Phil Leggiere interviewed Right Media’s director of Product Management, Alex Hooshmand and published the interview in the January 31st 2007 edition of MediaPost’s Behavioral Insider.
At the end of the interview, Phil asked about the new frontier of Behavioral Targeting, what is coming next. Hooshmand’s response, “we now have several clients who are using post-search behavior to target banner or display ads.” Let’s get into that.
So what are the options?
Obviously RightMedia has some offering within their exchange network although I have not been able to find anything more than that. Plus as a hermetically sealed network you are limited to being a buyer or seller within that auction environment. Works great for direct response, low-dollar advertisers and publishers with remnant inventory but not for the rest of the market.
One options is Post-search advertising. AlmondNet delivers post-search paid-listings to users based on previous search behavior across its distributed ad network. If a user searches on an item through conventional search, their search behavior is cookied and tracked. When they are encountered in the future they are targeted with relevant paid listings. This is a lot like behavioral targeting only with paid listings and with search instead of pixel-associated events.
MSN’s new AdCenter offers an advertiser the opportunity to target their search advertising by demographics, geography, day-part and several other parameters. So they are using browser-based cookies to single-out users for targeting. Crossing the chasm to then offer an advertiser the opportunity to subsequently advertise a banner ad to someone based on search response behavior would not be a hard leap to make. But that is my supposition and is not something that has been publicly been brought to market. But it will I am guessing.
Then of course there is my favorite, the creative approach that the early adopters are deploying. Search advertising with First Party cookie ad serving.
Advertisers that manage healthy search campaigns will usually employ the services of an ad server to track their campaigns – leveraging unique click-thru URLs and landing pages to track each keyword. This approach enables the advertiser to measure the effectiveness of every keyword. While the search engines may provide impression data on the keywords through their reports, and clicks, the ad servers can provide successful clicks and then post-click events (what happens after someone clicks on the keyword and enters the advertiser’s site) when the advertiser’s site is properly tagged with tracking pixels.
One of the benefits of using an ad server is to have the comparative reporting between a search engine’s reported clicks and actual clicks. Up until recently, Google reportedly had a click-fraud rate of approx 12%. Now it is 2% with the invention of something they refer to as “invalid clicks” making up the other 10%. Invalid clicks are screened out clicks that you no longer have to pay for. So they are making good on the evident occurrence of people clicking on multiple links before pages load, “stopped” browsers, spiders and bots, failed page loads and other behaviors that result in “fraud” click counts but unsuccessful events. Whereas, the ad server counts the click as resulting in someone landing on the advertiser’s web page.
But back to the integration of search and banner advertising. When using an ad server to manage search campaigns, a user receives a cookie when they click-thru to the advertiser’s web site. I know I have gone though this before so my readers should have this down. But the basics are as follows:
The cookie is placed on the browser so that they can be tracked through to the advertiser’s web page and the activity can be credited back to the keyword and search engine. As this user continues to surf the web they can be recognized and targeted based on that cookie with banner ads.
If it is a third-party cookie, it is event-based targeting. TACODA, Advertising.com are network examples and Boomerang are ad server examples that can apply this technology and can target a user based on their search behavior.
If it is a first party ad serving implementation – DirectServe – then the cookie that is applied is the advertiser’s cookie. The behavioral targeting features still apply insomuch that if all they do is visit the site, they can be targeted with future ads just like with the example described above for third party providers.
But with first party implementations, the user can also be targeted based on advertiser knowledge generated from the site visit. For example, if the user clicked on the search term and registered for information, purchased a product or applied for a loan then they are in the CRM system and are a known individual that can be included in a customer segment. Customer segments can be targeted with DirectServe, first party ad serving.
A user who clicks on a search term and visits the advertiser’s site; and who then completes some level of activity that results in their identification will get a first party cookie. This individual can then be re-targeted with ads anywhere across the internet at anytime as a customer or registrant. They can be up-sold, cross-sold or otherwise targeted as an anonymous member of a customer segment (brand preference, purchase frequency, buying habits, etc.).
Post-search behavior can be used to create the customer segments when the users arrive for the first time. For example, the segment examples can further be dissected to include keyword groups so that when targeted, the advertisements appeal to keyword groups that initially generated the user’s response. Once the user returns to the web page – just like when they first arrived at the site – a content management system can leverage the actual keyword to customize content delivery and properly display product information to maximize revenue or other desired response.
I’d love to hear from you on this. This can be done a number of ways. But the easiest that I have come across so far is to integrate the three – search, behavioral targeting and DirectServe/first party ad serving.
As I have described in the past DirectServe has three phases of implementation: (1) re-targeting, (2) cookie-writing and data delivery for analytics and (3) integration – CRM, Content Management and Site-Side Analytics. But for the purpose of this post and this example, I am really only focusing on re-targeting. That is as far as you need to go and you will already be way ahead of the curve.
What else can you do?
If you integrate your search with your ad serving, leveraging post-search capabilities to drive your behavioral targeting (prospecting) and customer re-targeting (DirectServe), you will generate data that you can analyze about customers that will enable you to better understand not just what search terms generate leads but what search terms generate customers, customer segment groups, customer values, repeat custom actions and long-term metrics. Grouping keywords together will help you determine long-term effectiveness of search campaigns. Furthermore, by integrating post-search with banner advertising, you will be able to recognize how search and banner messages combine to effective solidify messages and have the same impact that can be measured with the same metrics described above. You can go hog wild! But most importantly you can measure and determine how to better allocate media spend. If search works for you, you will know why and how. You will come to recognize how to compliment it with banner advertising.
Last thing. When you use first party ad serving, the cookie that you tag a browser with helps you to measure advertising audience. This means that when you advertise on Yahoo and you buy 1 M impressions, you will know exactly what % of that audience is comprised of existing customers and what % of that audience is comprised of people who have not been to your site before (or who have recently cleared out their cookie file).
What about search? The same holds true. Any of your customers who carry your first party cookie will also be measurable. If someone searches on a term and clicks through to your web site, and they are an existing customer already, they will be measured as an existing customer (their customer segment type will be measured) and you will know what % of the search audience you capture is already comprised of existing customers. Interesting tidbit. How much money do you spend with search recapturing recurring revenue?
Reactionary with Insight.
_uacct = “UA-980395-1”;