Frequency is a symptom, not a cause. Capping it treats the reading rather than the problem. Real fatigue shows up as click-through rate decaying against a stable audience, cost per thousand impressions rising, and new customer acquisition cost climbing while returning customer revenue holds steady. The fix is creative supply, audience size, or both.

What ad fatigue looks like in the numbers, and the order to read them

Read three signals, in this order. First, click-through rate against a stable audience. Stable matters: if targeting, placements and geography have not changed and click-through rate drifts down week over week, the people seeing the ad have seen it before and are responding less. Change the audience halfway through and the signal is worthless.

Second, cost per thousand impressions. Rising delivery cost against the same audience means the auction is charging more to reach people the system finds harder to serve. Read it as confirmation rather than the trigger, since it also moves with seasonality and competitor spend.

Third, new customer acquisition cost rising while returning customer revenue holds steady. That split is the one that matters. If both were rising, you would be looking at a pricing, offer or site problem. When only acquisition degrades, the top of the funnel is worn out.

Why frequency is a symptom rather than the cause

Frequency is impressions divided by reach. It goes up for two reasons: you are showing the same people more ads because you have run out of new creative to show them, or because there are no new people in the pool. Spend divided by available audience produces the number almost mechanically.

That is why the same frequency reading means different things at different brands. One account sits at a high frequency with healthy acquisition cost because the creative pool is deep and each person is seeing different assets. Another sits at the same number with three ads in rotation and a shrinking audience. The metric is identical, the situation is not, and no threshold applied to the metric alone can tell them apart.

Why a frequency cap can quietly cost you reach

A cap is a delivery constraint. Once your best-responding people hit the ceiling, the system stops serving them and either underspends or reallocates that budget into weaker pockets of the audience. The frequency chart looks healthier afterwards, because you told it to look healthier. Acquisition cost usually does not improve, and often gets worse, because budget moved away from the people most likely to buy.

This is where a cap gets confused with a fix. It suppresses the reading without adding one new asset or one new person to the pool. Fatigue is a supply problem, and supply is the thing the performance creative system Plaid Testing runs is built to keep filled, so that rotation happens because there is something new to rotate in.

What to do instead, and when a cap is genuinely right

Diagnose which shortage you have before spending against either. Introduce two or three genuinely new concepts. If they perform on entry, you were short of creative, and the answer is a queue that keeps feeding rather than a monthly batch. If the new concepts also decay quickly, the audience is the constraint, and the work is expanding it: broader targeting, new geographies, new placements, new channels.

Working with a womens fashion brand, we grew sales 99% while efficiency improved at the same time. New customer acquisition cost fell 21%, new-customer ROAS rose 58% and net profit rose 136%. Net margin is still around 3%, which is worth stating plainly.

A cap is the right call in one narrow case: a retargeting pool you cannot expand. Site visitors from the last fourteen days is a fixed pool, and without a limit you will serve the same people the same ad for no return. Cap there. On prospecting, treat rising frequency as a request for more creative or more audience, and answer it. Plaid Testing runs a free thirty minute audit covering tracking, account structure and three fixes, including whether the creative queue can keep up with the spend behind it.