If you want to know whether a feature works, give it to an agency owner who's skeptical by trade. On The Truth About Social Ads, Jason Smith told me he did exactly that with Advanced Signal.
Advanced Signal creates two custom events in Meta from Wicked Reports' verified data. NC is a new customer purchase. RC is a returning customer purchase.
That means you can tell Meta which one you actually want. Instead of optimizing for "purchase," you can set the ad set objective to new customers only.
Jason didn't flip it on and declare victory. He gave it about a month so the data was flowing and healthy. Then he tested it on one of his bigger clients.
He ran a couple of end of month sales using the NC event at the ad set level. He used RC for some retargeting.
When he first turned it on, it was slow. Then the results came in on the new customer side.
He was upfront that it wasn't a perfect test. Budgets weren't equal across ad sets and total purchase volume was a little lower. But the ROAS difference was big enough that he's been using the NC and RC events since.
I like that he led with the higher CPA. It's what makes the result believable, and it's exactly what you'd expect.
Meta's algorithm is trying to get you the fastest sales it can. The fastest sales almost always come from people who already know you. Your existing customers.
So you get a nice ROAS on the dashboard, but the business isn't growing. People tell me "my ROAS is high but my business is flat." My answer is that you're just recycling demand.
To get new customers, Meta needs horse blinders. It shouldn't even see when a repeat purchase happens. When you optimize for the NC event, Meta has to go find people who have never bought. That costs more per acquisition. Jason's theory is that it's finding the better buyer the first time instead of sifting through everyone to get there. His ROAS says he's onto something.
If you're paying more to acquire new customers, judge them on new customer metrics. nCAC, not blended CPA. New customer percentage, not total purchases.
Jason now includes new customer percentages in the client reports he builds with Claude from Wicked Reports and Meta data. That's the number that tells you whether the business is actually growing.
They're custom conversion events created by Wicked Reports Advanced Signal. NC fires for new customer purchases and RC fires for returning customer purchases.
New customers cost more to find than people who already know your brand. In Jason's test, CPA went up while ROAS on the new customer event rose 30 to 40%.
Jason waited about a month for the data to flow in and stabilize before running his test on a live client account.
See it working: Watch the demos · Free guide: The New Customer Attribution Playbook