How a value-based lookalike works
An ordinary lookalike treats every person in the source list as equally worth copying. A value-based one does not. You supply a value alongside each customer — from an uploaded file with a value column, or from purchase events that carry a value — and Meta weights the seed accordingly, so the traits of your better customers count for more when it decides who resembles them.
The output is still a cold prospecting audience in a chosen country and a chosen similarity band. Nothing about the delivery changes. What changes is who ended up inside it: an audience shaped by your most valuable buyers rather than by whoever bought most recently or most often.
Why value-based lookalikes matter
Most customer lists are lopsided. A small group of buyers accounts for a disproportionate share of the money, and the rest barely cover the cost of acquiring them. A plain lookalike averages all of that together and quietly asks for more of the middle. A value-based one asks for more of the top.
It matters most where order values differ widely — a travel operator selling both day tours and long expeditions, a clinic selling consultations and procedures, a remittance product with occasional senders and heavy ones. Where every customer is worth roughly the same, it changes very little and is not worth the setup.
Common mistakes with value-based lookalikes
The first is using revenue when you mean profit. A discount-driven bulk order can look like your best customer on a revenue column and be one of your worst on margin. Meta will faithfully go looking for more of them.
The second is using a single transaction as the value when the business is repeat-purchase. The customer who spent little once but returns every month is the one you actually want copied, and only a lifetime value figure will say so.
The third is uploading a file with values missing on many rows, or letting it go stale. Blank values weaken the weighting; an old file describes the customers you had two seasons ago.
How to act on it
Decide first what the value means and write it down: gross profit, contribution after refunds, or lifetime value over a defined period. Use the same definition every time you refresh, so the audience is not chasing a moving target.
Export enough history for the weighting to have something to work with, fill the value column on every row you can, and refresh on a schedule rather than when someone remembers. Keep an ordinary lookalike running alongside for comparison, and judge both on what the customers were actually worth after they bought, not on the cost per lead. As with any lookalike, the seed audience does most of the work; the value column only tells Meta which parts of it to take seriously.