Meta frequency as a fatigue signal: refresh creative at 3 or wait for CPA to move?

A lot of people treat frequency like a hard alarm: hit 3 in seven days, pull the ad, make a new one. For most small accounts I think that's backwards. Frequency mostly reflects audience size, not creative wear. If you put a modest budget into a tight local audience, frequency climbs fast even while the ad is still working, because there are only so many people to show it to. Say a dental clinic targets a 40 km radius and sits at 4 or 5 after two weeks, still converting at a steady cost. That ad is fine.

What actually signals fatigue is CTR falling while CPA rises over several days, not the frequency number on its own. Frequency is a decent warning light, but it's a poor reason to kill a winner. How do others set the trigger when the budget is small?

Agreed on audience size. With a small local audience, frequency mostly tells you how saturated the pool is, not how worn the creative is. I'd anchor the trigger to the ad's own first week after it leaves learning. Compare today's CPA to that baseline, and if the gap stays wide for several days while CTR drops on those same days, that's real fatigue. A 30% drift held over a week would be my line, though it varies by offer. Frequency only matters after that, as a reason to check whether new people are still entering the audience.

What baseline are you using for CPA, the first week or a rolling average?

The first week is shaky when conversions are thin, because one or two bookings can swing CPA by half. I'd only trust the numbers once there are around 15 conversions, and I'd compare 7-day windows instead of single days. For clinics with delayed bookings, the last 2 or 3 days also look better than they really are, so leave those out.

Fair on the 15 conversions, but what matters more is how many people actually see the ad each week, not the conversion count. Take a clinic with a 40 km radius and an audience of 20,000 people. Reach against that pool is what decides whether frequency means anything, so I'd check audience size before reading the frequency number at all.

Agreed on checking the pool first, but I'd compare reach against the pool over the same time window rather than staring at the raw 20,000. If one ad has hit 60% of a 20k audience in two weeks, a frequency of 4 mostly means you've already shown it to nearly everyone who'd book. At that point, widening the audience will do more than a new creative.

I wouldn't widen it. A clinic that only serves a 40 km area mostly ends up paying for people who can't get to the chair, so CPA usually gets worse before it gets better. And with 60% of the pool already reached, a fresh creative aimed at those same people is the cheaper test.

Depends on whether the people who book are a fixed group or a pool that keeps refilling. A clinic with a one-off offer, like a consultation or a whitening package, mostly converts people who've already seen the ad once or twice, so a fresh creative aimed at that same 60% is a reasonable test. But if most bookings are repeat visits, like cleanings every six months, the people in that pool who want an appointment are often already booked, and a new hook won't find new bookers. Then the real constraint is how many new people move into the 40 km radius each month, and widening the radius or trying a new audience matters more than the creative. So the split that decides it is first-time versus returning patients, and if you can't tell them apart in the conversion data, fix that first.