Scaling creative across multiple client accounts rests on three pieces of infrastructure: a strictly separate footage library per client, one brief template used identically on every account, and approval that happens per batch rather than per ad. Report publishable ads per week per client as the number that matters, not hours worked or assets delivered. The failure that ends client relationships is footage crossover, and it is worth engineering against before you optimise anything else.
Everything else about running creative for eight clients is a repetition problem. Crossover is a trust problem, and trust problems do not get a second attempt.
Keep client libraries separate, structurally
One client's footage appearing in another client's ad is not a quality issue. It is a confidentiality breach, and depending on the contract it is a breach with a termination clause attached. I have seen it happen twice, both times from the same cause: a shared "general B-roll" folder that somebody topped up with client footage because it was convenient.
Engineer against it in three ways.
Separate accounts or workspaces per client, not folders. A folder is a naming convention, and naming conventions fail. If your tooling supports a per-client boundary, use it even when it is less convenient.
Scope aggressively at the client level and loosely inside it. Within a client, over-scoping clips to individual products or campaigns shrinks the pool each brief can draw from, which produces variations that look like each other. Between clients, scope is absolute. Our guide to organising your library covers the internal version of this trade, and the rule I would apply is: scope hard at the client boundary, then leave generic footage unscoped inside it.
Never build a shared stock pool that mixes sources. If you want generic B-roll available to everyone, licence stock footage for it and keep it in its own library with the licence documented. Do not let client-shot material drift into it.
| Asset | Where it lives | Who can use it |
|---|---|---|
| Client-shot footage, any kind | That client's library only | That client's campaigns |
| Client testimonials and customer faces | That client's library only | That client's campaigns, and check the release form |
| Licensed stock you bought | Agency stock library | Any client, licence permitting |
| Agency-produced generic B-roll (hands, screens, streets) | Agency stock library | Any client |
| A winning ad's structure or hook pattern | Documentation, not the library | Any client. Patterns are not assets |
That last row is the one people get wrong in the other direction. Reusing a hook structure across clients is legitimate and it is a large part of what an agency is for. Reusing a client's footage is not. Keep the pattern library in writing, separate from the media.
One brief, used identically on every account
The reason to standardise is not tidiness. It is that a standard brief lets you compare keep rate across clients, which is the only way to find out whether a low keep rate is the account or the person.
Five fields, and the same five on every account:
- Offer. The specific thing the viewer gets. "Learn more" is not an offer, and it is the most common cause of weak output across an agency's whole book.
- Audience and their objection. One line each. The objection is what makes hooks different from one another.
- Angle. The single claim this batch argues. One angle per brief, three briefs for three angles.
- Constraints. Banned claims, required disclaimers, logo rules, competitor mentions, ratios, language. This block is per client and signed off once a quarter.
- Format. Length, ratios, placement, language, caption policy.
The constraints field is where the per-client variation lives, and keeping it as a standing block that the client's legal or brand contact approves quarterly is the single biggest saving available to an agency. A constraint stated in a brief is an instruction. The same constraint discovered in client review is a binned asset plus a delay plus an email thread.
Standardising also makes onboarding a new client a two-hour job: fill the constraints block, fill the offer, index the footage. Our agency creative workflow page walks the full sequence.
Batch approval with clients
Per-ad client approval does not scale past two or three accounts, because the cost is not the reviewing, it is the queueing. Five hours of client attention spread across sequential email threads becomes three weeks of calendar time, and at that point creative is decaying faster than you can ship it. 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, so a three-week approval loop means launching ads that are already halfway through their useful life.
What works, in order of how hard it is to negotiate:
- Per-set approval. One brief produces one set. The client approves or rejects the set, with individual outputs flagged rather than commented on. Publish, regenerate, cut.
- A default-approve window. 48 hours, stated in the contract. Silence means approved. This one sentence removes more delay than any process change.
- A quarterly constraints sign-off instead of per-ad brand checking.
- A standing exclusion list so the client tells you in advance what will be rejected rather than discovering it live.
- One named approver per client. Two approvers is a queue. Three is a committee, and a committee produces the ads nobody objects to, which are also the ads nobody notices.
Regulated clients are the honest exception. Finance, health and gambling often require documented per-asset review and you cannot design that away, though a tight constraints block still keeps the reviewable surface small.
Report publishable ads per week
Most agency creative reporting counts the wrong thing. Assets delivered rewards volume that nobody launched, and hours worked rewards slowness. The metric to put at the top of the client report is publishable ads per week, alongside keep rate.
Keep rate is published divided by produced, and it is the number that governs your true cost per ad on every account. If a client publishes 60 percent of what you produce, delivering 40 launched ads means producing 67, and everything you spend scales off 67. Track it per client and the pattern is immediately useful: a client whose keep rate is 45 percent has a briefing problem or an approval problem, and either way it is costing you the difference. Keep rate is worth measuring properly before you take on the ninth account.
A weekly report I would actually send, per client:
| Line | Client A | Client B | What a bad number means |
|---|---|---|---|
| Produced | 18 | 22 | Your cost base for the week |
| Published | 14 | 10 | Actual throughput |
| Keep rate | 78% | 45% | Under 60% is a brief or approval problem, not an editing one |
| Days from brief to live | 4 | 11 | Over a week and creative decays before it launches |
| Live variations in rotation | 12 | 6 | 8 to 20 is what an active campaign usually needs |
| Frequency on the top ad | 2.3 | 3.4 | Above about 3 means fatigue has already arrived |
Client B in that table is not an underperforming account. It is an approval bottleneck with a decay problem attached, and the fix is a default-approve window and a constraints block, not more production.
What the tooling does and does not cover
We make Genyad, so treat this as a disclosed interest. It handles the per-client separation and the production side: a library per client, indexing and transcription free, one credit per variation (about EUR 1.52 on the EUR 99 Growth pack), exports at 9:16, 4:5, 1:1 and 16:9 with no watermark, and scripts written natively in English, German, French, Spanish, Italian or Hindi rather than translated, which matters if your clients run multiple markets. Credits do not expire, which suits agency work where volume is lumpy. Our pricing page has the packs and there is a free tier of 5 variations without a card.
What it does not have, and this matters most for agencies specifically: there is no agency console with client sub-accounts, no consolidated invoicing, and no client-facing review seats on self-serve plans. So client approval happens wherever you already do it, and per-client separation means a deliberate library structure on your side rather than a feature that enforces it for you. There is also no AI avatar, no static banner output, no product-URL import, no feed or CSV template rendering, no predicted performance score, and no direct publishing to Meta or TikTok. If your agency's bottleneck is rendering 500 feed-driven variants, a template renderer is a better fit than we are.
Frequently asked questions
How do agencies keep client footage separate?
Use a separate library or workspace per client rather than folders inside one library, because folder conventions fail as soon as someone is in a hurry. Keep any generic B-roll in a licensed agency stock library that no client footage ever enters. Reusing a hook structure across clients is fine and is part of the value an agency adds; reusing a client's footage is a confidentiality breach.
Should every client approve ads individually?
Only where regulation or the contract demands it. Otherwise approve per set, with one named approver, a 48-hour default-approve window written into the contract, and a constraints block signed off quarterly instead of brand checks on every asset. Sequential per-ad approval is what turns five hours of client attention into a three-week delay.
What should an agency report to clients about creative?
Publishable ads per week and keep rate, plus days from brief to live, live variations in rotation, and frequency on the top ad. Assets delivered and hours worked both reward the wrong behaviour. A keep rate under 60 percent is a signal about briefing or approval, and it is worth raising with the client rather than absorbing.
How many client accounts can one creative strategist handle?
One to two accounts at a weekly test cadence, or up to three if they share a product category and offer set, because each account needs roughly five to six hours a week of briefing, batch review and reading results. Standardising the brief and the constraints block is what makes the higher end possible. The first sign the ratio is too thin is briefs that recycle last month's angle with a new hook.