Let's Talk About Lookalike Audiences

Martin Kilp Jun 28 2019 6 PM | 3 min read

If you’ve ever run a campaign on Facebook, chances are you’ve used a lookalike audience. It’s no secret to those of us in the paid advertising industry that these audiences can drive performance for your campaigns and significantly improve your return on ad spend.

In essence, you’re asking Facebook to put its social graph to work for you. Once you’ve mapped out your source audience (e.g., all players that have completed the tutorial in your game; or all players that have completed at least one in-app purchase etc.), Facebook will identify the common qualities of said audience. Next, they analyze 100s of user attributes to seek out users that most resemble your seed audience. In other words, you’re putting the heavy lifting on Facebook’s shoulders - you’re betting on Facebook to locate the most valuable users and maximize your return.

Needless to say that when lookalikes first became available, they proved exponentially more effective than broad or interest targeting. Advertisers were smart to funnel the majority of their spend via lookalike audiences. Over time, however, as advertiser adoption of lookalike audiences increased, the marginal competitive benefit to using LALs started to decrease. After all, if everyone’s running lookalikes, you start to lose your edge. And, that’s precisely why we need to talk about lookalikes. 

What’s Next for Facebook Audiences?

The answer: a nifty new Facebook feature called Ad Set Based Conversion Lookalikes (ABC LALs). The feature builds on the age-old lookalike concept by generating lookalikes not from seed audiences but rather from well-performing ad sets. The ABC LAL brings several advantages over conventional lookalikes.

Better Match Rates

Since your seed audience is made up entirely of Facebook conversions, its audience match rate is 100% by design, meaning that you’re providing Facebook with the richest possible data signal with zero noise. For comparison, the match rate for app event based seed audiences for gaming advertisers is typically around 60-70%.


ABC LALs continuously update whenever your seed ad sets receive new conversions. This automation saves you time as you no longer need to manually compile device ID lists from your mobile measurement partner. What’s more, this quickly becomes a self-fueling loop where the more traffic your seed audiences accrue, the more powerful your ABC LALs become at locating the most valuable FB users.

Dual Intent

Perhaps most importantly, the ABC LALs screen for two significant signals. Firstly, just like app event based LALs, they screen for conversion propensity (the likelihood that a given user will convert). However, unlike conventional lookalikes, ABC LALs also screen for ad clicking propensity. In other words, you’re asking Facebook to look for people that are both likely to click on your ads as well as convert. The intersection of these two intent signals is what ultimately drives substantially better performance than either signal alone.

Incremental Reach

In our experience so far, ABC LALs have minimal overlap (<5%) with other top performing lookalikes. ABC LALs allow you to tap into entirely new Facebook audiences to whom you haven’t previously served ad impressions. Consequently, you can also start experimenting with ABC LALs in parallel to your current lookalikes without fear of cross-contamination. 


Ad Set Based Conversion Lookalikes are an API-only Facebook feature that’s available on the platform. Our team of paid social experts is more than happy to help you get started with ABC LALs and level up your Facebook performance.

   Retail Re-UP: Adjusting to the Next Normal of Social Advertising CHECK IT NOW
Martin Kilp

Read Next