Mind Reading for Better Search
If you're buying for your 50- to 64-year-old "urban hipster" mom for Mother's Day, Gift Finder suggests an Elie Tahari turquoise knit Carol shirt from Bluefly. For a 25-year-old "adventurer/traveler" cousin who did something nice for you, it suggests the Women's Mountain Hardwear Poodle Hoody from Backcountry Outlet as a thank-you gift. Enter an age, a personality type, the recipient's relationship to you and the occasion and Gift Finder will suggest the presents.
What's brewing here is more than just a new way to buy gifts. What's brewing is a piece of the push for the most personalized search engine experience possible. Because whether they're buying a gift, conducting online research or just looking to play an online game, what nearly all searchers want to get to is a small number of Web sites that best approximate the exact thing they're looking for. And since everyone is looking for something else, the more personalized you can make the search, the better.
The problem being, of course, that the keywords typed into search engines are often vague or ambiguous. So giving a precise, personally meaningful response can be very hard. Is a search for "dog" a search for dog food or for a new pet? Is a search for "shoes" for Nikes or Armanis? That's why, either through directly asking searchers to clarify what they want to see, or through other means, the next phase in search engines is going beyond well-targeted results. It's about learning to read a searcher's mind.
And mind-reading search engines aren't just relevant for the search engines themselves. They also have tremendous ramifications for the businesses that market through them. Because, after all, mind-reading search engines can mean mind-reading search engine marketers, too.
Letting searchers see what they want to see. For now, the two ways of resolving ambiguity both involve, more or less, the searcher manually guiding the search engine to give her the results she wants to see. One way involves the searcher choosing from sites she has seen before; the second involves the searcher telling the search engine more specific information about her search.
The first way is search histories. The major premise behind search histories is that if a searcher deemed a result relevant once, there's a good chance that she was happy with the Web site and would want to visit it again. So a search for a keyword really might not be for all the sites on a topic - it might be for a specific site whose URL the user forgot, and is using the search engine to remember.
Or, even if the searcher forgot about that site altogether, chances are the keyword she's using a second, third or 40th time has the same meaning it did in the past - and the sight she visited earlier (but forgot about) will give her the information she wants. So the search engine provides a history of the searches the searcher has already done, including results she has clicked on before for the keyword she just entered.
Google's My Search History is one such search-history product (Yahoo, A9 and many other engines have excellent search history products, too). In My Search History, searchers can see the Google listings for all sites that they've visited through Google on a given keyword, from the time they signed up for Google services through the time of their current search. Then they can use those history listings as they would use the regular Google results. Searchers also can see their search histories by date.
The other method search engines use to learn, more specifically, what a searcher wants is letting the searcher fill in the information herself. Gift Finder would be an excellent example here.
Along with Gift Finder, sites that make searchers add some specificity themselves might include the narrow-focus search engines, like the shopping engines (of which Yahoo Shopping is one) and databases like IMDB (Internet Movie Database). After all, when was the last time someone entered the keyword "rocky" into IMDB hoping to find mountain-climbing tours? In a Google search, the same keyword would be more ambiguous.
For now, one aspect that all these methods have in common is that a searcher's manual input seems like the engines' most powerful way of knowing what a searcher really means. For the future, look forward to all search engines being intelligent enough to know what you mean on ambiguous terms, based on your prior search history (and perhaps other activity that's stored on your computer) and organizing results accordingly.
Take a search for the ambiguous keyword "hog." If you have a strong history of searching for motorcycles, the keyword "hog" is most likely a search for a site like Harley-Davidson.com. If you regularly search on farming keywords, the search for "hog" might actually be for a site that's related to pigs.
With enough intelligence, a search engine could present an entirely different results-ordering based on what's relevant to you. For the person who's been spending her time looking at motorcycles, motorcycle sites will appear first, pig sites second. For the searcher who spends all her search engine time looking for pig feed, that order would be reversed.
And that kind of sophistication could even, eventually, break down similar search results by taste (based on search history) - the way Amazon.com already does for suggesting products.
What targeted search gets you, what it doesn't get you. The most obvious way that better-targeted search results are great for advertisers is that, simply, they can be well-targeted. As it is, a search engine may be the best-targeted marketing forum around: everyone there is looking for a very specific thing - including specific products and services.
Compare this to, say, TV advertising, which can never be quite as targeted, because people don't watch television to learn about specific products. And, to approach your already-interested audience, all you need to do is purchase an ad on the right keywords. Any attempts to make search results more specific will only increase the power of targeting specificity that search marketing already offers.
Rather than bidding on a keyword, for example, there are ever-expanding ways to better target users, based on demographic and interest as determined by the search engines themselves. That way, to come back to the earlier example, pig feed wholesalers and Harley Davidson retailers wouldn't have to bid against each other on the same keywords (as they do now if they both want to bid on the term "hog"), because the search engine could determine, on its own, who's searching for a term and why.
Already, MSN Search has created Ad Center, its pay-per-click advertising program, which is able to target searchers based on gender, age, lifestyle category, wealth and geography. Locally based advertising is also an integral part of local search features on Google, Yahoo, MSN Search and other engines.
In addition to the benefits that the future of more targeted search results offer, more targeted search also frees up marketing time and energy that gets lost because, currently, searches aren't specific enough. Which is why only half the art of today's search engine marketing is driving the right people to your site; the other half is making sure that the wrong visitors don't visit.
Ad copy has to be specific enough to indicate exactly what you're selling, and long lists of negative keywords - keywords that you specifically don't want to advertise on for your campaign, so as not to get charged for useless click-throughs - need to be generated. The company that sells pig feed doesn't want to pay for mistaken site visits from motorcycle enthusiasts who came through a PPC ad on the keyword "hog." (This might explain why neither motorcycle companies nor pig farming sites advertise on the keyword "hog" in Google, at the time of this writing.)
The more specific the search engines are with their results, the less energy you need to put into driving bad traffic away - and the more you can dedicate to attracting the people you do want.
As search engines continue to produce organic results that are more personally targeted, search engine marketers will be able to piggyback on their success.
Good cookies aren't enough: a cautionary last word. One potential snag for the search engines - and the search engine marketers - to be aware of is that people change. A given search history doesn't always indicate future activity. People go on diets, they move, they switch careers - any number of factors could make a consumer change her buying habits and search interests.
Analysis of searcher cookies can help with many changes, like calculating a searcher's age and targeting information accordingly as the searcher ages. But there will always be a degree of unpredictability that needs to be watched out for (and, sometimes, embraced).
The most precise way of determining a searcher's interest at a precise time will almost always be, simply, creating more effective ways to let her tell you what she'd like to see. The fact that Gift Finder does that is what makes it so exciting and what paints such an exciting picture of the future of search.