What you can tell from outside an account
Quite a lot, as it happens. Public ad libraries show which creatives are running, roughly how long they have been live, and how many variants exist. Being served the same ad repeatedly, across devices, tells you about frequency. The presence of near-identical ads with different framing tells you about audience structure.
None of that is precise. All of it is directional.
What we saw
Four creatives, live for about eleven weeks, appearing in both cold and warm contexts. Over two weeks of ordinary browsing we were served the lead creative repeatedly — enough that a colleague who had never visited the brand's site recognised it unprompted.
That is a frequency problem, and frequency problems are expensive in a specific way: they do not look like failure. The campaign keeps running, results decline gradually, and the natural response is to spend more.
The arithmetic of fatigue
Below a frequency of about 1.5 you are relying on a single impression to do the work, which rarely converts anything considered. Above roughly 3.5, the same people keep seeing the same ad, costs climb, and results fall.
The zone in between is where trust gets built. Most accounts we audit have never looked at the number.
The structural issue underneath
Cutting prospecting to fund retargeting is one of the most common and most damaging decisions in D2C, because the dashboard rewards it. Retargeting always shows better return — it is measuring people who were already going to buy.
Starve prospecting and the retargeting pool drains. Six weeks later, blended cost per acquisition rises and nobody can point to a decision that caused it.
What we would change first
- Split creatives properly: cold audiences get the problem and the proof, warm audiences get objection handling and urgency.
- Refresh creative on a schedule, not when performance visibly breaks.
- Restore prospecting spend and judge it on blended cost per acquisition rather than campaign-level return.
- Stop discounting to people who already bought.
Caveat
We have no access to this account. Everything above is inferred from publicly served ads and public library data, and could be wrong if the brand is running structures we cannot observe.
The reason we publish it anyway: the pattern is common enough that if you run a D2C account, there is a reasonable chance some part of this describes yours.