One production quote promises senior human direction. Another promises automated speed. A brand commissioning content from a Chinese production partner needs more than a choice between those descriptions. What improves in the work, who owns delivery problems, and does the brand have a specific requirement about how the content is made?
ONCE proposes three separate reasons to purchase human involvement: an improvement to the content, responsibility for delivery, and an explicit production-origin requirement. Each deserves its own explanation and acceptance method. Neither AI use nor human participation determines the right price by itself.
A research result that does not price advertising services
The October 12, 2026 preprint Creative destruction: Human-AI competition in an experimental market, by John Conlon and Peter Schwardmann, studies US participants making computer games with an AI coding agent and consumers choosing games. Disclosing human involvement did not increase willingness to pay in that experimental market.
The authors describe a compressed setting: production lasted minutes, consumption was brief and payments were small. The design, results and limitations were checked on October 27. This was not an advertising-service procurement study and does not establish that directors are unnecessary or that every client ignores production origin.
The useful question is whether human involvement has a specified value. Brands also buy coordination, dependable delivery and work with a real product. Those responsibilities are not equivalent to a consumer paying to play a short game.
Buy a change in the work
Ask what a person will change: the product's role in the story, the order of a demonstration, a structural error, an actor's response or the hierarchy of information across formats. “Better taste” becomes assessable when it identifies a decision.
Consider a hypothetical coffee-appliance demonstration. An attractive initial sequence makes the contact point difficult to understand. A revised angle, edit or real-product insert could make the action clear. The improvement is something the client can examine. The number of times a director reviewed the sequence is an input, not sufficient evidence of the improvement.
For a concept-led film, the difference may be a stronger event or a clearer story rather than a technical correction. Agree on the central task before reviewing the sample, so that a new attractive detail does not quietly replace the question the work was meant to answer. This example is not an ONCE client case or a claim about a particular appliance.
Buy responsibility through an agreement
A sample cannot demonstrate future punctuality or how a supplier will handle an incorrect export. Review those commitments separately: who receives the product information, who confirms claims, who consolidates feedback, when decisions become fixed and who corrects defined problems.
Responsibility should match the work a supplier controls. It does not mean guaranteeing sales or taking over every brand obligation. Product statements need the appropriate brand owner, and specialist matters need appropriate review.
“Human QA” is particularly worth unpacking. Ask which objects are checked, what record is delivered and which corrections are included. A final playback review should not be mistaken for full product verification, permission review and multilingual checking. If the project needs those additional tasks, budget and assign them explicitly.
Buy production origin when it matters to this brief
A brand may need footage of an actual manufacturing process, the work of a named artist or restrictions on generated imagery. Specify where and why. The origin requirement for a documentary process shot may differ from that for a stylized background.
Do not assume that handmade content is always preferable. If the artist and process are part of the idea, their identity can be a legitimate purchase. If the task is simply to explain a product, adding a behind-the-scenes sequence does not automatically create value. The experiment neither rules out origin preferences in other markets nor proves one for this project.
Put three value columns beside the fee
| Reason to pay | Supplier explanation | How to examine it |
|---|---|---|
| Content improvement | The decision or passage changed | A comparable sample, revision and explanation |
| Delivery responsibility | The work owned and the correction commitment | Scope, named owner and agreed feedback process |
| Origin requirement | What must be captured or created in a specified way | Source information and agreed records |
One fee can serve several columns. Clarify overlap: an included product correction should not become a second charge under a different QA label unless the additional work is distinct. Align scope before comparing totals.
Use a paid, limited sample to examine the increment
For a new supplier, agree on a bounded paid passage with the same product inputs, communication task and output requirements. If comparing a baseline with an additional human-service option, record other changes in tools, inputs or conditions. They limit what can be attributed to the intervention.
An internal review can initially omit unnecessary price and title cues, record whether the content answers the task, and then examine production information and delivery commitments. This is a review sequence, not a reason to hide required disclosures or contractual information, and not a rigorous causal experiment.
If the core task shows no discernible improvement, ask whether the extra fee buys a needed delivery commitment or origin requirement. If neither applies, reduce that scope. If the improvement is useful, document the decision that produced it rather than prescribing the same additional hours for every future video.
For a production business, this turns human contribution into an intelligible service while leaving room to automate repetitive work. It is ONCE's proposed service-design direction, not a reported margin or customer outcome. Ask the next quote to explain three things: what changes, what is owned and what origin is required.
For related commissioning guidance, see ONCE's Video Production articles. The framework above evaluates the value of participation, rather than assuming it from a job title.
