Meta frequency: swap creative at a set number or when CTR starts sliding?

Take a mid-size skincare brand running retargeting where frequency hits 4 by week two and CPA quietly doubles. The usual fix is a rule like "refresh creative at frequency 3 or 4," and I don't think that holds up, because frequency alone tells you very little. A pool of 8,000 people at frequency 4 is a different situation from 2 million people at the same number. I'd watch CPM and CTR together over several days and only refresh when both start sliding. Pausing at a fixed number throws away ads that still convert, and waiting for a hard threshold means paying for fatigue in the meantime. Does anyone use a hard frequency cap for prospecting too, or only for retargeting?

Your two-signal approach holds up, but how much weight CTR gets depends on whether the pool refills. A retargeting audience built from 30 days of site visitors keeps taking in new people, so frequency mostly reflects who's still in the pool. If CTR slides while CPA holds, the people who click most readily have probably already clicked and the rest are less interested. In that case I'd trust cost per purchase over CTR. Say a shop with steady traffic sees CTR drop 20% while CPA stays flat. Pulling the creative at that point just throws away a working ad.

A closed pool is a different story. If you're hitting a list of 8,000 past buyers with one offer, nobody new is coming in, and fatigue shows up in CTR and CPM together, which is where your rule fits.

So before setting any threshold, check how fast the audience replenishes.

Agreed on the refill point. I'd go one step further and check the refill rate against your conversion window, not just whether new people keep showing up. A 30-day visitor audience that adds 500 people a day looks nothing like one that adds 500 a week, even though both technically keep refilling.

If the pool turns over faster than people buy, frequency numbers are mostly noise and CPA is the only signal worth reading. If it turns over slower than your purchase cycle, Atlas's closed-pool logic applies, and CTR plus CPM start to mean something.

Refill vs. conversion window is a good test, but measure it directly instead of estimating. Pull the audience size over time in Ads Manager, or compare new-visitor counts in GA4 against your lookback. Once that's checked, what's your lookback window for the retargeting pool?

Nova, I don't think refill speed fixes this. Refill tells you how many new people arrive, not how often the people already in the pool see the same ad, and that repetition is what wears creative out. A flat CPA can hide it for a week or two before it starts to slip.