Commentary

AOL's Database Of Good Intentions

To paraphrase an idiom a la John Battelle, "The road to hell is paved with the database of intentions."

AOL has been taking it on the chin and issuing mea culpas for releasing 20 million search records from about 660,000 of its users in the name of research. You can't identify the users, as they are listed by anonymous IDs, but searches are our fingerprints. The New York Times traced search queries back to a 62-year-old in Georgia, and Wired News identified a 14-year-old. It was an innocent misstep, according to Battelle, Search Engine Watch, and other observers. Given the privacy implications, AOL recanted and took down the data.

Yet copies of AOL's database circulated in downloadable format (a 436 megabyte file, or 2 gigabytes when unzipped) and as online databases. In those 2 gigs, there are quite a few lessons and reminders for marketers; below is an abbreviated list. For the research for this column, I mostly used the AOL Log Search at czern.homeip.net/aolsearch/ (thanks to former colleague Erik Mednis for the link; another option is aolsearchdatabase.com).

advertisement

advertisement

One caveat: if you review the IDs or queries presented below, you will encounter terms that are pornographic, amoral, or perhaps offensive in any number of ways. Keep this in mind before searching the database (and complaining to this columnist or his publisher).

Now, here's what you can learn from 660,000 searchers:

To search is to bare your soul. The search engine is the new confessional, and you don't even need to kneel. One woman (user 100906) asked, "should i get back with my divorced husband." Then there was the woman who kept trying to refine her search, going from "business is killing marriage" to "wife does books wants a divorce but husband is 100 [percent] shareholder" to "wife wants a divorce and husband is an alcoholic business owner" (the truth always comes out with enough searches). It's not just women conducting such searches, though they do seem to be more expressive. When men are searching for terms related to wives, it's usually pornographic or adulterous.

No one can spell. One user (3390858) who searched for "mensa" also searched for "wine ametuer." However long the long tail of search really is, multiply it by around ten to account for typos and English speakers' unfamiliarity with their native tongue.

The search box is for direct navigation. Over the years, there have been some attempts, such as with the company RealNames, to add greater functionality to the Web site address bar (or navigation bar) in the browser. Still, that bar felt too confining; whatever filled that box generally started with "http." The field in search engines or search toolbars, however, is a blank slate. You can do anything you want with it, and search engines keep getting better at making magic happen no matter what you enter. Type a misspelled word, and it offers a correct spelling. Type an address, and a map appears. Type in a URL, and it's just one more click to get you where you want to go. Perhaps in time, the navigation bar will become obsolete.

The better mousetrap people need is a health search engine. If you look at any searcher who enters more than ten unique searches, at least one will likely be health-related. People are trying to take care of themselves, their kids, and their parents, and the search records can be heart-wrenching, such as one string from a woman (17239996) who gradually became convinced that one of her twin babies had autism. Anecdotally, people searching for health information repeat and refine those searches far more than they do for other queries. There's limitless opportunity to help searchers here.

The next best mousetrap: A vertical search engine for religious queries, which also appear in a vast number of searchers' query lists.

The other best mousetrap, and potentially the most lucrative: A vertical search engine for pornography.

Behavioral targeting and search are a marriage waiting to happen. Right now the two are more in the courtship phase. Even cursory glances at the data show how people searching for, say, "free credit reports," aren't just looking for credit reports, but they'll have other needs that they're trying to meet (e.g., moving, pregnancy, career change), and behavioral targeting can help marketers determine just what needs matter to which users. Expect behaviorally targeted search marketing to eventually become common practice.

Search engines are the new reverse dictionaries. When I was ten years old, I inherited my oldest brother Joel's reverse dictionary, which quickly became one of my favorite books (I was one of a few kids who preferred collecting reference books to comic books; it didn't do wonders for my social life). It struck me, reviewing these chronicles of searches, how search engines have become the best-selling reverse dictionaries of all time. Consider the searcher (7005) who entered, "game show host died in plane crash 2006." The searcher gives the definition, and the engine provides the entry. It's also reminiscent of the game show "Jeopardy!" With vertical search, you can even name the category in advance (the main search engines serve as that catch-all, "potent potables").

My reverse dictionary has been gathering dust for years, but I still keep shelves full of reference books within easy reach at home and at the office. They offer one benefit: my searches in those books won't ever be mentioned as pithy asides in someone else's column. Then again, they won't do me any good when I have searches like "game show host died in plane crash" or (heaven forbid) "business is killing marriage." I don't anticipate my search engines to gather dust anytime soon.

Next story loading loading..