AI Ad Credits or Finished Ads? A Cost-per-Approved-Ad Framework for European Agencies (2026)
Short answer
The price of a generation credit tells you very little about what an AI ad costs your agency. Credits buy attempts. Your client pays for ads that run. The number to track is cost per approved ad: everything you spend to get from a brief to an ad that has passed your QA, your client’s sign-off and the platform’s review, divided by the number of ads that actually run. Once you have that number, choosing between running AI tools in-house, working with a white-label AI ad production partner or mixing both becomes a question of control, compliance, capacity and predictable pricing, not a comparison of credit packs.
Why the credit price is the wrong number
AI video tools are usually sold by credits, minutes or seats. That makes them easy to compare with each other and hard to compare with anything else. A credit pays for a generation, and most generations never reach a client. Some takes have the wrong hands, a drifting product or an off-brand voice. Some survive selection and fail in editing. Some pass editing and come back after client review. A few get rejected or restricted by the platform.
None of this is a flaw in the tools. It is how generative production works: you generate wide, then narrow. The cost sits in the narrowing, and it is paid in your team’s hours, not in credits.
What counts as an approved ad
Agree on a definition before you measure anything. For this framework, an ad is approved when it has:
- passed your internal quality check (product fidelity, on-screen text, sound, specs);
- been signed off by the client, after the revision rounds the brief allows;
- cleared the platform’s ad review in every format and ratio ordered;
- a documented disclosure decision, in case anyone asks later.
A draft that looks great in a review link but isn’t approved yet doesn’t count. Neither does a 9:16 cut whose 4:5 version still needs work.
The formula
Cost per approved ad = (credits + prompting and selection time + editing + QA + compliance documentation + revisions + platform-rejection risk) ÷ number of runnable ads
| Cost line | What to count | Where it usually hides |
|---|---|---|
| Credits | Subscriptions, credit packs and top-ups used for the batch | On a card nobody reconciles per client |
| Prompting and selection | Hours spent writing prompts, generating variants and choosing takes | Logged as “creative” or not logged at all |
| Editing | Cutting, pacing, captions, sound, colour, format adaptations | Spread across several people’s weeks |
| QA | Shot-by-shot product checks, text checks, platform specs | Done in a rush the day before launch |
| Compliance documentation | The disclosure analysis, consents and the record of what ran | Often skipped until a client asks |
| Revisions | Every round after the first client review | Absorbed as goodwill |
| Platform-rejection risk | Rework and delay when an ad is rejected or restricted | Lost flight days and media budget held back |
How to fill it in with your own numbers
Use a real batch you have already shipped, not an estimate. Five steps:
- Count the runnable ads. Only the ones that ran, per format. This is your denominator, and it is usually smaller than people remember.
- Pull the tool spend for the period and allocate it to the batch.
- Log hours by role for prompting and selection, editing, QA and documentation. Ask the people who did the work rather than reading the plan.
- Price those hours at your internal loaded cost, meaning salary plus overheads, not your client day rate.
- Add revisions and rejections. Count the rounds beyond the first and the ads that had to be reworked after platform review, then price that time the same way.
Divide the total by step 1. Run the same exercise on two or three batches and the figure becomes stable enough to quote from.
The lines teams forget
Selection is a job, not a click
Choosing the right take among dozens is skilled work. It needs someone who knows the brief, the brand and the platform. When that person is your senior creative, the hour costs more than it looks on the timesheet.
Product fidelity needs a second pair of eyes
Generated products drift between shots: a colour shifts, a logo warps, a finish looks better than the real thing. Catching this takes a deliberate shot-by-shot check against the real product. Skipping it is how an ad gets past review and then misleads the people who see it.
Compliance takes documentation, not just a decision
Since 2 August 2026, realistic AI-generated people, places or events in ads running in the EU must be disclosed by the deployer under Article 50 of the AI Act. Each ad therefore needs an analysis and a record. We cover the six checks in AI-generated UGC ads: 6 checks before you deliver a creative.
Platform review is part of the cost
TikTok, for example, says that undisclosed AI-generated content leads to an ad being “rejected or restricted”, and it also refuses ads whose content doesn’t match the landing page (TikTok). Each rejection means rework, a resubmission and, often, a campaign that starts late.
In-house, partner or hybrid: a decision matrix
Once you know your cost per approved ad, compare the three set-ups on the criteria your clients notice.
| Criterion | In-house with AI tools | White-label partner | Hybrid |
|---|---|---|---|
| Creative control | Full, if you have the senior time | Shared: you approve the angle, script and storyboard | You keep strategy and hooks, the partner produces |
| Compliance record | You build and maintain it | Delivered with each ad, you confirm the disclosure | Split by who produced each ad |
| Capacity at peaks | Limited by your team’s hours | Scales with the order | In-house for quick variants, partner for volume |
| Cost predictability | Variable: depends on takes, revisions and rejections | Fixed price per order | Partly fixed |
| Skills to hire or train | Prompting, editing, QA, compliance | Briefing and client management | Briefing plus light editing |
| Confidentiality | Native | Contractual: your brand on every file | Contractual for the partner share |
When in-house makes sense
You already have an editor and a creative lead with spare hours. The work is mostly simple variants of an existing ad, such as new hooks, end cards or languages. Realism and product fidelity are low risk.
When to outsource AI UGC ads
Briefs arrive in waves. Your senior team is the bottleneck. Clients ask for realistic presenters or product videos where fidelity and disclosure carry real weight. Or you want to quote a fixed price to your client and keep it.
When hybrid makes sense
Most agencies end up here. The agency owns the strategy, the hooks and the client relationship, and runs quick iterations in-house. The partner produces the ads that need a full production pass, with QA and documentation included.
The landing page has to keep the same promise
An approved ad that points to a page with a different offer, product or tone loses in two places. Platforms can refuse it for inconsistent information, and the visitors who click don’t convert. When the ad and the page are briefed together, the promise stays the same from the first second of the video to the form. One contract can cover both: the ad, and the landing page through our website production.
Reselling finished ads to your client
When you buy finished ads at a fixed price, the margin question becomes simple. Price the ad per approved deliverable, not per hour, and keep the revision policy you agreed with the partner. Our white-label margin method walks through how to set a markup per service. If the same client also runs paid campaigns with you, our online advertising team can run them under your brand.
How Genesis Digital Factory fits in
We produce AI video ads for agencies across Europe, in white label: UGC, product videos, image ads and, on request, cinematic spots. Human creative direction approves the script and storyboard before generation. Editors finish every ad, and quality control can send an ad back before it reaches you. Each ad ships with its EU AI Act compliance report, and a synthetic presenter never claims a lived experience. Every order is quoted at a fixed price within 48 hours, with kick-off within 72 hours. We have delivered more than 85,000 projects since 2005 from our production studios in Tunisia. See the full range of white-label production services.
Get a quote for AI ad production
Want to measure us against your own numbers? Run a trial project and put it through the formula above.
FAQ
Isn’t this framework just a case for outsourcing?
No. Plenty of agencies will find that in-house tools work well for simple variants. The point is to compare on cost per approved ad, control and risk rather than on the price of a credit pack, whatever the answer turns out to be.
How many batches do we need to get a reliable figure?
Two or three real batches are usually enough to see a stable range. Use shipped work, count only ads that ran, and price hours at loaded internal cost.
Can we keep our own tools and still work with a partner?
Yes. That is the hybrid model. Decide upfront which ads each side produces, and who documents the disclosure for each one, so nothing falls between the two.
Read also: AI Is Changing Ads by Default: What Agencies Should Check in Client Ad Accounts. Version française : Crédits de génération IA ou pubs livrées ? Un cadre de coût par pub approuvée.
