Creative is the only lever left
Platforms have automated much of the media buy. Targeting, bidding, placement and delivery now sit largely inside the ad system's own model, and the guidance from Meta and Google is to give that model room to work. Creative is the input the advertiser still supplies directly.
That happens to be the input carrying the most weight. NCSolutions studied nearly 450 consumer goods campaigns and found the creative itself drove about 49 per cent of the incremental sales an ad produced, ahead of brand at 21 per cent, reach at 14 per cent and targeting at 11 per cent. The platforms agree in their own documentation. Meta's guidance on creative diversification says a wider range of creative is how you get the most out of its automated ads system. Google's advice on Performance Max assets is that the more assets, the better.
The platforms publish fatigue signals and refresh guidance
Creative does not fail on a universal timetable. The platforms do publish clear warning signs and refresh guidance. Meta marks an ad as fatigued when its cost per result reaches twice what your past ads paid, and it warns you when it predicts that inside the first seven days of a campaign. TikTok's creative guidance is more direct about the remedy: when it detects fatigue, which it puts at roughly every seven days, add three to five new creatives. This is a property of audiences rather than of one auction. A meta-analysis of advertising repetition across decades of studies found response peaks after a handful of exposures and decays after that. TikTok's guidance also covers targeting, bidding and budget, so creative is the largest lever available rather than the only one.
The arithmetic at enterprise scale
Those numbers are written for a single advertiser running a single line of activity. An enterprise runs many at once, and each fatigues on its own clock. Six markets with three audiences each is eighteen live ad sets.
If all eighteen ad sets reached TikTok's typical fatigue interval in the same week, following its guidance would require 54 to 90 new creatives. Across a thirteen-week quarter, that is 702 to 1,170 assets. Not every ad set will fatigue on the same schedule; this is a planning stress test that shows how quickly creative demand compounds.
It is still worth writing down the other side of it: how many finished assets your current production contract delivers in the same quarter. For most enterprise brands the two numbers sit an order of magnitude apart. The gap shows how much of the media plan depends on existing creative lasting longer than platform guidance suggests.
Why the supply side stays flat
The ceiling is set by what a single asset costs to make and clear. When a video carries four-figure production cost and a two-week review chain, the asset count gets fixed in the plan at whatever the budget and the calendar allow, and it does not move when media spend goes up. We covered the cadence side of this in Post like a small brand and the review chain in Approve the workflow once.
The point here is narrower and specific to spending at scale. A fixed asset count turns into a ceiling on what a growing media budget can buy. Adding spend against a fixed creative pool increases frequency and raises the risk that cost per result will climb.
Raising the ceiling
The way out is to move the cost out of the per-asset loop. Brand rules, approved claims, tone and compliance constraints get encoded into a Creative Workflow once and reviewed once, after which each additional variant inherits its guardrails and costs very little to produce.
Mr. Brushable, a touch-up paint brand, shows what that does to supply. Eleven finished ads for that one product were produced on our platform across four days in August, and the three at the top of this post are among them. All three carry the same product and the same encoded rules. Each one opens differently. Producing eleven takes only makes sense when a take is cheap, and a review pass then decides which of them run.
Where those variants go matters as much as how many there are. The opening matters disproportionately. Meta's work with Nielsen across 173 BrandEffect studies found viewers who watched under three seconds produced up to 47 per cent of a campaign's total value, and TikTok reports that 63 per cent of its highest click-through ads show the product or the message within the first three seconds. So the cheapest variant worth having is a new opening on an asset that has already passed review. A workflow that produces twenty openings from one cleared cut gives twenty different chances to find a winner from the same approved production.
What to measure
Three numbers tell you where you stand, and none of them require a new tool to collect.
| Measure | What it tells you |
|---|---|
| New creatives per live ad set per week | Your actual replacement rate, against the three to five TikTok suggests |
| Share of spend on creative more than 30 days old | A rough exposure-risk proxy, at about four times the seven-day refresh interval TikTok suggests |
| Cost per finished asset, and time from brief to live | Whether the ceiling itself has moved |
The first two can be pulled from Ads Manager this afternoon. The third is the one a pilot is meant to change, and it is the fair way to judge one.
Where to start
Take the markets and audiences on next quarter's media plan, multiply them by the replacement rate in TikTok's guidance, and set the result against the number of finished assets your production line is contracted to deliver in the same period. That ratio measures creative-supply coverage: how much of the plan is carried by new creative and how much depends on existing assets running longer than the guidance suggests. Every other argument about the media plan sits downstream of it.