Who Is Speaking in Your Commercial's Voiceover—and to Whom?
Who Is Speaking in Your Commercial’s Voiceover—and to Whom?
A treatment line reads: "Voiceover: warm, confident, intimate—like talking to a friend."
That is a reasonable thing to want and nearly impossible to play. An actor receiving it has no character, no listener, no stake and no moment. They will supply all four themselves, and what they supply may have nothing to do with the film you are proposing. The adjectives are not wrong. They are the last layer, described before the first two exist.
Before you describe the voice, decide three things: who is speaking, who is being addressed, and when the words happen relative to the picture. A vocal quality cannot repair an undefined point of view. It can only disguise it well enough to get through the meeting.
Give the voice a position in the scene
Start with where the speaker stands.
A participant is inside the situation and knows what they lived. This is the most common and most mishandled position, because it invites you to write lines that require knowledge the speaker does not have. If your participant says something only a witness could have seen, you have not written a character; you have written a narrator in a character's clothes. The fix is usually structural, not verbal: either someone tells them, or they find out, or you move the line to someone else.
A witness is present but not central—the neighbor on the step, the barista who saw the whole thing. Witnesses get a specific gift, which is the right to interpret without owning. They can describe someone else's behavior and be wrong about it, and the film survives.
A narrator outside the action belongs to no one in the picture. That is a legitimate and expensive choice. The voice can explain anything, at any time, from any distance—and precisely because nothing constrains it, it drifts toward the sound of a company describing itself. If you choose this position, the film has to give the voice a reason to exist that is not "we need the information somewhere."
There is a fourth position worth naming because treatments stumble into it: an imagined voice. A conscience, a running inner monologue, a product that talks, a person narrating their own memory from inside it. This works when you decide whose head the voice is in. "An inner voice" without an owner is not a position; it is a mood.
Then pin down when. The same line has three different lives as a recollection, a live utterance and an anticipation. Past tense means the event is over and the speaker is now looking back across it. Present tense means they are in it. Future tense—or a line spoken at the door before the handle turns—means the event has not happened yet.
Point of view and tense are not stylistic garnish. They determine what the viewer knows, in what order, and therefore what the shot means.
Name the person being addressed
A voice with no addressee has no intention, and intention is what an actor actually plays.
Ask who is on the other end. A friend. A stranger. Themselves, quietly. The audience, openly. Each choice changes the sentences that are available.
"Conversational" is a container word, and it contains at least two incompatible films. One is a person confiding in someone who is already on their side; their sentences can start mid-thought, refer to shared history, and leave conclusions unsaid. The other is a person persuading someone who is not yet convinced; their sentences must define terms, pre-empt objections, and land their points in order. Both are conversational. They are not the same performance, and an actor given the wrong one will deliver a film you did not propose.
This is also where vocal adjectives stop being mysterious and start being consequences. Warmth is what you get when a speaker is on the listener's side. Confidence is what you get when the speaker is not asking permission. Intimacy is what you get when the speaker assumes the listener already knows something. If you want those qualities, write the relationship that produces them; then the adjectives describe the result instead of demanding it.
The listener does not need to appear on screen. But the treatment should make their existence legible—because a line written for a friend is audibly different from a line written for a stranger, and the difference is in the words you are about to hand over.
Decide when words and pictures agree
Now take one phrase and hold it next to the image that will accompany it.
The words can confirm the picture: what we see is what we hear, and the redundancy is doing the work of certainty. They can anticipate it: the voice says what we are about to watch, which creates a small forward lean. They can complicate it: the voice claims one thing while the image suggests another, and the viewer reconciles. Or they can open a gap: the voice speaks about something the image never shows, and the film asks the viewer to hold both.
Worth saying plainly: contradiction is not automatically wit. Irony between sound and image is a machine with parts—setup, tension, release—and if the film builds none of those parts, the audience does not experience a clever mismatch. They experience a mistake, or worse, nothing. Tension has to be planted before it can be harvested.
And keep two audiences separate in your own head. Your treatment is read by people deciding whether to make the film. Your viewer receives only the finished combination of voice and picture, with no explanation attached. If the words take a meaning you need to defend in a sentence of prose, that sentence is not riding along with the video.
Describe the performance without borrowing authority
When you write the direction, work from knowledge, intention and observable behavior.
Knowledge: what does this speaker know at this moment, and what do they not know? Intention: what are they trying to get from the person they are addressing—reassurance, permission, forgiveness, agreement, a laugh? Behavior: do they say it once or try it twice? Do they look at the listener or away from them? Is the sentence the first thing they say or the thing they finally get to?
That will get you further than any adjective list, and it will not smuggle in a reference the production cannot legally or practically deliver. Two shortcuts to avoid: describing the voice by naming a famous actor you would like the audience to think of, and describing it by demographic shorthand. Both substitute a borrowed impression for a decision, and the second usually imports a set of assumptions the film did not intend to make. If you cannot say what this person wants from this sentence, a name will not say it for you.
One boundary deserves its own paragraph, because treatments cross it constantly.
A scripted first-person voice is fiction, even when it sounds like a diary. If you write "I used to dread the drive home," and an actor performs it, no customer said that. Presenting the result as a real person's experience—or implying it in the treatment language, which is where the implication usually starts—is a claim you have not earned. If the speaker is a real customer, the words need to be theirs, or a properly attributed account of theirs. Otherwise, call the role what it is: a written character, performed. A testimonial is evidence. A performance is a performance. They can both be good; they are not interchangeable.
Finally, know where the narrator section ends. Once the position and address are set, phrasing, breath, emphasis and duration belong to the spoken-copy work. The treatment's job was to make the role directable, not to pre-perform it.
One arrival scene, two speakers
Here is an invented illustration, and it stays invented: no recording exists, nobody has been cast, and it is not drawn from any commercial I can point you to.
The picture is fixed. A guest arrives late to a gathering. Outside, in the light from a window, they straighten their collar, say something under their breath, and reach for the handle. Same shot in both propositions.
Proposition A puts the line in the guest's mouth, which makes it an on-camera line rather than a voiceover. The same analysis applies either way—position, addressee and tense are decided identically whether the words come from inside the frame or over it.
Proposition A. The guest's own voice, in the guest's mouth, addressing the friend they are about to disappoint: "I have a very good reason."
The speaker knows their own reason, knows the friend is annoyed, and can see the room only in outline from the step. The intention is to get ahead of the disappointment—to be forgiven before being accused. Two production consequences follow immediately.
First, tense. "Have" places the speaking inside the moment, which is what the picture shows: the guest at the door, still hoping. Written as "had," the line would place the speaking after the event, and the two would be compatible only if the voice were retrospective—the guest looking back from inside the evening, or from later—with the shot on the step becoming a remembered image rather than a present one. For a brief about owning a small mistake, that difference matters, because the retrospective speaker has already been through the door and the present-tense one has not.
Second, the picture complicates what we are hearing. We watch the sentence get tried out. A rehearsal does not prove the reason is false—the image establishes preparation, not guilt, and it would be a mistake to treat the two as the same thing. But it does change the category of the statement. A rehearsed defense is not a confession. The viewer notices that before they notice anything the voice says.
Proposition B. The waiting friend, in the next room, addressing another guest: "They always have a very good reason."
The knowledge is smaller and more human. This speaker knows the pattern, not tonight's reason. She has not seen the rehearsal—the audience has, and that asymmetry is now part of the scene: the viewer holds something the speaker does not.
The addressee is a third guest, which quietly makes the line about the speaker's own standing in the room as much as about the latecomer. The intention is to pre-empt the excuse, dryly, for someone who needs the context. The attitude is affection with a small blade in it. And "always" is only available to a speaker with history across many evenings. A stranger cannot say this line. Neither can a narrator outside the action, who would have to earn the familiarity some other way.
There is also a presence problem to accept rather than ignore. This speaker is not in frame. The viewer hears a person who is not there, over an image of someone who is. Films do this all the time, but it is a permission the viewer must extend, and it works better when the voice belongs somewhere specific rather than nowhere.
Neither proposition is a testimonial and neither is a finished recording. Choose the first if the brief is about owning a small mistake. Choose the second only if the film is about the two friends' shared understanding—because that is what the line is actually about, and no amount of warmth in the delivery will relocate its subject.
The brief you can hand over
Cut every adjective that adds no actionable distinction, and end with four facts and one phrase against picture.
Speaker: [who, and their position in the scene] Listener: [who they are addressing, and what that person knows] Knowledge at this moment: [what the speaker knows and does not know] Intention: [what they want from the listener] Line against picture: "[one phrase]" over [the shot it lands on]
Run it on the arrival scene: A late guest, inside the moment, speaking to the friend they expect to disappoint; knows their own reason and the friend's irritation, not the room; wants to be forgiven before being accused. "I have a very good reason" over the hand reaching for the handle.
That is a performance someone can play, and a filmed line someone can evaluate. It also makes the tense a visible decision—present here, because the picture holds the moment at the door—rather than something buried in the wording.
If you can, read the line aloud against the shot description before you send it, and keep any product or legal language exactly as its owner wrote it; a scratch read that changes required wording has not tested anything. Nor is any of this a promise about casting, rights, or how long the line will take to say. Those come after a person and a listener exist—which, at the treatment stage, is the whole job.
Frequently asked questions
Why is a direction like “warm, confident, intimate—like talking to a friend” hard for an actor to play?
It gives no character, listener, stake, or moment; the actor must supply all four, possibly unrelated to the film. These adjectives are the last layer, described before the first two exist. First decide who is speaking, who is addressed, and when the words happen relative to the picture.
What speaker positions can a voiceover occupy?
A participant inside the situation who knows what they lived; a witness present but not central who can interpret without owning; a narrator outside the action, which can explain anything but tends to drift toward a company describing itself; and an imagined voice, which works only if you decide whose head it is in.
How does the listener change the writing?
A voice with no addressee has no intention, and intention is what an actor plays. A friend, stranger, themselves, or audience each changes available sentences. “Conversational” can mean confiding in someone already on your side or persuading someone not yet convinced; those require different sentences and performances.
What should a voiceover brief contain?
Four facts and one phrase against picture: speaker and position; listener and what they know; knowledge at this moment; intention—what they want from the listener; and the line against the shot it lands on. Read it aloud if possible, and keep product or legal language exactly as its owner wrote it.
Can a scripted first-person voiceover be presented as a real customer's experience?
A scripted first-person voice is fiction even when it sounds like a diary; if an actor performs it, no customer said that. If the speaker is a real customer, the words need to be theirs or a properly attributed account. A testimonial is evidence; a performance is a performance.