Advantage+ audience with no interests, or keep interest stacks on a new Meta account?

Take a skincare shop doing about 40 orders a month on a fresh Meta account. The old playbook was three or four interest stacks, narrow audiences, and hope the pixel data trained the system. Now plenty of people just switch on Advantage+ audience with only age and location set and let Meta find the buyers.

Honestly, I think interest stacks mattered most when the algorithm had plenty of conversion data to learn from. At thin volume they're mostly a guess about who buys. Going broad only works if the creative does the targeting, because the hook has to make the right person stop scrolling.

I'd still start broad for 2 to 3 weeks, but only once the purchase event is firing cleanly. What are you seeing at low volume?

Thin volume is a bigger problem with interest stacks than targeting quality. At 40 orders a month, three or four separate ad sets each land around 10 purchases, so none of them gets enough conversions to leave the learning phase. You end up paying for four partial samples instead of one decent one, whereas a single broad ad set gives Meta one pool to learn from.

Going broad does put more weight on creative, though. For the first two or three weeks you're really testing hooks, and five near-identical ones on 40 orders is too many variables to read. Two or three genuinely different angles, like a before-and-after, an ingredient explainer and a customer quote, will tell you more than a dozen small tweaks.

Are those interest stacks still running as separate ad sets right now?

Marco, I'm not sure the "one broad ad set gives Meta one pool" idea holds. Advantage+ audience still shows the ad to whoever Meta predicts will buy, so it isn't really one clean sample. And its learning depends on the purchase event firing properly. At 40 orders a month, that's the first thing I'd verify before trusting either setup.

Interest stacks don't give you a cleaner sample, they just swap the bias around. Instead of Meta guessing who buys, you're guessing, and at 40 orders a stack built on one assumption about "people who like natural skincare" can be just as skewed, only smaller. Either way you're learning from a narrow slice. I do agree on the purchase event, though. Check that first, whichever setup you pick.