Fellow Search Insider Gord Hotchkiss has been writing about search personalization for quitesometime. Gord has been preaching that the biggest (and most important) opportunity for innovation in the search space is around personalization. I agree that we'll see steady investment and advances in this area, but I'm less bullish than Gord on the prospects of personalized search to truly benefit the digital ecosystem.
For starters, there have been some compelling arguments around why search personalization may not be the cure for irrelevant results. These concerns center on limitations of search engines to decipher things like multiple users sharing a computer or the fickle and fleeting nature of human intent. Issues like these raise questions about the potential accuracy of personalized search. But I'm not betting against the likes of Google, MSN, and Yahoo when it comes to addressing these challenges.
So, rather than contemplate whether or not personalized search can deliver accurate results, I want to focus on whether or not personalized search can deliver valuable results. And what happens when search results become both accurate and valuable.
What do you consider a valuable search result? Finding the answer to your question? Tracking down the Web page you were looking for? Discovering content you didn't know existed? Perhaps all of the above at various times?
And there you have the conundrum that is search personalization. While it may someday be possible for search engines to deliver results tailored to a specific individual, there are times when that individual doesn't want to see results tailored to him or her. An obvious example is SEO practitioners who want to see what the general public sees atop the search rankings.
So why couldn't a search engine simply allow you to choose whether you're looking for an answer to a specific question, a Web site you've already visited, or content you're not sure exists? Yahoo was on to this with its Mindset beta, which allows users to self-select shopping vs. researching queries. And Ask has some nifty tools for refining your search.
Why haven't these features caught on? My guess here is that, at the end of the day, most people just don't know what they want. And, as much as I hate to bring any more attention to Lord Saatchi's misguided ramblings -- which were properly panned by Chris Copeland in a recent Search Insider column -- a subtle tweak to his thesis illuminates the point here: "People do not know what they want until a brilliant person [or search engine] shows them."
This certainly seems to be the approach Google is taking. Isn't that what Universal search is all about? Google now returns a set of search results it determines you will want with a mix of Web pages, video, news, etc.
In Gord's column for Search Engine Land, he outlines three ways search engines can "help define intent without depending on further information from the user:
·"They can look at your past history and learn more about you by what you have already done
·They can look at the context of the task you're currently engaged in, hoping that it will give some clues to what you're looking for
·And finally, if they know something about you and your social, geographic and demographic cohort, the engine can hope that there is a similarity of thinking within that cohort, at least when it comes to common interests and intent"
While all this is great when it comes to delivering accurate -- and even valuable -- search results, I think the end result is a series of self-fulfilling prophecies.
Let's work through an exercise to see how this plays out...
Jimmy searches for "crack corn" (and yes, he does care -- about the results).
For that query, most everyone in Jimmy's peer group was looking for the Eminem song (Site Group A) but Jimmy's past search history had him clicking on listings about his original tune (Site Group B) -- yes, even Jimmy is prone to a little vanity search every now and then. This time, however, Jimmy submitted the query immediately after visiting a gardening site, so the engine knew to display a site related to seed corn (Site Group C).
Seems like a victory for personalized search -- the result is both accurate and valuable. So Jimmy clicks on a site from Group C. And the scenario gets played out hundreds of times over the years with Jimmy clicking between Site Groups A, B, and C depending on his motive and mood.
At some point these site groups will be the only options presented to Jimmy. And as Jimmy chooses from these limited set of results, it just serves to reinforce his behavior. Sooner or later, Jimmy will not realize that other choices exist. And the same will go for Jimmy's peer group. In true Darwinian fashion, will all other sites eventually go extinct -- or at least fall to the bottom of the SERPs? Will it be survival of the fittest, er... personalized in cyberspace?
And let's suppose all the clicks on Eminem lyrics from Jimmy and his peer group label them as rap music fans. Now rap starts to pervade results for other queries. It is not unreasonable to think that this could create a rap fan where one didn't exist previously. If you think rap music being blamed for violence is crazy, imagine what happens when the search engines start being blamed for rap music.
Along these lines, here's another extreme example -- what happens when a new street drug called "crack corn" becomes popular and all of Jimmy's friends and peers begin to click on results about how to make it? Is it only a matter of time before those results pervade Jimmy's SERPs and he gets hooked on "crack corn"?
OK, am I over-exaggerating here? Probably. But am I on to something? Possibly. So how will I take it if Gord doesn't respond? Personally.