We use cookies to personalize content, to provide social media features and to analyze our traffic. We also share information about your use of our site with our social media, advertising and analytics partners. For information on how to change your cookie settings, please see our Privacy policy. Otherwise, if you agree to our use of cookies, please continue to use our website.

PurchaseMatch: How GOOG Could Hit $750, or How Yahoo / Microsoft Could Best Google

Here's my advice that, adopted by Google, could add 50% to their valuation. Or if adopted by Yahoo or Microsoft, it could put them back in the game. space shuttle taking off
Let advertisers bid a premium for clicks from users with a recent online purchase.
Tomorrow I have post coming out at SEL arguing that the wisdom of the direct mail greats can teach us paid search marketers a great deal. In other words: if the likes of Claude Hopkins, Albert Lasker, John Caples, David Ogilvy, Leo Burnett, and Maxwell Sackheim were alive today, they'd be (a) buying AdWords, and (b) kicking butt. In the SEL post, I turned up my nose at demographic targeting for paid search. Direct marketers have long known that demographics do a weak job of predicting response, whereas purchase recency does an incredible job. I mentioned this idea in passing on the SEL post and wanted to expand on the notion further here. Privacy concerns aside for the moment, consider the following proposal:
  • This hypothetical program is called PurchaseMatch. (I just made that up.)
  • Advertisers are only eligible participate in PurchaseMatch if they share conversion data with the search engine. (Hattip, Tony White and Abacus.) For Google, an Adwords conversion tag or Google analytics tag would qualify.
  • PurchaseMatch is optional; eligible advertisers don't have to use it.
  • Just as advertisers today have the option to bid at the AdGroup level (dumb) or at the keyword level (smart), advertisers would have the ability to differentiate their bids by sixteen different PurchaseMatch segments.
  • Here are proposed PurchaseMatch segments:
    Days Since Last PurchaseEXACT match PHRASE matchBROAD matchNON match
    Today E1 P1 B1 N1
    This Week E7 P7 B7 N7
    This Month E30 P30 B30 N30
    This Year E365 P365 B365 N365
    Here's an example to clarify. Suppose you're advertising the phrase "Canon Digital Camera".
    • If the engine detects the SERP was served to a user who bought anything at all online in the last year, that'd be a N365 user.
    • If the user had bought anything online in the last 30 days, that would also be a N30 user.
    • If the user had bought from an online store after clicking on an ad that broad matched "Canon Digital Camera" in the last 30 days, that would also be a B30 user.
    • If the user had bought from an online store in the last week following a click on an ad which exact matched "Canon Digital Camera", that would be a E7 user.
  • The core idea: let advertisers bid more for ads based on user type, where the user type is a combination of purchase recency (day, week, month, year) and ad phrase relevance (exact match click before purchase, phrase match click before purchase, broad match click before purchase, any paid click before purchase).
Yes, this is too complicated. Smarter folks than me could likely simplify it intelligently. But bidding by purchase recency would be marketing rocket fuel. If, say, the economics of the phrase "Canon Digital Camera" made sense for a retailer today at $1.00 CPC, I'd guessing that under PurchaseMatch they'd be delighted to bid something like this:
Days Since Last PurchaseEXACT match PHRASE matchBROAD matchNON match
Today E1:
$20
P1:
$14
B1:
$10
N1:
$5
This Week E7:
$10
P7:
$5
B7:
$4
N7:
$4
This Month E30:
$5
P30:
$4
B30:
$3
N30:
$2.50
This Year E365:
$1.75
P365:
$1.50
B365:
$1.50
N365:
$1.20
Of course, I just made those bids up. If PurchaseMatch existed, smart bid management algorithms would use statistical optimization to determine the relative value of inbound clicks from different segments of users, and bid accordingly. But I have no doubt that data would show that smart retailers could pay twenty-fold more for an E1 click versus what they'd pay for a generic click. Some of you might be scratching your heads and thinking, "Hey, shouldn't your guess for the value of an E1 be worth less than a E30, because the person just bought the widget and so have no need for another one for some while?" While that makes intuitive sense, in almost every case classic RFM has shown that response rate monotonically decreases with increasing R. Others might be scratching their hands thinking, "Hey, why should the engines optimize for the direct response crew, as DR is only 10% of total US adspend, and the general advertising folks who comprise the 90% love their demographics?" I'd answer that I think Larry Page is right: the future of advertising belongs to direct. And I think the privacy issues could be surmounted... What do you think? As a direct marketer, would you bid a premium based on recency data? Why or why not? As a consumer, is PurchaseMatch just too creepy? Why or why not?
Join the Discussion