Paid Meta (Facebook & Instagram) Audience Targeting Best Practices Post-iOS14 & Other Privacy Changes

The launch and widespread adoption of iOS14 reduced Meta's ability to track user interests and behaviors at the level of detail they were able to previously. This significantly decreased the effectiveness of Meta's ad targeting algorithm, and led to performance declines for most advertisers.

In response, Meta has been continually updating their algorithm to perform better without the ability to directly track user interests and behaviors on Meta or within other tech platforms and brands (such as Google).

As advertisers and brands adjust to this constantly evolving landscape, many are confused on how to best structure campaigns and audiences, and why strategies that previously were frowned upon are now the new norm.

See below for a summary of audience targeting best practices based on our experiences over the past few years, and click the button above to download the PDF version of this post.

*Note that we will use "Facebook" to encompass Facebook and Instagram in this post.

Facebook Audience Targeting Best Practices Post-iOS14

• Focus on broad audiences between 1M and 10MM people. 
• Avoid behavioral or interest (and thus most Lookalike) targeting.
• Generally avoid narrow geo-targeting as this often causes audiences to fatigue over time. However, larger regional or country-wide geo-targeting can work well. 
• Key optimization driver is a continual flow of high-quality creatives that the algorithm can use to discern interest and behavioral targeting within a very large audience of potential prospects.

Why the Shift to Broad Audience Targeting in Facebook?

• The large-scale adoption of iOS14 in the summer of 2021 led to a massive user data loss for Meta. Thus it was no longer able to understand behaviors, demographics, or interests the way it did previously. 
• Therefore traditional interest, behavioral, and lookalike targeting stopped performing as well as it used to.
• In response, Meta updated their algorithm as an “AI” capable of machine learning. 
• A basic tenet of machine learning is that the more good conversions the system gets on a daily or weekly basis, the better it performs.
• Large broad audiences ensure that the algorithm continually has enough potential prospects to test and learn from so audiences don’t continually fatigue.

What Does Facebook Broad Audience Targeting Mean for Lead Generation Accounts?

• Marketers should no longer rely on lookalikes, behavioral, or interest audiences as - even if they perform for awhile - they tend to fatigue over time.
• Extremely broad audiences oftentimes work well if they are given strong creatives and are allowed to learn over a period of months (the longer they run, the better they perform).
• Using different creative types and messaging themes is a good way to segment / target within broad audiences.
• Segmenting broad audiences by large regions also can work well. 
• Creative is extremely important, and videos / slideshows and reels creatives tend to be top performers.

Case-in-point: We work with a healthcare client to drive leads for multiple clinics across the US. A few years ago, it worked best to target lookalike audiences built from lists of highest lifetime value (LTV) patients, and to target a narrow radius around each clinic. Now, we've found that our paid social ads perform best if we target a broad audience (18 to 64) and to expand our radius around clinics. As a next step, we are even considering targeting entire DMAs to give the algorithm as much data to learn from as possible, and thus limit fatigue.

What Does Facebook Broad Audience Targeting Mean for eCommerce Accounts?

• Marketers should no longer rely on lookalikes, behavioral, or interest audiences as - even if they perform for awhile - they tend to fatigue over time.
• Broad audience ASC (Advantage+) shopping campaigns tend to perform best, but drive more view-through conversions than many other campaign types.
• Broad audience DABAs can perform well when segmented by the correct themes / product types / business goals.
• Creative is extremely important. A healthy mix of quality video carousels, static carousels, reels, and static images is a winning combo. Static images (especially for promos and product shots) still are a top performer in eCom.

Case-in-point: Across all of our eCommerce accounts, we've found that targeting lookalike audiences (including high LTV or top tier customer lists) and targeting interest-based audiences (including competitor interests) no longer work as well as targeting all genders and ages across the US. Instead of segmenting by audience demographics, we've instead begun segmenting by a combination of:

• Campaign type (ASC, DABA, DPA, etc),
• Offer / promo (free trial, 50% off, etc),
• Product type or collection (modern farmhouse decor, anti-aging, etc.),
• Price point (products under $50; decor from $7.99, etc),
• Creative (video, static product shots, influencer reels, product collections like rattan, etc),
• Intent (Top of Funnel, Middle of Funnel, Bottom of Funnel), and
• Goal (Purchase, Add to Cart, Landing Page View)

We've also begun creating a funnel (or segmenting) within the above segments. A few examples:

For brands with evergreen offers, we have created a full funnel structure for each offer via a mixture of segmenting by campaign type, intent, and creative.

We also have seen success when setting up the DPA as a funnel via a mixture of segmenting by intent and goal. For example, we break the DPA into 2 campaigns: Low Intent and High Intent. We then set up the Low Intent campaign to target an add to cart goal. Whenever someone adds to cart, they're then moved into the High Intent campaign that targets a purchase goal.