Sunday, February 10, 2008

The Battle for the Best Buggy Whip

Google and Microsoft are in the news right now, Redmond having dropped $40B in market cap in response to their bid to buy up the Yahooligans for another $45B in order to produce a "credible number two" to Google's far-and-away number one in the online advertising space.

They're all fighting to be the best buggywhip manufacturer in town, particularly when it comes to the mobile space.

Horizontal search, which Google unquestionably excels at, has its uses, but those uses become more and more limited as information (much of it useless) becomes increasingly "available" (i.e. findable). Simply finding a chunk of information that matches your inevitably incomplete attempts at what you think it might look like isn't usually enough, not when the matching process is driven by how many randomly-chosen others point at this chunk, without regard to who's looking for it.

A couple of concrete examples: if you just know me by first and last name, you'll have one hell of a time finding anything that's actually about me with Google−there's a guy who's worked for Reuters for years with exactly the same name, who gets pointed at a lot more than I do (I'm working on this, I assure you). Most of what you'll turn up is about him, unless you know more about me than that.

I live in Santa Cruz. Santa Cruz, California. When I try to find out something about Santa Cruz,I frequently find myself wading through a stuff about Santa Cruz County, Arizona, Santa Cruz de la Sierra, Bolivia, Santa Cruz de Tenerife in the Canary Islands, Santa Cruz Island (aka Indefatigable Island) in the Galapagos archipelago, and so on. These are all places, which−while I'm sure they're quite interesting, kinda−are not locales that I'm likely to be looking for information about. Because I'm me, but Google doesn't really know that (although they have enough information to know it better than they do).

I travel a lot. Type "plane flight to Brussels" into Google, and you'll certainly turn up pointers to places that could get you one. Quite a few of them, in fact, both of the "sponsored" (i.e. AdWords ads) as well as of the "organic" (i.e. directly generated by the search) variety. Now, your work really starts. Out of the, oh, three-quarters of a million responses there, which one gets you the cheapest ticket on the nicest airline leaving (or arriving) at the time you want, and so on...?

Thus, the limitations of the horizontal approach to both search and advertising. Luckily, in this particular instance, there are vertical search engines, like Kayak and SideStep, effectively meta-search engines, which go out and troll every travel and airline site they're aware of (and they're aware of quite a few more than I am) and collect the results, allowing me to sort and filter them in a pile of ways. Then, there's a whole other site which can tell me where the best and worst seats are on a given model of aircraft, and ones that can give me suggestions (and reviews) of hotels in Brussels (and let me book reservations), and ones that can provide some ideas of things to do, or places to eat while I'm in Brussels...

But we're back to horizontal again. Tons of stuff to wade through. Because it's not about me.

What if, my putting an entry into my calendar that indicated I was going to be in Brussels for a week made certain information about me available: when I'm going, my history (and ratings) of my past hotel stays in other places, my various travel affiliations (frequent flier clubs, etc.), the kinds of food I like or the kinds of things I like to do (based on an "interests" list, or restaurant reviews), and so on...? What if that information enabled the right kind of vendors to come looking for me? I'd need to have some way of matching potential responses against the information I'd provided−but if a response came in from, say, a restaurant that had overall poor (or no) ratings, or one that didn't match well with my history, it'd be "ranked" lower than responses which were a good match, or which had many positive ratings.

Since responses would come in after my "expression" of a concrete need, there'd be a good incentive for prospective vendors to "cut me a deal". Knowing (from my history) that I like Japanese food, if there were a good sushi bar in Brussels, they'd probably be interested in offering me a free bottle of sake to come in, and−if their reviews were good−the odds are that I'd take them up on the offer. Moreover, there'd be an incentive for them to offer me something in exchange for reviewing them on an independent site, information which could feed into other people's future "findability".

Because the new model isn't going to be about searching: it's going to be about allowing yourself (or aspects of yourself, more properly) to be "found" by prospective vendors and to allow them the chance of being "found" by you. Advertising is going to be a lot more targeted, much more so than even AdWords can manage, since its responses suffer from all the limitations of horizontal search in general, and then some.

Simple search-generated ads are like billboards: they appear in front of your eye by relative happenstance, and if they happen to actually be of interest to you, it's pretty much by chance. While there's a certain probability that I'll be driving past a billboard for a McDonald's at a moment when I'd both be interested in a Big Mac and have the time to stop for one, it's not all that large. And when I get a thousand ads for plane tickets to Brussels, the odds of any given one being the one I pick are looking like the blllboard's chances of getting my market share.

This won't apply only to travel, eating out, and so on: it can be extended to things like music, movies, and almost anything that can be sold, whether a good or service. By looking at my buying history (what I'm willing to share of it, via reviews or otherwise), my interests (again, what I make available, and there's already a lot on FaceBook, Plaxo, etc.), and by comparing that information to similar information from others, you can identify potential "outliers" in my constellation of interests. Amazon already does this with their "Recommendations", and not too horribly−they're frequently recommending CDs or horror movies to me that I actually already own−but it can be a lot better. They could know (from the information I share about my music collection) what I already own. You see some interesting progress in the direction of music from sites like, the Music Genome Project, iLike, and others.

As I said at the outset, this approach (some aspects of which Doc Searls has referred to as "vendor relationship management") will be increasingly important as our access to information becomes increasingly mobile. Because of factors like small screen size, there's not going to be a lot of room on mobile and converged devices for ads. Nor, because of the I-need-to-get-it-done-now, task-oriented nature of the use model of such devices, is there going to be a lot of patience for advertising that isn't very specifically targeted at what I'm likely to be interested in right now. What would be nice would be something akin to an email or SMS notification on my phone, at lunchtime on a day when I had some free time, for places (in a specific radius of where I am right now, as provided from my GPS information, say) which wanted to show me just how good their miso ramen is, with a tag on each showing the number and distribution of reviews the places had gotten.

So, good luck to Microsoft and Google. While they're duking it out over the billboard rights on the Information Highway, I suspect someone else is going to be sneaking in with a better idea.