Half of all creative is retired before it turns four weeks old
Creative lifespan compressed much faster than production budgets did. The arithmetic only closes if you change how a shoot is planned, not how fast the editor works.
Production teams are being asked for something that sounds unreasonable until you look at where the numbers come from. Ship more, ship faster, ship in every format, and accept that half of it will be dead inside a month.
Motion's 2026 benchmarks put the retirement figure plainly: roughly half of all creatives are retired before day twenty-eight. Brands spending over a million dollars a month ship around 18.85 new creatives a week. The smallest tier ships 2.8. Top performers at every level run two to three times those volumes.
Against that, most brands still plan production the way they did when a hero asset ran for a quarter. One shoot, one hero cut, a handful of resizes, and a gap of several months before the next one. That model does not produce enough surface area to survive four weeks of an automated feed.
Resizing is not volume
The tempting shortcut is to hit the number by cutting one film into nine aspect ratios and calling it nine creatives. It does not work, and the reason is specific rather than aesthetic.
Fatigue is audience-level decay against a concept, not against a crop. When someone has mentally tuned out an idea, showing them a squarer version of the same idea does not reset anything. Practitioner guidance through 2026 is blunt about this: generating more variations of a fatigued concept accelerates decline by exposing the tired idea to more people faster.
Format diversity still matters, separately. A static that works in feed does not work in Reels. A thirty-second in-stream edit does not work in Stories. You need both axes, which is exactly why the arithmetic is hard: concepts multiplied by formats, refreshed monthly.
The unit of production stopped being the film. It became the concept, and a concept has to arrive already cut for every surface it will live on.
Where the shoot day actually goes wrong
Almost every over-budget production we see went wrong before anyone arrived on set. The shoot was planned to capture one story beautifully. It needed to be planned to capture components.
Modular production means shooting with recombination as the goal: multiple hooks for the same body, several endings, product moments captured clean against neutral backgrounds so they can be dropped into edits that do not exist yet, and vertical framing treated as the primary composition rather than a crop applied in post. A day planned that way yields dozens of genuinely distinct concepts. The same day planned traditionally yields one film and a folder of resizes.
The hook is doing most of the work
Meta's Andromeda system weights creative signals more heavily than its predecessor, which pushes the decisive moment earlier in the asset than most brand teams are comfortable with. Hook rate is the first metric to move when something is failing, well before click-through or conversion register the problem.
This makes hooks the highest-leverage thing to vary and the cheapest. Six hooks against one body is six tests for barely more than the cost of one. Most teams do the reverse: they perfect one opening and then produce six bodies, which is the expensive way to learn less.
What good looks like
A queue, not a campaign. The teams that beat fatigue share one structural habit: the next variation is tagged and ready before the current one degrades. They are never producing in response to a problem, because by the time fatigue is visible in a dashboard, budget has already been spent on an asset that stopped working.
Judge the system on absolute winners rather than hit rate. A lower percentage across higher volume still produces more winners, and winners are what you are actually buying. At roughly twenty launches per winner, the number of winners you need each quarter tells you exactly how much to produce. Work backwards from that and the production plan writes itself.
None of this makes creative less important. It makes it more important, and considerably more industrial. The craft did not go away. It moved from the final cut to the design of the system that generates final cuts, which is a less romantic place for it to live and a far more valuable one.