A few years ago having a database of customers’ contacts for creating lookalike audiences gave an edge in running ads. It gave a much better CPA. Though, these days, targeting in Facebook (Meta) Ads shifted from user-defined to AI-defined.

As you can see on the screenshot, there were 2 ad sets: 1 – targeting lookalike, 2 – “Generic” (targeting the whole country excluding lookalike audience). The generic ad set has even lower CPA than the ad set that targets lookalike. CPM in ad sets that target lookalike audience is significantly higher. I noticed this in many other cases in this and other accounts.

What it means to advertisers these days is that you don’t need the customer’s database to start getting good results with Facebook Ads. You just need to generate some initial conversions data for AI. Then, it serves ads to people who are the most likely to take the action you want them to. It also helps to scale ads – the fewer restrictions you set for targeting, the more room for optimization AI has.