– ‘Look Alike’ Audience Targeting, without Cookies



Facebook recently announced ‘Look Alike Audiences‘, or the ability to target ads to users who look like your current customers. Simply upload an email, phone or other “list” of your customers and Facebook will find other Facebook users that “look like” those users. [You can also upload browser UID's to FB, linking offline and online profiles in some instances. Unless you work in the ad tech/online media biz you are not aware of this ongoing tracking capability. FB is walking a fine line with its users over their privacy, when you come and go from FB you pass over your browser history. This allows sites you visit to re-target you on FB. Once FB integrates ATLAS into its platform you can expect even more granular BT tactics]SC

Look alikes have been around for a long time. They are used heavily in direct marketing where “lists” are the lifeblood of a mailing or calling campaign. They have been used in online advertising for the last several years as well. While Quantcast is probably the best known for look alike modeling, most ad networks, DSPs and DMPs also have similar capabilities: Optimal, Audience Science, Turn, BlueKai/Datacratic, RadiumOne, to name just a few. A primary difference is what they use to model against. Facebook uses social profile data (stuff users explicitly divulge) whereas most networks, DSPs and DMPs use anonymous cookie based data and historical profiles (largely based on assumed interests from implicit behaviors, like viewing a web page). [and these days FB partners with DSPs and DMP's by sharing data for advertisers. They are not separate ecosystems, this is the pot of gold FB pitches to Wall St] SC 

What do all of these look alike providers share in common: user level data and the privacy baggage associated with that data (opt-ins, opt-outs, PII, anonymity, privacy policies, etc.).  It is possible to build your look alike models without any privacy red flags. No cookie data or lists of PII needed. Just some URLs. There are three ways to use URL lists for Sequential Media’s Look Alike modeling:

Existing Sites on a Media Plan

Let’s say you buy the same twenty sites over and over. They work well for hitting your audience and performance goals. Provide your list of proven sites to Sequential Media and we can build a look alike campaign against those sites.

Top Performing DSP or Ad Network Sites

If you work with a transparent network or DSP, ask them for a list of your top performing sites. Sequential Media can build a look alike campaign from those sites. [Good luck getting that list, once you have it you dont need the DSP. Pubs prefer to work direct with advertisers and will package a deal comparable to what you were getting with a third party reseller, including additional exc. branding/integration opps] SC 

Referring Sites from Site Analytics

If you have access to your “referring sites” report within Google Analytics, Webtrends, SiteCatalyst or other site analytics package, then you can export a list of sites driving the most traffic to your website. Sequential Media can build a look alike model and find similar sites.

All of Sequential Media’s Look Alike modeling is based on our anonymous, patented crowd targeting algorithms. We can generate a model in a matter of hours and uncover numerous needles in a haystack…lots of great long tail sites to supplement your existing campaign.



You can be the first one to leave a comment.


Leave a Comment