Executive Summary
- • Privacy changes reduced some observable conversion and matching signals used in audience creation.
- • Broad Targeting allows Meta's AI to use your Creative as the primary targeting signal.
-
•
Action: Test broader targeting when narrow audiences limit delivery or add unnecessary assumptions.
There is no universal winner between broad targeting and Lookalike Audiences. Broad gives Meta more room to optimize, while a strong lookalike can still be useful when its source data is relevant and large enough.
Why Lookalikes Failed (The iOS14 Effect)
Lookalike audiences rely on seed data such as customer lists or conversion events. Privacy choices and missing or outdated identifiers can limit matching. Check the quality and relevance of your own source audience rather than assuming a universal match rate.
"Restricting the audience with Interests or LALs is actively preventing the AI from finding cheaper conversions." — Meta Engineering Team
The Power of Broad Targeting
Broad targeting gives Meta more flexibility within the location, age, and other controls you keep. Delivery can use creative response, conversion signals, and platform behavior to find likely responders.
How it works:
- You launch an ad about "Dog Toys."
- Meta shows it to a small sample of users.
- Dog owners click on it.
- Meta's AI learns: "Oh, dog owners like this."
- It finds MORE dog owners, even if they aren't in your "Dog Interest" list.
This is called Creative Targeting. Your ad copy and creative *are* the targeting.
Sources & References
- 1. Meta AI Sandbox. "Generative AI and the Future of Ad Targeting." (2024).
- 2. Common Thread Collective. "The Death of the Lookalike Audience." (Case Study).