Five issues with Google’s Personalized Search – SEO isn’t dead

by Andreas on December 31, 2009

Earlier this month Google announced that even Google searchers who are not logged into their accounts will now be seeing personalized search results. Up to this point you needed to be logged your Google account to see customized rankings tailored to your search history.

Google has taken the initiative to let us all know that all of this is harmless and only serves to improve the quality of the search results:

Among the YouTube comments as well as various SEO blogs the usual reactions of doom and gloom have started to appear: SEO is dead. Rankings are inaccurate. Google knows too much. Etc.

I don’t necessarily agree with this. This has happened before with the introduction of Google Universal Search, for example. And it didn’t mean the end of the world then either. I think we need to take a step back and analyze some issues about this new trend before throwing in the towel.

Issue #1: Personalized search results should not be the standard

The examples Google uses here are fairly constructed. It’s always seems to be the same simplistic idea: One person is searching for “golf” as the sport; the other is looking for the VW car model. In this case Google’s personalize search would easily help segregate out results that are less compelling. Great example, but how many searches actually fall into this model? Can Google truly use prior search history to determine which results should be displayed for the current search?

Let me be more specific: Today somebody searches for “kohler bath fixtures” and tomorrow for “real estate san jose”. What can be learned from this information? Not a lot. Even for a human – compared to a non-human algorithm – it would be extremely difficult to establish a valid search profile around this history. Is somebody looking to purchase a property in San Jose? Or does somebody want to sell their San Jose house? Is that person looking to buy Kohler fixtures? To repair a fixture? Or this just research for a spouse? Or could the fixture have something to do with the property in San Jose?

And that’s my main point there: How does the first query influence the second one? I.e. how many of the search queries are actually qualified to be used in personalizing the results? I personally don’t have access to these statistics, but would have to guess that the percentage must reside in the single digits in terms of a valid cause and effect. So, I don’t really foresee the end of SEO here.

Issue #2: Personalized search results narrow my perspective

If I often click on New York Times news results that means that NYT results will be pushed up in my search result pages in the future. I’m sending a signal to Google that this web site is relevant for me. But do I actually want that? If I’m searching for political themes, wouldn’t it make sense to expand my view to include CNN and FOX perspective rather than push only what I think is relevant?

Of course, from an efficiency standpoint it makes sense that an often-clicked web site gets better rankings and helps me find what I’m looking for quickly. But ultimately this won’t diversify my perspective of the web content.

Issue #3: personalized search is a usability problem

Many of my search queries have nothing to do with one another. Sure, that might have to do with the fact that I search for clients or projects that aren’t related. This might also have to do with the fact that I’m a father, husband, entrepreneur, bicyclist, musician, etc. I’m not just wearing one hat.

Let’s say I first search for “wedding photographers” because one of our clients requires some competitive research done in this sector. Then I search for “buying dresses” because my daughter needs a new outfit for a dinner party. Although there is none, the obvious connection here would be “matrimony” and the assumption could be that I’m looking for a wedding dress. In this case, instead of being beneficial, it would drive me nuts to see all “buying dresses” results for my daughter come up as wedding gear.

In other cases, especially for queries that might serve local results, this personalizing might make a lot of sense. So I would love to see the following implemented:

1. Google has to tell me which factor(s) the ranking adjustment is based on
2. I should be able to prevent or override the ranking modification

Let me be more specific: After my search for “buying dresses” Google could ask me something like “Did you mean search results relating to the concept <wedding> or general results?” Then I could clearly answer this question and determine the influence of these factors. Otherwise I might end up concluding that Google isn’t a good search engine.

Issue #4: Personalized results don’t help me educate myself

The idea isn’t a bad one: I type in “golf” and Google understand immediately that I mean the car and not the sport because I have been searching for “vw diesel” as well as “used diesel golf” in the past months.

But this doesn’t help me learn how to make my Google queries more effective. Google is lightening my workload by doing this for me but, at the same time, taking away control and the necessity to interact with a search engine. An effective relationship between the user and the search engine needs to be bilateral.

Issue#5: Personalized results are not so completely different

For those who insist that search engine results are becoming worthless, in a sense they are right because a ranking no longer is something that can be used as a concrete benchmark. However, everything is relative. A result that comes up #1 for one searcher will not end up #50 for another user just because of the influence of personalized search.

Personalized search is more a question of a shift within the rankings. Whoever has a top ranking for a certain keyword phrase will keep that top ranking. It might not always be precisely #2 but it will continue to hover in that vicinity.

What is the take-away?

I predict that personalizing within search will continue to increase in importance. For certain searches this might be a good thing, for other it might be counterproductive.
My concern is rather that Google is making assumptions about me and my search behavior. And by acting accordingly and modifying the search results page, this may have a negative effect on my search experience. Google results are based on statistical analyses and projections, so mistakes will be made.

I can see a similar trend with Amazon.com where I have ordered a medical manual for an acquaintance in the past and I still receive screen-fulls of ads for the latest medical releases that never interested me and no longer interest this acquaintance either. And Amazon should be learning from me that these books don’t interest me since I have never clicked on one of them.

So, let’s just keep our radar up as we follow all of the trends Google has in store for all of us in 2010.

Andreas Mueller is the President/Founder of Bloofusion.

All posts from | Follow @bloofusion on Twitter

Leave a Comment

Previous post:

Next post: