An ad creative approval process that survives a weekly test cadence has three properties: it approves sets rather than individual ads, it runs one pass instead of rounds, and it checks written constraints instead of taste. Budget about five minutes of review per output and put the whole batch in one calendar block. Once production stops being the expensive step, approval is the bottleneck, and nearly every approval process in use was designed for a world where six ads a month was a lot.
The symptom is recognisable. Production takes an afternoon, approval takes nine days, and the test that was meant to run weekly runs monthly. Nothing in that sequence is fixed by producing faster.
Why approval is now the bottleneck
A team shipping 40 published ads a month at a 60 percent keep rate produces 67. At five minutes of review each, that is 5.6 hours: an afternoon. Now add a second reviewer, a comment round, and a scheduling gap between the two, and the same 67 outputs consume three weeks of calendar time while using the same 5.6 hours of attention. The cost is not the reviewing. It is the queueing.
Keep rate, meaning published divided by produced, is what makes this concrete. It also means approval is not a neutral gate: it sets your production volume. If review rejects at 55 percent, you have to produce nearly twice as many ads to hit the same published number, and you pay for all of them. Keep rate is the production metric worth tracking precisely because approval controls it.
The cadence pressure is real too. Our 2026 ad fatigue benchmark puts CTR decline at 15 to 20 percent in a creative's first two weeks, with week three landing 45 to 70 percent below the launch baseline. A nine-day approval queue is not a process inefficiency at those decay rates. It is a media cost.
Approve sets, not ads
Per-ad approval charges you a fixed overhead per asset: reload the brief, reload the brand rules, reload the platform policy, form an opinion, write a comment. For one ad a week that overhead is invisible. For 67 a month it is most of the work.
Per-set approval reverses it. One brief produces one set of 10 to 14 outputs. The reviewer opens the whole set at once, holds the brief and the constraint list once, and makes one of three calls on each output: publish, regenerate, or cut. No comments, no threads, no "can we see it with the other music".
| Per-ad approval | Per-set approval | |
|---|---|---|
| Unit of decision | One asset | One brief's output set |
| Reviewers | Usually 2 or more, sequential | One, plus legal on the constraints block quarterly |
| Rounds | 2 to 3 | 1, with one regeneration pass allowed |
| Overhead per asset | 10 to 20 min of context reloading | Context loaded once per set |
| Calendar time for 67 outputs | 2 to 3 weeks of queueing | One afternoon |
| Typical reject reason | "Not quite right" | Named constraint or brief mismatch |
| Effect on keep rate | Drifts down, unexplained | Visible per set, improvable via the brief |
Two rules keep it honest. Review against the brief, not against your preferences, because taste-based rejection is unbounded and produces no lesson. And cap regeneration at one pass: an output that needs a second attempt is an output the brief failed to specify, and the fix belongs in next week's brief rather than in this week's queue. Our regenerate or edit guide covers which of the two is the right call, since editing an almost-right ad is often faster than another round of generation.
Move brand constraints into the brief
This is the change that does the most work, and it is unpopular because it moves effort earlier.
Anything that would cause a rejection should be written down before production starts: banned claims and comparatives, required disclaimers, the logo rule, music licensing, whether a competitor can be named, the aspect ratios, the language. Keep it as a standing constraints block. Legal approves the block once a quarter rather than approving ads one at a time.
The economics are one-sided. A constraint stated in the brief is an instruction that costs nothing. The same constraint applied in review bins a fully produced, fully reviewed asset and sinks every minute in it. Teams that make this one change typically watch keep rate move from somewhere in the forties into the seventies, without anyone getting better at editing.
A worked example. Forty published ads a month, review at five minutes an output, and a loaded hourly cost of about EUR 45 (salary plus overhead, not just salary):
| Keep rate 45% | Keep rate 75% | |
|---|---|---|
| Produced to publish 40 | 89 | 54 |
| Review hours at 5 min each | 7.4 | 4.5 |
| Review cost | about EUR 333 | about EUR 203 |
| Production hours at 1h each | 89 | 54 |
| Production cost | about EUR 4,005 | about EUR 2,430 |
The review line barely moves. The production line falls by more than EUR 1,500 a month, and the only thing that changed was where the constraints were written. Our video ad cost calculator will run this with your own figures.
What to stop reviewing entirely
A review that checks everything checks nothing carefully. Here is what I would delete.
- Which hook is best. You cannot tell, and neither can I. The test tells you. Review that the hook is compliant and on-angle, then let all the compliant hooks run. Picking hooks in review is the most confident wrong decision in creative ops.
- Music choice, unless it is a licensing question.
- Frame-accurate cut timing. Nobody in the feed notices a cut landing 4 frames late. Reviewers notice because they are watching for it.
- Whether it matches the last winner. This is how accounts end up with fourteen versions of the same ad and rising frequency.
What stays on the list is short: is the claim true and permitted, is the disclaimer present, is the offer stated correctly, is the audio intact, does it match the brief's angle, is it the right ratio and length. Six checks, each answerable in seconds by someone who is not guessing at intent.
The review time to budget
About five minutes per output, all in. That covers watching a 15 to 30 second ad twice, running the six checks, and recording the decision. It does not cover writing feedback, which is why per-set approval does not include feedback: the feedback goes into the next brief, in aggregate, once.
For planning: 12 outputs is an hour, 30 is two and a half, 67 is an afternoon. If your actual review time per output is 20 minutes, you are not reviewing, you are art directing after the fact, and the cost of that is the ads you did not ship. Our creative testing workflow sets out where the review block sits in a weekly cycle.
Where this gets harder
Client approval is the honest exception. If a client contract requires per-ad sign-off, you cannot batch your way out of it, though you can usually move to per-set sign-off with a named constraints block and a 48-hour default-approve window. Our agency creative workflow covers that negotiation, which is mostly a contract conversation rather than a process one.
Regulated categories are a genuine constraint. Finance, health and gambling claims often need documented per-asset review, and no process design removes that. What still works is the constraints block, because it keeps the reviewable surface small.
On tooling, we should be plain about our own product. Genyad handles the generation and the export side, and on self-serve plans there are no client-facing review seats, no client sub-accounts and no consolidated invoicing, so an agency running client approval through Genyad is exporting files and reviewing them wherever it already reviews. There is also no predicted performance score, which means nothing in the tool will tell you which output will win. The test does that.
Frequently asked questions
How long should reviewing a video ad take?
About five minutes per output, covering two watches and a fixed checklist of six items: claim accuracy, disclaimer, offer wording, audio, angle match, ratio and length. If review is taking 20 minutes an ad, the reviewer is making creative decisions rather than checking constraints. Feedback belongs in the next brief, not in a comment thread on this asset.
Should every ad get individual approval?
Only in regulated categories or where a client contract demands it. Otherwise approve at the set level, since one brief's worth of outputs share the same claims, angle and constraints, so reviewing them together removes the context reloading that makes per-ad review expensive. One accountable reviewer per set, one pass, one regeneration allowed.
Who should approve ad creative?
One named reviewer per set who owns the decision, with legal or brand approving a standing constraints block quarterly rather than signing individual ads. Sequential multi-reviewer approval is what turns 5 hours of attention into three weeks of calendar time. If two approvers are unavoidable, put them in the same sitting.
How does approval affect production cost?
Approval sets your keep rate, and keep rate divides every other cost. Rejecting 55 percent of outputs means producing nearly twice as many ads to publish the same number, so briefs, production and review all scale off the larger figure. Moving constraints from review into the brief is the cheapest way to change that ratio.