the data is biased by Google Instant's recommendations. So it will be tricky to analyze if people search mostly by brands. Only Google has the full answer to that question.
Log in to post a comment