Pricing exploration separately from finish
Generation got cheap and finishing did not. A two-phase quote with a selection gate, so a client's appetite for options stops eating your finishing capacity.
The pattern in every AI campaign that actually landed is the same: generation for exploration, human craft for finish. Nike's Serena Williams project ran an enormous volume of simulated matches to find the material; the finished film was still finished by people. Coca-Cola's holiday campaign compressed a timeline that had run about a year into roughly a month, and drew significant public backlash alongside the speed gain.
The commercial problem is that those two activities have moved in opposite directions. Exploring twenty directions is now a smaller cost than it was three years ago. Finishing one of them — grade, sound, typography, cleanup, continuity, the fourteen small decisions that make a cut look deliberate — costs roughly what it always did, because it is still someone's afternoon.
Quote both under one number and you have built a subsidy. The cheap half funds the expensive half, the client optimises for the half that feels free to them, and your finishing week gets eaten by options nobody is going to use.
Why the blended quote fails specifically
A single project fee sends the client one signal: everything inside the scope has zero marginal cost. So the rational client asks for more options. Not because they are difficult — because you told them options were included, and more options genuinely does improve the odds of finding something good.
What they cannot see is that the twenty-first direction does not cost you a generation. It costs you a review pass, a shortlist conversation, an approval cycle, and a slice of the finishing week that was already fully booked.
That relocation shows up in the industry-side numbers. Superside's Breakpoint research found 80% of creative teams at or beyond capacity and 70% of creative leaders reporting burnout after AI adoption rather than before it. The commercial reading is narrower than the headline: the constraint did not disappear when generation got cheap, it moved onto the desk of whoever has to choose between twenty options and finish one of them. That desk is the one nobody itemises on a quote.
A two-phase quote puts a price on the relocated constraint.
Two phases, two units
Phase 1 — Exploration. Unit: the direction. A direction is a distinct creative approach, not a variant of one. Three tones of the same concept is one direction; three concepts is three. Priced per direction, delivered as previews, deliberately unfinished.
Phase 2 — Finish. Unit: the finished deliverable. Priced per asset at your difficulty tier, covering everything that turns a selected direction into something shippable.
The two are separately quoted, separately invoiced, and separated by a gate the client has to walk through on purpose.
| Belongs in exploration | Belongs in finish |
|---|---|
| Concept and framing variants | Colour and grade |
| Model routing tests | Sound design, mix, music |
| Rough sequence, no polish | Typography and lower thirds |
| 480p preview passes | Final export at delivery resolution |
| Style and tone comparison | Continuity fixes between shots |
| Shortlist and recommendation | Cleanup, retiming, stabilisation |
| Written finding on what works | Versioning, aspect ratios, handoff |
The line between them is one question: does this activity narrow the field, or does it improve one thing that has already been chosen? Narrowing is exploration. Improving is finish.
The gate
The gate is the whole mechanism. Without it, exploration bleeds into finishing and you are back to one blended blob with two invoices.
The gate is a written selection: the client names which direction proceeds, in writing, and that selection is the input to the Phase 2 quote. Not a call, not a Slack thumbs-up on a preview link — a named direction, dated.
Three rules make it hold:
- Phase 2 is not quoted until the gate closes. You cannot price finishing until you know what is being finished, and quoting it early gives the client a number to anchor on while the brief is still moving.
- Reopening exploration after the gate is a new Phase 1 block, at the published per-direction rate. This is the clause that stops the sixth round of "actually, what if we tried…" from landing inside a fixed finishing fee.
- Exploration deliverables are explicitly not shippable. Say so in the quote. Previews are 480p, unfinished, and not licensed for distribution. This matters more than it sounds — clients who see a good rough cut will ask whether they can "just use that one," and the honest answer is that it is not finished, not graded and not the delivery resolution.
Running the gate on preview links rather than file transfers keeps the whole thing tidy; a client review loop built on previews and share links is the operational version.
Pricing each phase
Exploration prices off directions and a time box. A block of, say, five directions delivered in three days is one line with one number. It is deliberately affordable, because its job is to be an easy yes and to make additional exploration a visible purchase rather than an invisible favour.
The input cost is real but forecastable: Versely bills every generation in credits with no free allowance behind it, and exploration is where the volume lives. The one exception is the editor's 480p preview pass, which is free and carries a short per-user cooldown, so iterating on a rough sequence before committing to a final export does not consume export charges. The final export is charged once regardless of clip count, which is why an exploration phase run through the editor is cheaper to operate than one run as a series of full-resolution renders. Editor previews and final export sets out the shape.
Finish prices off the deliverable and the difficulty tier, and it should be the larger number of the two. If it is not, one of two things is true: you are underpricing craft, or you are doing exploration work that should have been billed in Phase 1.
The ratio between the phases is a useful diagnostic. On a healthy short-form project, exploration is typically the smaller line by a clear margin. When exploration approaches or exceeds finish, either the brief was unusually uncertain — legitimate, and worth naming — or the client is using Phase 1 as a substitute for having a point of view, which is a scoping conversation.
What this does to revisions
Splitting the phases quietly fixes the revision problem, because it gives you two different revision definitions instead of one that fits neither.
In exploration, a revision is a new direction, and it has a published price. There is no meaningful "revising" a rough concept test; you are either exploring more or you are done.
In finish, a revision is a change to a selected deliverable, capped by round in the usual way. That is the version a revision policy was designed for, and it is much easier to enforce when the client has already signed a selection naming what is being revised.
The failure that a blended quote produces — round five arriving as "let's go back to concept three" — becomes structurally impossible. Concept three is a Phase 1 line item, and going back to it is a purchase.
Where the finishing work actually is
If you have never itemised Phase 2, it is worth doing once, because the list is longer than most people's instinct and it is the argument for the price.
Continuity between separately generated clips. Grade matching across shots that were never lit together. Sound, which generated output either does not provide or provides generically. Typography, which is usually composited on the timeline rather than generated, for good reason. Aspect-ratio versions. Resolution decisions — whether to upscale the render or generate at full resolution is a finishing choice with real cost implications. And handoff, if a human editor is taking it onward: handing generated clips to an editor cleanly and layer separation are both Phase 2 work that a blended quote never named.
None of that got cheaper. All of it is what the client is actually paying for, and a two-phase quote is the first document most clients see that says so.
FAQ
Does the client actually accept two invoices?
More readily than one large one, in most cases. Two phases with a decision point between them looks like a managed process, and the first number is smaller than a single blended fee, which makes it easier to approve. The clients who resist are usually the ones whose procurement needs a single PO — issue one PO with two milestones and the structure survives intact.
What if the client wants to skip exploration entirely?
Then they have a direction already, and you go straight to Phase 2 at the tier rate. Write down what they specified, because "we already know what we want" has a way of becoming "we want to see some options" in week two, and at that point it is a Phase 1 block rather than a scope dispute.
How many directions belong in one exploration block?
Between three and six for most short-form work. Below three you have not explored, you have confirmed a preference. Above six the client cannot hold them in their head, the shortlist conversation gets worse rather than better, and you are generating options to look busy. If a brief genuinely needs more, sell a second block.
Can exploration be run at final quality to save time later?
Rarely worth it. Finishing a direction that gets rejected is the exact waste this structure exists to prevent, and rough output is often better for the selection conversation because nobody argues about a colour grade that visibly is not finished yet. Keep exploration at preview quality and let the gate decide what earns the finishing pass — the editing timeline is re-renderable, so nothing you build in Phase 1 is thrown away when the selection lands.