Intel·Creative Strategy

The Creative Rotation Mistake. And What It Costs at Black Friday.

Going into the highest-CPM period of the year with stale creative compounds the problem: you pay peak prices for an audience that has tuned you out. The ad library shows who has already made this call — and who hasn't.

AdIntel·28 October 2026·Meta Ad Library analysis

Key findings

  • Black Friday CPMs are the highest of the year. Going in with stale creative compounds the problem: you pay peak prices for an audience that has already tuned you out.
  • PrettyLittleThing has been running two body copy templates since 10 July. The same copy — 'If you needed a reason to shop, here it is' — is still live on ads launched 22 October. That's 110 days.
  • ASOS has launched four distinct creative waves since July. Their ninth creative concept of the autumn dropped on 5 October. Their audience has fresh material to engage with going into peak.
  • The discount escalation signal: when creative fatigues, brands compensate by deepening the offer. PLT went from 30% off in July to 70% + extra 10% in October. The creative didn't change. Only the number did.
  • DPA (dynamic product ads) follow different rules — the product rotation handles freshness. It's your hook and brand creative that need active rotation cadence.
  • Optimal cadence ahead of BF: one new creative concept per week from five weeks out. That's five waves before the offer lands, each giving the algorithm a fresh signal to learn from.

Creative fatigue is a problem year-round. In the five weeks before Black Friday it becomes a more expensive one. BF CPMs run 30–50% above the September baseline across most DTC verticals — the cost of reaching your audience goes up precisely when your creative is most likely to have exhausted its effective frequency. The two problems compound. You pay more to show an ad your audience has already stopped seeing.

Most brands treat creative rotation as a reactive problem — they pull creative when CTR falls and replace it when something breaks. That logic works when CPMs are stable. Heading into Black Friday it does not, because by the time the data signals fatigue, you are already inside the peak CPM window and launching new creative from scratch. The brands that perform best in BF week are not the ones that rotate fastest on the day. They are the ones that have been rotating consistently for the past month.

What 110 days of the same copy looks like

PrettyLittleThing's active library right now contains two body copy templates. The first has been running since 10 July.

PrettyLittleThing — active since 10 July 2026

“If you needed a reason to shop, here it is”

Still running on ads launched 22 October. 110 days, same copy. Only the emoji has changed (✨ → 🪐).

PrettyLittleThing — active since 10 July 2026

“Because the best outfits start with the best deals”

Second persistent template. Also still live on ads launched 22 October.

110Days PrettyLittleThing has been running the same two body copy templates. Their UK audience has had over three months to stop noticing them — right before the highest-CPM period of the year.

The creative itself has not changed. What has changed is the discount percentage in the headline: PLT launched at 30% off in July, moved to 20–30% off in October, and is now running “up to 70% off almost everything — plus extra 10% off” in the same week. That escalation is the tell. When copy stops converting, the instinctive fix is to deepen the offer. The creative fatigue drives the discount spiral, not the other way around.

What four creative waves looks like

ASOS has run four distinct creative waves since July. The most recent launched on 5 October — their ninth and tenth distinct creative concepts of the autumn, after a wave of eight new concepts on 29 September and a follow-up batch on 1 October.

ASOS — wave launched 29 September 2026

“Ready for your closeup?”

One of eight distinct creative concepts launched in a single wave. 'Autumn, your way', 'Step into the new', 'Cue: autumn', 'Capture autumn', 'Strike a pose' running in parallel.

ASOS — wave launched 5 October 2026

“You just found your new favourite outfit.”

Latest creative wave. No offer language. No countdown. Each wave gives the algorithm a fresh learning signal before the BF window opens.

Eight concepts in one wave is aggressive — most brands cannot sustain that production volume. But the structural pattern matters more than the count: ASOS is entering BF week with an audience that has seen fresh creative in October, September, and before. Their algorithm has recent learning data. Their frequency per creative unit is low. When their BF offer drops, it lands against a backdrop of varied, non-fatigued messaging.

The four things this tells you

1

Frequency and CPM compound — that is the actual cost

High frequency on stale creative means your effective CPM is higher than your reported CPM — you are paying to show an ad to people who have mentally filtered it out. When BF CPMs add another 30–50% on top of that, you are compounding an already-degraded signal at the worst possible moment. The brand entering BF week with low creative frequency gets more value from every impression it buys.

2

Discount escalation is a creative fatigue signal, not a strategy

When PLT moved from 30% to 70% + 10% over four months while keeping the same copy, they were not executing a planned offer escalation. They were compensating for copy that had stopped working. The offer does the job the creative used to do. That is a costly substitution — you burn margin to replace what a new creative concept would have recovered for free.

3

DPA rotation is not the same as creative rotation

Dynamic product ads refresh automatically because the product changes — a new SKU, a re-ranked bestseller, a different price. That freshness is real but narrow: it refreshes the product, not the hook. Your DPA copy ('If you needed a reason to shop, here it is') can still be 110 days old even when the products cycling through it are new. DPA and brand creative need separate rotation logic.

4

The optimal cadence is one wave per week from five weeks out

That is not five full production runs — it is five creative refreshes, which can be as light as new copy on an existing asset, a new format variant of a proven hook, or a UGC cut with different talent. The goal is to give the algorithm a fresh learning signal each week so that by BF week it has recent data on what resonates. Brands that launch their first new creative in October after months of BAU are starting the algorithm's learning curve at the worst possible time.

The mistake is not running creative too long in isolation. It is running it too long into a period when CPMs are rising and creative fatigue is most expensive. PrettyLittleThing is not a brand with a bad creative team. They have a structural cadence problem: two templates have carried the load since July, and they are now entering the highest-CPM fortnight of the year with an audience that has seen those templates a hundred times.

The ad library makes this visible in a way that internal reporting does not. Inside the account, you see CTR trends and frequency scores. From the library, you see the date the creative launched and how many distinct concepts a brand has produced since. Those are different pieces of information, and the second one is harder to see from inside.

Run your own

Want this kind of breakdown on your own competitor set? Snapshots pull any brand’s active Meta ads and structure the data for you.

See how Snapshots work →
← More Intel