Every shot must count.
You will, no doubt, target your online ads contextually, by placing them on car and home improvement sites and wherever else makes sense. You'll overlay this with behavioral targeting, and deliver your message to people who research garage door openers online, or do whatever else you determine shows potential interest in your product. Alone and together, these approaches will certainly narrow the field.
As much as we'd like to claim otherwise, however, there will still be considerable waste: you will, for example, reach car owners without garages, people without the means to afford your luxury item's premium price, and stuffy "old money" folks who won't throw out working stuff just to have the latest thing. Unless you are lucky enough to have the more detailed kinds of information that can be gathered with registration pages, greater precision will be elusive using online methods alone.
The Internet, by itself, can tell you only what it gleans from users' online actions, or by asking users online questions (and having them answered). To date, however, except when something valuable is offered in return (be it product or information), and the circumstances seem to warrant giving up the particular information asked for, the Internet has proved to be a rather poor venue for getting the answers to detailed questions - a circumstance exacerbated by frequent abuses of those who do give them.
But perhaps there is another source of information. Traditional research is less intrusive than online questioning, and reality checks are infinitely easier. To take an extreme example, while the Internet cannot tell you that males of a certain age and income, living in a particular ZIP code, are exponentially more likely to own a basset hound and eat smoked tofu, while also being measurably more inclined to buy small red two-seater sports cars, traditional research, if such people exist, has long since found them, and it knows where they live.
What, then, if you were to jump the fence, grab some of the data amassed by traditional researchers, and add it to your pot? Suddenly you'd be operating in a different league. No longer would you be dealing with a basic undifferentiated stew of information, but one in which every ingredient could be distinguished. Suddenly you'd have the data you needed, in the form you need it, to deliver your message to precisely the people most likely to be receptive to it.
The folks at USATODAY.com are already doing it. They use the softest of "soft registration" to gather their readers' ZIP codes, and then overlay traditional market research firm Claritas' extraordinarily detailed segmented demographic information about those ZIP codes onto the online data that they already have about their individual readers.
In that single stroke, USATODAY.com transforms a simple piece of anonymous information, which most people provide willingly, into a powerful tool for maximizing advertiser value. USATODAY.com's advertisers can now do what no print advertisers, restricted by the economic practicalities of long print runs, nor those relying solely on online methods, can do: using the most basic information, they can target narrow sections of the community with extreme precision.
Combine, then, the online information you already have with traditional research's precise demographic information, and you will be able to zero in, not only on those with the cash to buy your product, but with the purchasing characteristics that would make them susceptible to its charms, in the first place.
One ZIP code might, for instance, be full of rich retirees who have stopped spending except when things break: no use to you at all. Another ZIP code, however, might have vast swathes of suburban developments with two-car garages facing directly onto the street, inhabited by well-paid and highly enthusiastic buyers of marginally useful status symbols: these (and virtually every large city in the United States are filled with such places) are, of course, your perfect target.
Take the information that traditional researchers can provide, combine that traditional information with online behavioral and contextual targeting, and you will be able to pinpoint an audience for your ads that no single method could dream of reaching in isolation; it is, I suspect, likely that your only problem will be to finance enough door openers to keep up with demand.
When would this work? It seems to me that just about any branded consumer product that is or could be advertised on the Web might benefit. It might not work as well for something more generic - although if, for example, pork producers decided to advertise online they could use it to avoid pitching to vegetarians and those whose religion forbids consumption of their product - something that the managers of their recent poster campaign wretchedly failed to do when they plastered Pork, the other White Meat on every phone booth on Manhattan's West 47th Street, which the most basic research would have told them is the epicenter of New York's heavily Orthodox Jewish jewelry trade.
This is not a one-size-fits-all world, of course. For many advertisers, the methods available online will provide enough precision for their purposes, and the extra expense of overlaying traditional data may not be for all. But as the line creeps upwards on the brand identity/product distinctiveness graph, so too will the value to the advertiser increase in precision offered by combining online and traditional data - a value that cannot be matched by either targeting method on its own.