Value-Based Lookalike

A value-based lookalike is a lookalike built from a source that carries how much each customer is worth, so Meta models new people who resemble your biggest spenders, not just anyone who bought.

A plain lookalike treats a 500 rupee customer and a 50,000 rupee customer as the same "buyer," so Meta goes off to find more buyers of any size. A value-based lookalike changes the seed it learns from. You feed Meta a source that includes each customer's value, your purchase events with value attached, or a customer list with a value column, and Meta builds the audience weighted toward the people who look like your high spenders. You have stopped asking for more customers and started asking for more of your best ones.

For a business with a wide spread of customer values, this can quietly reshape who your prospecting even finds. It is the audience-side partner to value optimization: one tells Meta to value big orders when it optimizes, the other tells it to go find the people who look like the ones who place them. Used together, they point the whole machine at revenue rather than at raw order count.

It lives or dies on the value data in your source. If your events or your list do not carry real, accurate values, Meta has nothing to weight on and you get an ordinary lookalike wearing a fancier name. And you need enough high-value customers in the source for the model to find a pattern, because a handful of whales is not a foundation for a lookalike, it is just noise the model cannot learn from.

A value-based lookalike only finds your best customers if your data actually tells Meta who your best customers were.