Your ads are not underperforming. Your creative ran out

Here is a sequence we have now watched play out on maybe thirty accounts.

Cost per acquisition starts climbing. Someone rebuilds the audience structure. It does not help. The campaigns get split out for more control. That makes it worse, because now the budget is fragmented and nothing exits the learning phase. Bidding strategy changes. Attribution windows get argued about. Six weeks pass and the account is performing worse than when the intervention started.

Throughout all of it, the same four ad creatives are running that were running in January.

Why the targeting is almost never the problem

Ad platforms have spent years moving optimisation away from manual audience definition and into the delivery system. On a broad campaign, the algorithm is finding the buyers. What you supply is the thing it tests and the signal it optimises toward.

Which means the creative is not one input among several. On most accounts it is the main lever left that you control, and the one most teams treat as a fixed asset rather than a consumable.

Fatigue is not a feeling, it is a curve

Creative fatigue shows up in the data before anyone notices it in the account. Frequency climbs against the same audience. The click-through rate on a specific ad decays while impressions hold. Cost per result rises on that ad before it rises on the campaign.

The diagnostic is straightforward: look at performance per creative over time rather than per campaign. If your best-performing ad is also your oldest, you are not looking at a strong ad. You are looking at an account with nothing newer to compare it to.

The uncomfortable arithmetic

A campaign running at meaningful spend can exhaust a creative in a few weeks. Test properly — several concepts, several variants each — and the monthly requirement is not two or three assets. It is dozens.

Most brands cannot produce dozens of assets a month at the quality of a brand film, and should not try. What they can do is produce dozens at the quality of a good phone video, which on paid social frequently outperforms the polished version anyway.

The agency that tells you your hero film will carry a quarter of paid social has either never run paid social or is selling you the film.

Where the volume actually comes from

Three sources, in rough order of cost efficiency.

One shoot, planned for volume

A production day planned around the deliverable list rather than the hero film yields months of cutdowns. This is the single biggest lever and it costs almost nothing extra if the planning happens before the camera is booked — the argument we made in planning cutdowns before the shoot.

Creator content

Creator-made assets tend to outperform brand-made assets on paid social for a reason that is not mysterious: they look like the platform rather than like an advert. They are also produced at a volume and cost a brand team cannot match internally.

Systematic variation

New hooks on existing footage, different opening frames, different text treatments, different lengths. Not infinite — a bad concept does not improve with a new caption — but a working concept has more life in it than most accounts extract.

Test concepts, not colours

The most common waste in creative testing is variation that could not plausibly change the outcome. Button colours and headline synonyms do not move a paid social account.

Test things that represent genuinely different propositions: a different problem framed at the start, a different audience addressed, a demonstration versus a testimonial, a price-led message versus a quality-led one. Those produce differences large enough to read through the noise at realistic budgets.

What to change first

Before touching audiences, pull a report of performance by creative with a date axis. If the assets are old, the diagnosis is finished and the work is production, not media buying.

That is usually where we start on an account audit, and it is why our performance marketing and content creation teams are not separate engagements.