Sell capacity, not deliverables, in a retainer
Asset-count retainers get gamed once variants are cheap. Build a capacity-unit retainer: what a unit buys, how unused units roll, and where the ceiling sits.
A retainer that promises twelve assets a month was a sensible contract when producing an asset meant booking a shoot. It is now an invitation to a specific argument: the client asks for twelve concepts, then twelve aspect-ratio variants of each, then a hook test with four openings per concept, and points out — correctly — that the contract says twelve assets and they have asked for twelve assets.
You cannot win that argument by tightening the definition of an asset. Every definition you write, the client's next request sits just outside it. The problem is the unit. Asset counts price an output that got cheap to produce and expensive to decide about, and no amount of clause-writing fixes a unit that measures the wrong thing.
Capacity units measure the thing that is actually finite: your team's month.
What broke, precisely
Three things happened at once.
Variants stopped being a cost. Generating four openings for the same concept is a small marginal spend. So "one asset" and "one asset with four hook variants" now sit at nearly the same input cost and wildly different amounts of your attention — review, selection, versioning, delivery, and the conversation about which one to run.
Selection became the bottleneck. Superside's Breakpoint research found around 80% of creative teams at or beyond capacity and roughly 70% of creative leaders burnt out, despite AI adoption. The constraint moved downstream. Filtering, judgement and brand governance are what fill the week now, and asset counts do not touch any of them.
Difficulty spread widened. An ambient b-roll clip and a hero product shot with legible pack copy are both "one asset." One takes a couple of attempts and one takes a fortnight of nobody being happy. A retainer that counts them equally is mispriced in both directions.
What a capacity unit is
A capacity unit is a fixed block of production effort, defined once, that the client spends however they like within stated rules.
Define it three ways at the same time — by effort, by an example, and by exclusion:
One production unit represents approximately half a day of studio production capacity. As a guide, one unit covers one Tier 2 deliverable including one revision round, or three Tier 1 deliverables, or four aspect-ratio variants of an already-approved deliverable. Units do not cover strategy workshops, paid media management, or deliverables requiring assets the client has not supplied.
The "approximately" is doing real work and should stay. The moment a unit becomes an exact measure, you have rebuilt asset counting with extra steps and the client will start auditing whether a particular job was really a full unit.
Underneath the client-facing definition, you need an internal conversion table so quoting is consistent. Build it off difficulty rather than runtime:
| Work item | Units |
|---|---|
| Ambient or atmospheric clip | 0.3 |
| Single-subject clip, no speech | 0.5 |
| Product interaction clip | 1 |
| Hero shot with real SKU and pack copy | 2 |
| Aspect-ratio variant of approved asset | 0.25 |
| Hook variant set, four openings | 0.5 |
| Additional revision round beyond the first | 0.5 |
| Recurring character across a sequence | Quoted separately |
That last row matters. Anything genuinely bespoke sits outside the unit system, quoted on its own, or one client's ambitious quarter consumes a retainer that was priced for routine work.
Sizing the unit and the month
Two numbers to set: how much a unit costs, and how many are in the month.
Work backwards from a monthly figure that makes sense for the client's tier. Freelance AI video retainers are commonly advertised from around $1,500 a month upward, and productized agency programmes sit well above that, so place the client on that spread rather than picking a number in isolation. Worked through: a mid-tier programme at $4,000 for 10 units puts a unit at $400. Check that against your own floor: half a day of studio time plus the credit spend for the attempts inside it, plus the review and delivery overhead nobody remembers to count.
Two sizing rules that save trouble later:
- Never sell more units than you can deliver in 60% of the month. Retainers do not fail on capacity in a normal month; they fail when two clients both have a heavy week. Build the slack in at the sizing stage, because you cannot add it later.
- Price the unit so the second block is an easy purchase. The upside of a capacity retainer is that overflow has a published price. That only works if the price is one the client can approve without a procurement cycle.
Rollover and the ceiling
Unused units are the first thing a client asks about and the fastest way to build a liability you cannot staff.
The rule that works in practice: units roll forward one month, capped at 50% of the monthly allocation, and expire after that.
That gives the client a real answer — their light month is not simply confiscated — while stopping the accumulation that ends with a client arriving in December holding four months of banked capacity and a campaign that needs all of it at once. The cap is the important half. Unlimited rollover converts a capacity retainer back into a liability with no delivery window attached, which is worse than the asset-count contract you replaced.
Two supporting clauses:
- No borrowing forward. Clients may spend rolled-over units, not next month's. Forward-borrowing is how a retainer becomes an interest-free loan of your time.
- Rollover does not survive termination. Say it explicitly. Otherwise a 30-day notice period arrives with a demand to burn five banked units in four weeks.
The other half of the guardrail is the upper bound. Without one, a capacity retainer is a promise of unlimited availability at a fixed price. Write three limits:
Monthly ceiling. Total units deliverable in a month, including rolled-over ones. Something like 150% of the base allocation. Past that, work queues into the following month rather than being refused, which keeps the conversation about scheduling rather than about willingness.
Turnaround floor. A minimum lead time per unit type. Units buy capacity, not instant capacity, and without a stated turnaround the retainer silently includes rush work. Rush is a separate line at a premium.
Concurrency limit. How many jobs can be in flight at once. This is the one people forget, and it is what actually protects the week — five simultaneous jobs each waiting on client feedback is more disruptive than eight sequential ones.
Reporting against units
The one genuine downside of capacity pricing is that the client loses the simple monthly proof they used to have. Twelve assets was a number they could see. Units are abstract, and abstraction erodes trust unless you replace it with something better.
Send a monthly statement with four lines: units allocated, units consumed with a one-line description of each, units rolled forward, and units expiring next month. That last line is what makes clients spend their capacity rather than hoard it, which is good for both sides — an under-consuming client is a client who cancels in month six because they cannot see the value.
Attach the output alongside it. Client reporting for content freelancers covers what belongs in the document beyond the ledger, and on your side tracking credits per client and per deliverable is what tells you whether a unit is still priced correctly. Versely bills every generation in credits with no free allowance behind it, so a client whose work skews toward high-attempt shots consumes more input cost per unit than one whose work does not — visible in the ledger long before it is visible in the P&L. The plan and credit pricing page is the input side of the unit price.
Migrating an existing retainer
Do not renegotiate the fee and the unit at the same time. Convert at the same monthly number the client already pays, sized so their historical average consumption lands comfortably inside the allocation. The pitch is flexibility, not savings: they can now spend their retainer on variants, on one ambitious piece, or on a burst of tests, without a scope conversation each time.
Change the price at the following renewal if it needs changing. Doing both at once turns a structural improvement into a price increase with a confusing explanation attached, and the client will assume the units exist to hide the rise.
Keep the revision policy exactly as it was through the transition. A revision policy that stops scope creep is orthogonal to the unit question, and changing two contract mechanics in one month makes it impossible to tell which one caused a problem.
FAQ
Do clients understand units, or does it just confuse the sale?
They understand them faster than the abstraction suggests, because most have bought something similar before — support hours, print allowances, cloud credits. The version that confuses is the one with no concrete examples. Always show the "one unit covers roughly X, or Y, or Z" line in the proposal; that sentence does most of the explaining.
What stops a client from spending every unit on the hardest possible work?
The conversion table. Hard work costs more units, which is the whole point — the system is difficulty-weighted, so a client who wants only hero shots gets fewer of them. That is honest pricing, and it is the conversation that asset counts never allowed you to have.
Is this better than a per-deliverable rate card?
Different, not better. Per-deliverable is stronger with spiky or unpredictable demand and with clients who buy occasionally; capacity works when demand is steady and the mix keeps changing. Per minute, per deliverable, or retainer walks the same month through all three units, and what goes in a local video retainer is the simpler scope-based entry tier for clients not ready for this structure.
How do I handle a client who consistently under-consumes?
Talk to them in month two, not month five. Under-consumption is a churn signal, not a windfall. The usual causes are that they do not know what to ask for or that approvals are stuck internally — both fixable, and both invisible if you are quietly banking the margin. How agencies scale client video output covers the demand-generation side of keeping a retainer full.