Turn Audience Research Into a Commercial Situation, Not a Stereotype

A research summary is not a cast list. It can tell you that a circumstance existed and that certain people described living inside it. It cannot tell you who your character is, what they say, or what they'll do once they're in the frame.

The translation runs in three moves, and most of the work is keeping the moves from collapsing into one another. Recover what the study actually recorded, including the parts the summary dropped. Turn the recorded tension into circumstances a person can be standing in. Then write a second version with a different person making a different decision, and read the difference. If the two versions only differ because the two people are different kinds of people, the first one was a category wearing a name.

The teaching scenario

Everything below — the repair service, the diary study, the two characters — is invented to demonstrate the method. No real study, participant or client is described. Substitute your own approved report; the moves don't change.

Suppose an agency is handed a fictional summary for a fictional appliance-repair company. It reads:

Customer insight. Two-week diary study, ten customers who had booked a repair in the past year, one metropolitan area. Seven of the ten described the four-hour arrival window. Three stayed in for the whole of it. Customers are time-poor and want speed.

That paragraph is a compression of a compression, and it has already cost you something. The first job is getting it back.

Recover the finding before you borrow its headline

Read the full report, not the slide. Note who took part, where, when, and how the material was produced. Ten customers, two weeks of diaries, one follow-up call each, one metropolitan area, all of them people who had already bought the service once. Every one of those conditions limits what the study can be used for, and every one of them is doing quiet work in your head whether or not you noticed.

Then sort the material into layers, because the layers answer different questions.

The participant's account. A diary entry is someone's own writing about their own week, composed after the fact. It is evidence of what a person chose to write down. It is not a researcher's neutral observation, and it is not a measurement. One diarist wrote that she "didn't dare start anything." Another described staying home all morning. These are accounts, and they're the raw material.

The research team's grouping. Somebody read those diaries and decided that the window, rather than the broken machine, was the anxious part of the week. That is a judgment, made by a person, from a small set of accounts. It's a reasonable judgment. It is not the same kind of statement as the diary line, and it should never be quoted as though it were.

The summary's compression. "Time-poor customers who want speed" is a third thing again — a headline written to travel. Test each word against the notes. Time-poor isn't in the diaries; what's in the diaries is a four-hour block in which nothing else can be started. Want speed points at a different fix from the one the accounts describe: knowing when someone will arrive is not the same as shortening the wait. Whether a shorter window would also help is a question these ten diarists never answered.

The UK government's service manual guidance on analysing a research session sets out the same order of operations — write down what was seen and heard, then group and sort those notes into findings (gov.uk, checked 8 September 2026). It's a UK user-research method rather than advertising-effects evidence, and it doesn't claim that observations arrive free of bias. What it offers here is the sequence: record first, interpret second, and don't let the second step erase the first.

Turn the tension into circumstances, not identity

A usable finding doesn't hand you a personality. It hands you a relationship: what someone is trying to do, what's in the way, and why the product is present at all.

From the invented study: the person is trying to get a machine working again. The obstruction is that nobody can say when anyone will come. The product is present because it holds the booking and the window. That relationship is a circumstance, and anybody can be in it. A retired man, a night-shift nurse and a new tenant all fit inside it without any of them becoming a type.

Now give your invented person an immediate objective and a particular environment. Fictional version one:

Nadia has had the repair booked for eleven days and has been washing things in the sink. The window is 8 a.m. to 12. Her class starts at 12:30 and is a forty-five-minute ride away, which puts her out of the door by 11:45 with nothing to spare. Her flat has an entry phone that she has to answer herself. She stays, and by 11:35 she is down to the last ten minutes she can still leave in. She doesn't leave, and the class she paid for starts without her.

Everything in that paragraph beyond the four-hour window is a writing choice. The eleven days, the class, the ride time, the entry phone, the flat — all invented. Say so in the treatment's rationale, in plain words, so nobody downstream reads them as findings.

Now the shortcut you'll be tempted by, because it's shorter and it sounds like insight: Nadia, 34, a busy professional who expects everything on demand. Delete it. It doesn't come from these ten diaries. It doesn't distinguish Nadia from several million other people. Worst of all, it explains the scene by category, which means the scene itself no longer has to explain anything — you've outsourced the work to a label and the drama evaporates.

There's a related trap in the absence of detail. The first version of this premise had Nadia on the top floor of a building where nobody else was home. That was a fabrication, and it was a fabrication with a purpose: it made staying the only possible action. One person with no options isn't a situation, it's a trap. If your premise needs your character to be unable to do anything else, you didn't find that restriction in the research — you installed it, and the scene is now rigged.

Write the second response

Same service, same four-hour window, same week, same entry phone. Different invented person, different action.

Theo asks the neighbour downstairs on Wednesday evening whether she'd mind listening for the door. She says yes. On Thursday he leaves at nine, comes back at one, and finds the machine fixed and his key pushed through the letterbox.

Two things follow from putting these side by side.

The first is a test of premise one. Theo's route was available to Nadia — she could have knocked on the same door. She didn't. That means the difference between the two premises is a decision made inside a shared circumstance, which is exactly what you want. If instead Theo's option had only been open to Theo because of who Theo is, premise one was resting on category rather than situation. Read the two side by side and ask which of those two readings is true. You'll usually know within a paragraph.

The second is that neither premise is the honest one. Theo's plan doesn't save a morning; it borrows the neighbour's. He's traded his own four hours of waiting for a favour he now owes, plus the small risk he took with his front door key. That's a real cost, and it belongs in the rationale rather than being tidied away. Premise A shows someone absorbing the cost alone; premise B shows it moved onto a relationship. Both are true of the circumstance, and the brief decides which one you're making. If the service's claim is a narrower window, premise A has somewhere to go. If the claim is that you can nominate someone to let the engineer in, premise B does. Pick on that basis, not on which scene feels truer.

While you're here, check for the other stereotype, the one nobody flags: helplessness. If every invented customer in this circumstance can only sit and wait, you haven't described a situation, you've described a mood. Waiting is one option among several, and the study's own accounts don't establish that it's the only one.

Keep the source-to-scene boundary readable

The rationale is where this either holds or falls apart. Annotate the chosen premise in three columns and keep every line in the right one.

The third row deserves a note of its own. Absence of a complaint in ten diaries is not proof that nobody has that complaint; it's proof that these ten people didn't raise it. Both of those statements are fine to write, and only one of them is true.

Then go through the treatment and strip out any research label that appears to endorse something untested. A citation next to a product claim, a casting choice, a line of dialogue, or a predicted response reads as validation whether you meant it that way or not. The research can support why this circumstance may be worth a scene. It cannot support this will work.

If you can, put the portrayal in front of someone with relevant lived experience of the circumstance and ask them to attack it. Then hold their answer at the right size. "That's not me" doesn't make a scene false, and "that's exactly me" doesn't make it representative — one reader is one reader, not certification. What you're listening for is structural. If the objection is that the character has no life beyond the appliance and the window, the fix is not a softer adjective. Go back to the premise and give her something else to want.

What you're handing over

A treatment built this way ends with a premise whose invented parts are visibly yours. Something like:

Inside a four-hour window, Nadia has already given up the morning she paid for, and she finds she can't start anything she'd have to finish. (Invented scene: the character, the class and the flat are writing choices, not findings. The four-hour window and the morning spent staying in come from the study's accounts. "Half-finished" is also a writing choice, made from the inference that waiting has a texture — it isn't a word the diaries use.)

That's a whole premise — a person, an obstruction, a decision — and the note underneath it tells everyone exactly where the research stopped and the writing began. It doesn't claim to represent the ten diarists, or the customers who never kept a diary, or anyone else. It claims to sit inside a circumstance the study described, which is the strongest thing a piece of audience research can actually authorise.

The window came from the study. Nadia, the class and the half-finished shelf did not. Keep those two sentences close together in your own head, and the scene will keep its nerve.

Frequently asked questions

What can audience research authorize for a treatment?

It can describe a circumstance that existed and what certain people said about living inside it. It cannot supply the character, dialogue, or action, and it cannot authorize predictions that a scene or product will work. A citation next to a product claim, casting choice, line of dialogue, or predicted response reads as validation whether intended or not. The research can support why a circumstance may be worth a scene, not that the execution will succeed.

How should a diary line, a research grouping, and a summary headline be treated differently?

A diary entry is a participant's own writing about their week, composed after the fact; it is evidence of what they chose to write, not neutral observation or measurement. A research grouping is a judgment made by the research team from a small set of accounts, and it should not be quoted as if it were a diary line. A summary compression such as 'time-poor customers who want speed' is a headline written to travel; test each word against the notes. In the invented example, 'time-poor' is not in the diaries, and 'want speed' points at a different fix from knowing when someone will arrive.

How do you turn a research tension into a situation rather than a stereotype?

Start from the relationship: what someone is trying to do, what is in the way, and why the product is present. In the invented repair example, the person is trying to get a machine working, the obstruction is that nobody can say when anyone will come, and the product holds the booking and the window. Give the invented person an immediate objective and a particular environment, then label everything beyond the source as a writing choice. Avoid category labels like 'busy professional who expects everything on demand,' and avoid fabricated restrictions that make only one action possible—one person with no options is a trap, not a situation.

Why write a second version with a different person making a different decision?

It tests whether the first premise rests on a shared circumstance or on category. In the invented comparison, Theo asks a neighbor to listen for the door and leaves; his route was available to Nadia, who stayed. The difference is a decision inside a shared circumstance. Neither premise is automatically the honest one: premise A shows someone absorbing the cost alone, while premise B moves the cost onto a relationship and adds a favor owed plus a risk with the front-door key. The brief decides on the basis of the service claim, not on which scene feels truer.

What belongs in the source-to-scene rationale?

Use three columns: source finding, creative inference, and proposed action, and keep every line in the right one. Absence of a complaint in ten diaries is not proof nobody has that complaint; it only shows these ten did not raise it. Strip out research labels that appear to endorse untested product claims, casting choices, dialogue, or predicted responses. If someone with relevant lived experience attacks the portrayal, hold their answer at the right size: 'that's not me' does not make the scene false, and 'that's exactly me' does not make it representative. Listen for structural objections—if the character has no life beyond the appliance and the window, go back to the premise and give her something else to want.

Further reading

• Describe the Performance You Want, Not Just the People You Want to Cast — Take your premise here when you need a performance line a director can actually shoot, with a stimulus and an obstacle instead of a mood to demonstrate.