Direct Actors and Non-Actors in the Same Commercial Without Forcing One Performance Style

A mixed-cast commercial usually goes wrong at the moment someone says "just be natural." It sounds like the kind thing to say, a way of putting everyone at ease. It's actually a demand that two people with different jobs produce the same output — and only one of them has a technique for it. The actor's naturalism is trained, adjustable, repeatable across takes. The person appearing as herself has something else, more valuable and less controllable: her actual way of answering a question about her own work.

So the fix is not to find a middle style both can meet. Stop sorting the scene by person and start sorting it by element. Some parts of a scene have to be the same every take: the handoff, the question, the sentence about what happens next. Other parts only work if they aren't the same every take: the length of an answer, the pace, whether someone looks up before she speaks. Assign each requirement to an element that can carry it, and the whole question of "one performance style" evaporates. Nobody imitates anybody, because nobody has the same job.

That is the answer. The rest of this article is what it costs to hold.

Sort the people first, then sort the scene

Before the shoot, three facts need to be on paper.

Who is playing an invented role. Who is appearing as themselves. And which single piece of information the scene has to land.

Take a constructed example, invented for this piece — no craftsperson, no participation agreement, no testimonial, no footage, nothing obtained. A neighborhood shoe-repair shop. An actor plays a new customer who walks in with a pair of boots and a worn heel. The craftsperson appears as herself, behind the counter. The commercial needs two things made visible: the boots passing from the customer's hands to the counter, and a clear statement of what happens next — the shop looks the boots over, calls the customer with a price, and no work begins until she approves it.

Now split the requirements by kind rather than by person:

A repeatable cue: an action or a line that must be the same every time. The handoff. The customer's opening question.

An agreed factual contribution: information the person actually holds and has agreed to state. How this shop assesses a repair.

An unknown response: whatever the craftsperson says back. Its shape, length and warmth are not decided in advance because they can't be.

Most production trouble comes from sliding one of these into the wrong column. The unknown response gets written down until it becomes a repeatable cue delivered by someone with no technique for repeating it. The agreed information gets withheld from the participant in the name of freshness, so she has to guess what the scene needs. And the repeatable cue gets handed to the person least able to hold it steady.

Two warnings belong here, because they're easy to get backwards. First, an actor is a real person. Her work isn't less honest for being trained, and giving her a playable task is not a lesser form of direction than leaving her to figure it out. Second, appearing as oneself is not a truth guarantee. If a line was written for the craftsperson and she delivers it, it's still a written line — putting it in a real person's mouth doesn't convert it into her personal view. If the commercial is going to imply that a statement is hers, she has to be the one who decides whether it is.

The repeatable version, honestly labeled

The structured route scripts the encounter as a sequence of clear actions: the customer enters with the boots, asks the agreed question, the craftsperson gives the approved explanation, the customer asks a confirming question, the boots go onto the counter. Every beat is a cue. Any take can be reshot, reviewed and matched to the last.

What that buys is unusually concrete. Coverage. Interchangeable takes. A guaranteed arrival at the necessary next-step information, no matter what anyone's day has been like. If the commercial must include a precise claim — the process for estimates, when the call comes, what approval means — this route can state it exactly, once, twice, six times, until the sound is clean.

What it costs is the reason you mixed the cast. If every word is fixed, the craftsperson is reciting, and recitation is a skill she wasn't hired for. On the fifth pass her delivery flattens, and the flattening is not a failure of character; it's what happens to anyone asked to reproduce a line they didn't write. Worse, the scene stops doing anything an all-actor cast couldn't do — and honestly, an all-actor cast could do it better. A fully scripted non-actor is a casting decision dressed up as a performance approach.

There's a specific temptation to name here. When the scene needs a particular feeling — conviction, warmth, a flicker of pride — it's easy to write it as a line and tell the person to mean it. Don't. Either the creative can survive without that feeling, or the team asks the participant whether she has something like it and takes what's actually there. Quietly scripting someone's personal feeling and presenting it as spontaneous is the one move that makes the whole mix dishonest, because it borrows the credibility of a real person to sell a written moment.

If a precise statement is non-negotiable, say so out loud. Label it approved information. That changes what the craftsperson is being asked to do — she's being accurate, not natural — and accuracy is a much fairer ask of someone who never agreed to act.

The responsive version with a fixed purpose

The other route keeps the frame and opens the middle. The counter, the boots, the opening question and the handoff stay fixed. The account stays hers: she explains the assessment in her own words, in her own order, and she may correct the customer's assumption. In this invented scene, the customer assumes the whole heel gets replaced; the craftsperson might say it may only need a rebuild once she looks at it. That correction is worth more than a scripted version of it, because it establishes that the shop looks before it prices.

What this route asks of the people is asymmetric, and pretending otherwise breaks it.

The actor's job gets harder, not easier. She isn't recreating a scripted reaction; she's holding a situation. What she needs from the director is the specificity of the circumstance — who this customer is, what she doesn't know, what she wants — and a playable responsibility. Listen. Make room. Ask the agreed question and then stop talking. Carry the transition to the handoff. All of those are actions she can perform. "Get her to open up" is not an action; it's an outcome, and an actor cannot play an outcome. Give her something to do and the interaction stays alive without her driving it. Her discipline, in this version, is largely a discipline of not filling silence.

The craftsperson's job is different in kind. She's being asked for recall, not invention, and recall has a failure mode: it degrades under repetition and under instruction. The first telling of how a repair gets assessed is usually the best one. The fourth is a summary of the third. If the schedule assumes eight takes of a personal account, the schedule is working against the thing it's paying for.

The edit inherits the consequences. In the responsive version, an exchange that happened once may have to carry the scene. You can't ask for a cleaner version of a genuine reaction at take three without it becoming a performed reaction — so if a piece of exact wording is required for the commercial to be truthful, don't try to extract it from the open layer. Move it to a layer that can be repeated.

Which is the meeting point, and it's worth showing as a shape.

Read down the right-hand column and the shape of the problem is visible. The responsive route buys you a real account and pays for it with edit flexibility. So the design question becomes: which piece of exact information absolutely cannot bend?

Where the two meet

For most mixed-cast scenes, one element has to be exact, and it usually isn't the participant's speech. It's the next-step information. The clean solution is to put that line in the actor's mouth, as a question.

The customer, still holding the boots, having heard whatever she heard: "And you call me before you start?"

The craftsperson's answer is now a small confirmation — a yes, maybe a nod, maybe "before we touch them." It's genuine, it's short, and it can be reshot without asking her to perform. Then the boots go onto the counter. The exact wording lives in the repeatable layer, where an actor can hold it steady, and the open layer stays open. If the process itself must be legible as text — on a sign, a ticket, a written line — that's another repeatable element, shot separately, which is a production decision for the team to make with the shop's actual materials rather than a line to pin on the craftsperson.

None of this is a guarantee. It's a proposal, and it holds only if the scene's dependency is honest about itself. Which is the last thing to check.

Test it against an ordinary answer

Three tests, run before the shoot rather than in the edit.

A brief answer. The craftsperson says "we look at them, we call you" and turns back to the bench. Does the commercial still work? If it doesn't — if four words are a disaster — then the scene was never really about the assessment. It was about a performance you didn't direct and can't repeat.

A correction. The customer's assumption gets corrected on the day, and the correction runs forty seconds longer than anyone planned. Does the commercial still work? Corrections are usually the most useful footage in a mixed-cast scene, but only if they were allowed for.

No emotional moment at all. No warmth, no story, no lift. The boots go onto the counter and the next step is clear. Does the commercial still work? If the answer is no, the creative has a dependency it hasn't declared, and the mix is being asked to produce a feeling rather than a fact.

One more boundary, stated where it belongs rather than buried: participation-sensitive questions — what a person is told, what she can refuse, how footage of her can be used — aren't performance decisions. They belong to the responsible team, and this article offers no consent, welfare or release conclusion. A proposal that rests on the participant not knowing what the scene is about isn't a performance structure at all; it's a different kind of question, and not one the shooting script gets to answer.

Throughout, the underlying limitation is worth keeping in view: the shop scene above is invented to make the structure legible. What would test it properly is a director's first-person account of a real mixed-performance spot — what was planned, and what changed on the day. That material isn't here.

What is here is the division. The handoff is repeatable; the boots pass from the customer's hands to the counter, take after take if needed. The question is repeatable; the actor can ask it cleanly eight times and adjust how she gets there. The answer is hers, and it should stay whatever length it turns out to be.

That's the version that holds. Not because the craftsperson delivered something moving, but because the scene was built so it didn't need her to.

Frequently asked questions

Why does 'just be natural' fail with a mixed cast?

It asks two people with different jobs to produce the same output, and only one has a trained technique for it. The actor's naturalism is adjustable and repeatable; the person appearing as herself has her actual way of answering. Instead, sort the scene by element: repeatable cues, an agreed factual contribution, and an unknown response that cannot be predetermined.

What are the three elements a mixed-cast scene should separate?

A repeatable cue is an action or line that must be the same every take, such as the handoff or the opening question. An agreed factual contribution is information the person actually holds and has agreed to state, such as how the shop assesses a repair. An unknown response is whatever the person says back, whose shape, length and warmth are not decided in advance.

How can an exact next-step claim be included without scripting the non-actor?

Put the exact wording in the actor's mouth as a question: the customer, still holding the boots, asks, 'And you call me before you start?' The craftsperson's answer can then be a short confirmation, a yes, a nod or 'before we touch them.' It stays genuine and reshootable, while the exact wording lives in the repeatable layer.

What three tests check whether the scene has an undeclared dependency?

First, if the craftsperson gives a brief answer like 'we look at them, we call you,' does the commercial still work? Second, if the customer's assumption is corrected on the day and the correction runs long, does it still work? Third, if there is no emotional moment at all, no warmth, story or lift, but the next step is clear, does it still work? If not, the creative is depending on a feeling rather than a fact.

What limits and boundaries does the article state?

The shop scene is invented to make the structure legible; what would test it properly is a director's first-person account of a real mixed-performance spot, which is not present. Participation-sensitive questions about what a person is told, what she can refuse and how footage can be used belong to the responsible team. The article offers no consent, welfare or release conclusion.

Further reading

• Direct the Hands in a Product Commercial, Not Just the Product — Once the handoff belongs to the repeatable layer, this gives four beats for writing it and notes how the other person's movement changes the offer mid-pass.
• Show How the Crew’s Presence Shapes an Observational Series — When the scene borrows real work as material, this maps what would happen anyway against what the crew arranges, and keeps participation questions unresolved.