Facebook audience targeting expansion for interest-based audiences, yay or nay?

Feb 3, 2020 | Facebook

Traditionally Facebook has treated interest-based audiences as very strict filters. If you wanted to target people interested in the ‘Boston Red Sox’ and ‘Clam Chowder’, the people in that audience would be people who have shown interest in the ‘Boston Red Sox’ and ‘Clam Chowder’.

Then Facebook introduced audience targeting expansion, a check box that in theory allows Facebook to target the people on Facebook who might not necessarily be fans of the ‘Boston Red Sox’ and ‘Clam Chowder’ but the people who Facebook thinks are close and will behave similarly to your ads as your targeted interest based audience should.

Our contacts at Facebook describe targeting expansion as:

“The purpose of this tool is to enable a situation in which the campaign budget can flow to the most valuable opportunities, regardless of the characteristics of the person being served the ad. This is because using detailed targeting provides the least cost effective results for DR outcomes. If we think about this historically, media buyers and planners would define their target audience and buy against them on the platform. Audience expansion is leveraging machine learning to better predict which people are most interested in buying a product. 

How does this work:

  • You set up a campaign to target users that are interested only in eyewear.

  • The way our platform works, we rely on several signals both on and off platform (e.g. things users clicked on, pages users follow, etc.) that determine that characterization of interest in eyewear.

  • There may be users who are interested in eyewear but may not be in that official detailed targeting category because the signals are not the same (e.g. user shopped for eyewear as a gift during a seasonal moment like Christmas vs. user browsed on Lens Crafters and 1-800-Contacts habitually. Again, this is just an example of how different signals can result in different categorizations.

  • Turning on audience expansion could allow for all audiences described above to be targeted.

Best practice: when working with advertisers with extremely small budgets, we shouldn’t be using detailed targeting at all. Rather, leverage other types of audiences where signals are inherent such as custom audiences and lookalike audiences. We want the biggest bang for our buck and the threshold with stat sig results is typically when budgets are bigger so our platform has more opportunity to find cheaper conversions.”

SO, like every feature, update, opinion or breathe from Facebook we over analyzed and tested targeting expansion to make sure we fully understood what it was and how it worked before we made recommendations to clients on whether or not to use it. 


The results from our testing were very inconsistent client to client. We determined audience targeting expansion for interest-based audiences is NOT an all client or none situation but instead it must be tested in all cases and those results should be used to determine whether or not it should be turned on.

Where to use it: 

We strongly suggest only using expansion when doing interest-based targeting. 

For retargeting, audience expansion makes no sense because expanding a retargeting audience creates a mix of two audiences and objectives – retargeting and prospecting. On top of that, if you’re looking to run true retargeting and are expecting to use the analytics to drive other decisions, this is certainly a way to muddy up your results.

Same goes for retention campaigns. Expanding your retention audience creates a mix of two audiences and objectives – retention and prospecting. Diluting your retention list, and again altering the learning from true retention campaign analytics. 

Running audience expansion in a prospecting lookalike audience will just make bigger lookalike audiences. A 1% lookalike audience, with expansion turned on, would be some sort of weird hybrid 2% lookalike. This may very well be effective, but we have not tested this yet and can’t stand by doing it as a best practice. 

How to use it:

We recommend testing each interest based audience twice simultaneously – once with audience expansion turned on and once with it turned off. After both audiences have run for a bit, determine a winner and pause the losing audience.

This certainly sounds like a pain and a lot of extra steps, but we promise that some audiences can perform significantly better expanded that it is totally worth the extra few steps.

We cannot stress enough, the results with audience expansion will be different for every campaign and every audience within every campaign. Do not just turn it on and forget it, you need to test it to determine if it’s right for your audience and your campaign.


Submit a Comment

Your email address will not be published. Required fields are marked *