You're buying creator ads at the wrong unit cost
Creator-made ads get priced per asset and judged per ad. Four numbers — hook rate, hold rate, CPA delta, re-cut cost — decide which ones you scale.
Creator-made ads are bought per asset and judged per ad, and neither is the unit that decides whether to scale one. The unit is cost per surviving variant: what you actually paid, in creator fees plus edit hours, for each cut that reached volume in the ad account. Four numbers get you there — hook rate, hold rate, CPA against the studio baseline, and re-cut cost per surviving variant — and three of them are already in your reporting.
The market signal behind the question is unusually clear this week. Digiday reported on November 20 that brands paying creators to produce their campaign creative has become a primary driver of creator-economy investment, citing a CreativeX study of 115,990 campaigns representing $11 billion in media spend. The same study puts a number on something most media teams feel and rarely measure: assets attached to a campaign are up 30% over five years, and in the average campaign just five assets absorb 88% of the media investment.
Sit with that second number. You are producing more creative than ever, and the delivery system concentrates almost all of the money into five pieces. Which means the interesting question is not "how do we make more creator ads". It is "how do we find the one that survives, and what did it cost us to get there".
The asset you buy is not the unit that runs
A creator deliverable is one file with one aspect ratio, delivered against one brief. What lands in the ad account is usually three or four variants: a vertical cut, a square, a shorter hook-first version, sometimes a text-led still pulled from the frames. Those edits exist because placements demand them, and each one costs editing time.
So there are two prices for the same piece of content, and they usually disagree. The first is the number on the creator's invoice. The second is the invoice plus the editing plus the media that was spent proving some of those cuts do not work — divided by the cuts that did.
The second number is the one a media buyer can defend in a budget meeting. The first is the one that gets quoted because it is easy to find.
One — hook rate, and why the denominator matters more than the benchmark
Hook rate is not a platform metric. Meta does not report it. What Meta reports is 3-second video plays, and your hook rate is that number divided by impressions. TikTok gives you a two-second and a six-second view, which are different denominators again. Any tool that shows a hook rate percentage is doing that division for you, sometimes against a different base.
The definition matters because the three-second play itself is a low bar: Meta counts a play that reaches three seconds, with no requirement that anyone tapped the video. A hook rate built on autoplaying impressions is measuring how well the first frame interrupted a scroll, not how well the ad was received.
That is still useful — it is the only cheap signal you get about the first two seconds, and creator content usually wins it, because a person talking into a phone camera looks less like an ad than a product on a seamless backdrop. The trap is comparing your creator cohort's hook rate to a benchmark you found online. Compare it to your own studio cohort, same placements, same window. A number that moves when you change the placement is not a property of the creative.
Two — hold rate breaks when the durations differ
Hold rate is usually calculated as ThruPlays divided by 3-second plays. Meta defines a ThruPlay as a play to completion, or for at least 15 seconds; if the video is 15 seconds or shorter, it counts only on completion.
Read that definition twice, because it hides an asymmetry that quietly decides creative arguments. For an eight-second creator cut, a ThruPlay requires someone to watch the whole thing. For a forty-second studio spot, a ThruPlay requires 15 seconds — a bit over a third of the runtime. Two videos with identical audience interest will produce very different hold rates, and the longer one gets the easier threshold.
Which means the standard hold-rate comparison systematically flatters long assets and penalises short ones. In practice, creator cuts are short, so they show up at the bottom of a hold-rate table and get cut for a measurement artefact. Before you retire a creator asset on hold rate, check the runtime next to the studio asset you are comparing it to. If the difference is more than a few seconds, you are not comparing attention, you are comparing how much video each one had left to give.
Three — the CPA delta only counts inside one objective
This is the number that actually decides scale, and it is the one most often computed wrong. Cost per acquisition for the creator cohort versus the studio cohort only means something when both run against the same optimization event, in the same placement mix, over the same window, on comparable audiences. Change the objective and you have changed the population, not the creative.
When the comparison is clean, the interesting case is the mixed result: creator asset wins the hook rate and loses CPA. The instinct is to kill it. Check the destination first. A creator cut can carry the click brilliantly and hand a cold visitor to a page written for someone who already wanted the product — that is a landing problem wearing a creative costume. The pattern is well documented at the top of the funnel: the most-clicked video ads frequently return the worst ROAS, which is a reason to distrust click-derived metrics, not a reason to trust the hook and stop measuring the rest.
The other mixed result is rarer and worth naming: creator asset loses the hook and wins CPA. Usually retargeting, usually a face people recognise, usually worth keeping and not worth re-cutting for more placements.
Four — re-cut cost per surviving variant
This is the number almost nobody keeps, and it is the one that decides whether creator production is genuinely cheaper than a studio shoot.
Add the creator fee to the editing hours, per placement variant. Then divide by the variants that passed a spend threshold — pick one and keep it, so that a cut which ran for a day and died does not dilute the denominator. What you get is a real unit cost for usable creative.
Run it honestly and the conclusion is often uncomfortable. A cheap creator license that needs four edits, of which one survives, can land above a studio asset that ships once and runs. That does not make creator content a bad buy. It means the saving is in the licensing and the volume, not in the post-production, and that a creator programme without an editing budget is a programme that will produce one usable cut per shoot and then stall.
The decision matrix
| Hook rate vs baseline | CPA vs baseline | What to do | Why |
|---|---|---|---|
| At or above | At or below | Scale, and re-cut for the remaining placements | The asset earned placement variants; spend the edit hours |
| At or above | Above | Diagnose the destination and the offer before touching budget | The creative is doing its job; the drop is downstream |
| Below | At or below | Keep, do not re-cut | It converts on a weak hook — usually a familiar face in retargeting |
| Below | Below on spend but above on CPA | Retire and log the reason | Two failures on one asset is a brief problem, not an editing problem |
The last row is the one that gets skipped. Retirements are cheap to do and expensive to forget: without a reason attached, the same brief gets rewritten six months later and produces the same asset.
More variants is not the answer
The instinct in every creator programme that starts working is to buy more. The CreativeX data reported by Digiday suggests the ceiling is not in production volume: five assets take 88% of the investment in an average campaign. Delivery concentrates on the pieces it can read, and it reads a small number of them.
For a media team, that reframes the job. You are not trying to fill a library. You are trying to shorten the distance between "we bought twelve creator cuts" and "we know which one carries the account" — and then to produce the next version of that winner cheaply.
Run the read in one pass
- Label the cohorts before launch. Creator assets in their own ad sets, or at minimum with a naming convention that survives reporting exports. Untagged cohorts are unrecoverable after the fact.
- Pull the placement breakdown for both cohorts, same date range.
- Compute hook rate as 3-second plays divided by impressions. Compute hold rate as ThruPlays divided by 3-second plays, and put the runtime next to it.
- Compare CPA only within one objective and one optimization event.
- Add the creator fee and the editing hours for every variant, then divide by the variants that passed your spend threshold.
- Write down what you killed and why, in one line each.
Six steps, one sitting, no new tooling. The value is not the dashboard. It is that the next creator brief carries the cost of the last one.
The real reason creator programmes die quietly
They rarely die because the ads underperformed. They die because nobody computes the replacement cost, so when the batch runs out, the media team has spend data and no unit economics, and finance sees a line item that looks like a discount on production with no evidence attached.
The four numbers above are not a measurement stack. They are the argument you will need in month four, when someone asks why you are buying creator assets instead of booking another shoot. Have the cost per surviving variant ready, and the answer is arithmetic.
If you are running creator and studio creative in the same account and cannot say which one is cheaper per usable asset, that is the audit worth booking — we do it as part of paid media and digital ads work, and it usually starts with the export you already have. If you are still deciding which creator assets are worth paying to license at all, the selection and licensing framework is a separate problem, and the sound and music question sits right behind it.
Frequently Asked Questions
What is a good hook rate for a creator-made ad?
There is no universal number, and anyone quoting one to two decimal places is selling something. Hook rate is 3-second video plays divided by impressions, so it moves with placement, format and audience temperature before it moves with the creative. The only useful comparison is against your own studio baseline, on the same placement mix, the same objective and the same window. Treat it as a relative signal inside one account, never as a benchmark across accounts.
Why is cost per asset the wrong way to price creator content?
Because most assets never run at scale. A creator deliverable usually becomes several edits — vertical, square, a six-second cutdown — and only some of those get real delivery before the algorithm concentrates spend. If you price per asset you are paying for the ones that never ran. Price per surviving variant instead: total creator and edit cost divided by the cuts that actually passed a spend threshold.
Can you compare creator-made ads against studio ads in the same campaign?
Only inside the same objective and optimization event. A creator cut running on an engagement objective and a studio cut running on a purchase objective produce two different cost-per-result numbers from two different populations, and comparing them tells you nothing except which campaign structure you built. Keep both cohorts labelled and compare them within one objective, one placement breakdown and one date range.



