"There's a story brewing about a next very big business it's building—one that competes with one of Yahoo's flagship ad products and would kill us."This entails the clever use of what in the 1980's would be called a feedback loop - in other words, the concept has been around a long time. It's not any easier than it used to be to collect postal mail addresses; the only step you skip is the data entry.
"As the marketer's product is sold, the retailer asks the consumer for their email address, phone number, or home address."
OK. No. First, you'd have to give the same email or phone number to the company marketer as you use on facebook - and many people avoid giving physical address and phone information to facebook. All that is necessary for this to be less powerful than one would like? The fact that some people don't. Will marketers have enough sophistication to increase the error inherent in the analysis to accomodate accurate understandings of peoples' deliberate misinformation?
Second, there is already a strong way to close the loop: you offer frequency and loyalty programs (and cohesion and community).
Third, even if you are sophisticated in handling unknown data and are getting a higher participation rate than loyalty programs, you don't know whether seeing the ad and clicking through is why the person bought the thing, or whether it was in-store merchandising or price comparison, or the weather.
Yahoo has an opportunity here. How many people have wished for true community but been appalled by Facebook's machiavellianism and Google's massively awkward concept of the "ideal UX"?