Back in the old days, things were simple: you entered a search query, it returned a page of results, and you could rest easy in the knowledge that all across the country, other people doing the same exact search were seeing the same exact results page. SEOs did painstaking work to build rankings for their clients, and the results of their labor could be seen clearly by anyone, anywhere.
Search engines—Google in particular—have of course become much more sophisticated since then. Personalized search was introduced to signed-in Google users in mid-2005 and sought to improve the user experience by tailoring results based on search history. By the end of 2009, personalized search became the norm for everyone using the search engine, regardless of whether they were logged in.
Today, search engines take into account a wide variety of factors in tailoring search results to each individual user. It’s no longer a simple matter of search history—now, everything from your locality (city, state) to personal endorsements (Likes, tweets, +1s) to social connections (what sites people in your network visit) can influence what your search results are for a given query.
Given this continuing trend towards greater search personalization over the years, it should have come as no surprise when search competitor, DuckDuckGo, publicized the results of an experiment showing that the exact same search query could bring up wildly different results, even among users who are not logged into Google.
From the article:
“I didn’t expect so much variation for signed out US users,” DuckDuckGo’s founder Gabriel Weinberg told TPM in an email. “You expect some personalization when you’re signed in. But if you’re signed out or in incognito mode, you expect to get the ‘regular results.’ What we found is there are no more ‘regular results’ on Google.”
The fact that DuckDuckGo decided to conduct this test and publicize the results in late 2012, almost three years after Google announced the launch of universal search personalization, is telling. It suggests that perhaps we’re still somewhat unclear on the extent to which personalized search has become the “norm.” While we might reasonably expect a search for “pizza delivery” to yield personalized results, the same may not be true for a query like “barack obama.”
What is also telling is the fact that the article presents increased personalization as a bad thing—bad for SEOs and bad for users. As a competitor to Google, DuckDuckGo has an obvious reason to go for this angle. But are they on to something? Will increasingly personalized search engine results create filter bubbles that isolate us from other viewpoints—or are they merely the logical extension of what a search engine is meant to do in the first place?
Why Personalized Search
It’s important to remember what the goal of a search engine is: to provide the user with the information they’re looking for as quickly and efficiently as possible. If I enter queries into a search engine and consistently get irrelevant results, I could (hypothetically) get fed up and decide to switch to another search engine. And with that in mind, the move towards personalization makes total sense—personalized results are more relevant to me.
For the most part, we accept this wisdom. When someone does a search for “sports bar,” they aren’t looking for a place that’s 1300 miles away—they want to be presented with a set of realistic (nearby) options, as well as any relevant data which might help them decide where to go (ratings, photos, etc).
Similarly, if I share an article about the Mars curiosity rover on one of my social networks, I might find it incredibly convenient that this article gets a boost in my personalized search results, especially if I decide I want to re-read it or show it to a friend a couple months down the road.
But what about search queries related to hot-button topics, such as “contraception” or “gun control”? This is where we start to see concern regarding the implications of a highly personalized set of search results—one that may effectively insulate a user from exposure to a wide range of viewpoints on a particular issue.
This would be a valid point, except that I wonder if perhaps it’s confusing cause and effect. For example, let’s say there is a hypothetical person out there who really trusts CNN.com: he has CNN.com as his home page; he visits CNN.com multiple times during nearly every browser session; he frequently clicks on links to CNN.com news stories shared by friends within his social networks. And, when he does a search for something, he will consistently click on CNN.com links—even if it means scrolling down to the fifth or sixth search result. Given that data, it seems logical that Google would tailor his personal results to give slightly more weight to CNN.com—this makes Google more valuable to him and thus helps to ensure his continued use of the Google in the future. In this sense, Google’s personalized results are actually a reflection of an existing user bias—not the cause of that bias.
At the same time, it’s likely that the DuckDuckGo folks are overstating the extent to which the search results might reflect a partisan bias. I haven’t done a large-scale test myself, but a quick search for “obama” on DuckDuckGo yields top 3 results that are identical to those on Google. After the top 3, the results differ, but still include links to biography.com, washingtontimes.com and chicagotribune.com. So that’s 6/10 results that are exactly the same, albeit slightly reordered. Here are the key differences among the remaining four results on each page:
- Top 10 DuckDuckGo results include a link to “obama.com,” which 301-redirects me to a page where I can donate money to the Obama campaign. Regardless of your political leanings, this doesn’t strike me as a particularly useful result based on the query.
- Top 10 DuckDuckGo results also include a link to change.gov, which one can see is an incredibly out-of-date website—in fact, it hosts a splash page that urges users to go to whitehouse.gov instead.
- Top 10 Google results include a link to www.nydailynews.com, a news source that is local to me based on my location.
- Top 10 Google results include links to the official Barack Obama Facebook and Twitter accounts.
Overall, the Google results seem not more partisan, but more useful. And when I have “personalized results” enabled, the last two results on the page are the only ones that change—to links that were recently shared within my Google+ circles. When you consider it, this is also useful: it allows me to quickly and easily access content that people I trust deem interesting or important.
Is Personalization Good or Bad For SEO?
People have heralded most every major change to search engine algorithms over the years as “the death of SEO,” and the shift towards more personalized results is no exception.
But personalization hasn’t killed SEO. As SEOmoz pointed out back in 2009 when “personalization for all” first launched, the basic process of optimizing a site for search engines has largely remained unchanged: create pages that are structurally accessible to search engines, target those pages with valuable keywords, and build comprehensive, high-quality content around those keywords that’s worthy of endorsements. Then, make an effort earn those endorsements—whether they be links or Likes.
In some ways, personalization has made the job of an SEO much easier. Rather than having to get a client’s website ranking #1 worldwide for “cardiologist” or “divorce lawyer” (which would probably be incredibly difficult), we can focus instead on a more targeted local search campaign. That’s not to say it won’t still be challenging or competitive, but a smaller target group necessarily means less competitors.
In other ways, personalization has also complicated things. SEOs will now need to be proficient in social media marketing tactics, or at least work closely with someone who is. With personalized search, social factors become an increasingly relevant ranking factor, and thus the line between search and social continues to blur.
Love it or hate it, search personalization is here to stay. And while we may need to adapt to this by broadening our skill set, the basic principles of SEO are still useful and still apply.