Oct 82012

Behavioral Targeting for Search Ads

At the risk of being pilloried: Social signals are not the be-all and end-all of the paid or organic SERP. For all the hyper-ventilation on the topic, social signals are but one signal among many to the ultimate treasure map.

Personalization is the ultimate treasure map.

Search engines seek to provide a SERP that best responds to the intent and interests of each individual user. An answer when the user's intent is informational and unambiguous (think: "Madonna's birth date"); a range of potentially valuable options when the user's intent suggests as much (think: "car insurance").

The engines have progressed from serving the same results to everyone, to serving personalized results based on geography, social signals, and past click-through behavior.

But couldn't Google take it a great deal further than that?

Google knows not only whether this person tends to click on Walmart links, but whether they buy from Walmart when they get there. They further know where else this person will shop, and where they never shop.

Does this browser ever transact online? For what kinds of sites?

Imagine what smart advertisers could do if those types of targeting options were available in paid search advertising? My ad copy could vary depending on the browser's customer profile. I might serve different copy and pay different amounts for the privilege, depending on whether they were a loyal customer of mine (brand heavy ad text) or a loyal customer of my competitor (Like my competitor? You'll LOVE me!") or someone with divided loyalty (promotional offers, perhaps?)

For online-only business, the value of browsers known to transact online would be materially higher than the value of those who don't. The value of folks who never transact online might be significantly more valuable to brick-and-mortar businesses with a less crucial online presence. You might pay more to make sure you retain your most valuable customers when they search than you would for someone who is a bargain hunter who returns everything.

Behavioral data has always been FAR more valuable than demographic, psychographic, or contextual data. Imagine the power of combining the in-the-moment intent of search with Google's browser data and your own customer data!

RTB comes to Google and Bing...

Is this a Brave New World? Terrifying? Exciting? Inevitable?

I'm taking bets :-)


6 Responses to "Behavioral Targeting for Search Ads"
This would be very exciting and it would help remove a lot of the frustrations of PPC advertising (they've given us a taste of controllable, performance based marketing which makes the things we can't control all the more frustrating). Two questions on this: 1. Why aren't the engines offering this already? 2. And what is going to change to make them offer this?
Thanks for chiming in, Richard. I'll hazard guesses to your on-point questions just because it's fun. 1) Bringing your own data would require engines to allow DSPs to serve ads against SERPs, that's a big leap given the money at stake; 2) What will get them to do this in some way or another (short of creating a Google Ad Exch for SERPs) will be assurance that they won't get sued for privacy violations and data suggesting they'd make more money doing so. Would be wild though, wouldn't it!
John Shea says:
Great article and topic George. I'm curious, if you could only have one which would you pick: 1) Ability to vary PPC bids based on overall online purchase recency, frequency and value. 2) Ability to vary ad copy based on merchant-specific purchase recency, frequency and value. For your readers, This reminds of one of my all-time favorite RKG Blog articles -- Purchase-Match: http://www.rimmkaufman.com/blog/purchase-match/13042008/ All the best, John Shea
Hi John, thanks for your comment. I'd choose #1 by a wide margin. Prominence on the page, I'd argue, has a greater impact on CTR than the ad copy, particularly at the thresholds moving from side to top placement or off the page to on the page. The variance in the value of traffic is best leveraged by bidding. Man, Alan was a great writer. Thanks for the reminder!
Toni says:
Hi George, I'd really like to see even some of that happen. :) Isn't this what Google is essentially doing with ECPC already, albeit on a limited scale? Of course it doesn't take into account your individual behavior (except when logged in?), but the aggregate behavioral patterns matched to your browser version, OS, time of day, location, etc. Of course we don't have much control over it or even access to the data... :)


Check out what others are saying...
[...] Behavioral Targeting for Search Ads, Rimm Kaufman [...]

Leave A Comment