What Does a Commercial Visual Researcher Do—and How Do You Brief One?

A commercial visual researcher investigates the questions a treatment has not yet answered, then returns material with its sources and reasons still attached. The job is not to fill a folder with beautiful images. It is to reduce the number of things the director is guessing about — and sometimes to say, with evidence, that a question cannot be answered from existing material at all.

That definition decides how the brief should read. You are assigning an investigation, not ordering a mood. The two failures are symmetrical: a brief made of adjectives produces a folder of images; a brief that names the exact pictures you already want produces confirmation instead of research.

Job titles and working arrangements vary from company to company. Some researchers sit inside a production company, some with an agency, some work independently for a single treatment. What follows is the shape of the assignment rather than an org chart.

1. Name what the research could change

Start with the film you are actually making, and then with the visual understanding you don't have. Those are two different sentences, and both need writing down.

"Warm, kinetic, surprising" is not an assignment. Each of those words is a verdict. The researcher cannot test a verdict, only illustrate it, and illustration is the thing that arrives as a folder. What the researcher can test is a question about behavior, material or camera.

Take a decision you're facing and split it:

• A camera question. Can this be done in one move, or does the pass need a cut?
• A material question. What does this surface do under the light we've chosen?
• A performance question. What does untimed human movement actually look like at full speed?

Each of those can be answered, partially answered, or found to be unanswerable. "Cinematic" cannot be any of the three.

Here is an invented example, stipulated rather than observed: a fictional spot for a cookware brand, in which an enamel colander — deep blue outside, cream inside, 24 cm — becomes the center of a choreographed group movement. Six performers in one ordinary kitchen pass it hand to hand along a line; the camera tracks with the colander; more people enter frame as it travels; it ends on the table, holding washed greens. The constraints are fixed before research starts: real-time capture with no slow motion in the final edit, one continuous take per setup, one physical colander used for the whole shoot, and the object identifiable in every frame.

The director's unresolved questions, then, might be these. At a normal kitchen field of view and full speed, does the colander stay readable as it crosses a pass — does the blue hold against the background, does the silhouette survive motion? How does ordinary window light sit on enamel, on the sheen and through the holes? How do people who are not dancers make a hand-off look anticipated rather than rehearsed?

Notice what has already been decided too: real kitchen, single take, real time, one physical object, a specified ending. Those decisions are the map. A researcher who never receives them will spend a week on studio-built sets and slow-motion choreography that you have just ruled out.

One more thing to say out loud in the brief: research is not the collection stage that begins after the creative thinking is finished. It can come back and tell you that your question was framed badly — that you asked for references of people passing objects when the real variable is what the second person does before the object arrives. Build in the expectation that a finding may change the question. Otherwise you will only ever be sent the answer you already described.

2. Turn the question into selection criteria

The useful sentence in a brief describes a behavior or a relationship a reference should make visible. The unhelpful sentence describes a feeling.

"Warm" becomes: the moment before a hand closes on the object, two or three people in the same frame are already moving toward where it will be. "Kinetic" becomes: the object changes hands twice inside four seconds without the camera losing track of it. Both of those can be checked against a clip. Neither requires the researcher to guess what you meant.

Alongside the behavior, state the constraints that come from the actual product and the actual brief. The colander is deep blue and must remain distinguishable from the background for the whole move. The take is continuous. The final edit is real-time. The performer group is six people, none of them trained dancers. Constraints do two jobs at once: they narrow the search, and they give the researcher grounds to tell you when your question is impossible.

Then describe what would make a result misleading. This part is usually missing, and it's the part that saves the most time.

• A clip where the pass is hidden by a cut answers nothing about whether a pass reads at speed.
• A clip of the same object in a different colorway or material — steel rather than enamel — answers the timing question but not the surface question.
• A clip of trained performers implies a rehearsal capacity the production may not have, and quietly sets a standard against which real footage will look broken.
• A clip shot hand-held and reactive implies a film you have not chosen. Adopt it as the model and the treatment changes without anyone deciding to change it.

Framing an exclusion as a behavior is stronger than banning a style. "The object must remain readable during the movement" rules out more useless material than "no unfamiliar visual styles," and it doesn't prevent the researcher from bringing you something you didn't imagine. A ban on a look is an aesthetic judgment delivered before the evidence arrives.

It also helps to say which partial answers would still be worth having. A reference that shows the timing but uses the wrong object still answers the performance question. A reference with the right object but a cut on the pass still answers the material question. Naming which partial answers help keeps a thin search from being wasted — and keeps the researcher from treating every unmet criterion as a reason to send nothing.

What the brief should not do is pre-select the answers. If you have already found the three shots you want, you are asking for validation, not research. Give the question and the constraints; leave the route open. You are paying for the part you couldn't find.

3. Access, boundaries, and the shape of the return

Research and clearance are separate tasks. Locating a file, describing what it shows and judging whether it answers your question is research. Whether that file can be reproduced, licensed or used as final footage is a different question, usually answered by someone with the standing to answer it. Keep them apart in the brief or you will get an assumption doing the work of a permission.

Share only material your collaborator is authorized to use in the way you're agreeing. That includes the tempting things: the unreleased product sample, the client's packaging design, the director's personal footage, an animatic, an actor's tape. If something is restricted, say so plainly, and say what happens to it afterward — whether the researcher keeps it, returns it, or deletes it at the close of the assignment.

Scope the assignment honestly. If it is research only, then locating and describing is the deliverable, and original illustration or design is a different commission with a different author and a different review. Don't fold it in silently because a found image suggested a drawing.

In this invented return, the researcher sends back three things.

Item A — the timing reference. Twelve seconds, side-on, unedited, real-time. Six people pass a heavy object along a line, and each person's free hand is already moving before the object arrives. Nothing in the frame is styled. The rationale is short: this answers the performance question directly, because the anticipation is visible at full speed and the pass is not concealed by a cut. The flagged uncertainty is equally short: the recording is low resolution, so nothing about surface or color should be judged from it.

Item B — the attractive one. A clip from a dance film. One take, an object crossing frame, genuinely lovely. It is shot at a high frame rate and played back slowed, on a wide lens, with atmosphere in the air, performed by trained dancers. Its rationale: it answers "what could this feel like at its best" and does not answer "can six untrained performers do this in real time." It also conflicts with two stated constraints, which is exactly why it is here — the researcher is showing you the objection rather than hiding it in an untitled folder.

Item C — the gap. The researcher could not find material showing a deep-blue enamel colander passed among more than two people at full speed with the object staying legible. What exists either cuts on the pass, uses a computer-generated object, or uses steel, whose moving highlights behave differently against a kitchen. They name what was searched, what was excluded and where the search stopped.

Three things should accompany every meaningful selection.

A rationale of a sentence or two: what this item answers about our question, and what it doesn't.

The original source context, not just a frame. Where it came from, who made it, when, whether the situation was staged, and what surrounds it in the original. A still pulled from a ninety-second sequence can say the opposite of the sequence, and a crop can hide the very staging that produced the effect.

Visible uncertainty, stated as such. An unconfirmed frame rate. Color in a scan that can't be trusted. A moment that might be a practical effect or might be post-production. Uncertainty isn't a defect in the return; it's information the team needs before it leans on the finding.

Failed routes and unanswered questions belong in the return when they change a decision. "I looked for X and found nothing adequate, and here is why the available material fails" is a result, not a shortfall. What the return should not do is imply that collecting a file clears its reproduction or turns it into final footage. A link is not a license, and a downloaded clip is not an asset.

You will notice there is no number in any of this. Ask for coverage of the questions rather than a quantity of images. A quota turns the folder into the deliverable, and then the pressure to reach the number produces items that answer nothing — the researcher's version of writing to length. If a question has no adequate material, that is the finding.

4. Read the return against the original question

When the material arrives, ask of each selection: what can we judge now that we couldn't judge before?

Item A lets the team judge whether a full-speed pass reads at all, roughly where a cut could fall if the take has to break, and what a minimum rehearsal would need to produce. That's decidable this week.

Item B lets the team see that the quality they responded to — the float, the suspension — comes from slow motion and trained performers, both of which the brief excluded. So the reference is either retired, or it becomes an argument for changing a constraint, which is the director's decision and no longer a research finding. Either outcome is useful. What would be useless is treating B as the target and discovering the mismatch on set.

Item C moves the decision off the search entirely: budget a test, or accept a design change such as framing the colander larger through the pass, or lighting the kitchen so the deep blue separates from the background. The absence of a reference has produced a concrete choice.

Throughout, keep the director's reading separate from what the source actually shows. The observerable fact about Item B is that an object crosses the frame in under a second, recorded at a high frame rate and played back slowly. "It feels euphoric" is an interpretation — a good one, worth recording, and worth marking as an interpretation. The same applies to Item A: the timing pattern is observable; "the group feels like a family" is a reading. When the two blur, the team ends up defending a feeling as though it were a measurement.

And when a return is thin, invite the researcher to explain the mismatch rather than fill the folder to a number. There are three honest responses. The finding supports a treatment choice, and you lock it. The finding narrows the next search, and you say precisely what is still missing and why. Or the finding shows you need separately scoped development — a test shoot, a previz, an original illustration, each with its own authorship and its own review.

What none of those responses requires is unlimited revision. If the brief promises that research continues until someone is satisfied, the assignment has no stopping condition, and the researcher absorbs the cost of your uncertainty. If a question keeps coming back thin, the honest moves are to change the question, budget the development, or make the decision without a reference and say so.

The brief, in five lines

The decision. We are trying to judge whether the colander stays readable through a full-speed pass in one continuous take.

Already decided. Real kitchen, real time, one take per setup, one physical object, the ending on the table.

A useful reference shows. The hand moving before the object arrives, and the object changing hands without the camera losing it.

A misleading reference shows. A cut on the pass, a different material, trained dancers, or slow motion standing in for timing.

Send back. Each selection with a two-line rationale, its original context, and its uncertainties — including the routes that failed.

A brief that names a decision, a body of usable evidence and a return format gives the researcher something to argue with, which is most of the value. And keep one thing in view when the folder lands: the colander nobody could find is worth more to the film than a hundred beautiful images of a different object.

Frequently asked questions

What is a commercial visual researcher actually assigned to do?

Investigate the questions a treatment has not yet answered and return material with its sources and reasons attached. The aim is to reduce what the director is guessing about, and sometimes to report that existing material cannot answer a question.

How should a brief avoid producing only a folder of images?

Name what the research could change and turn the question into selection criteria about behavior or relationships. Include fixed constraints and describe what would make a result misleading. Do not pre-select the exact pictures you already want, since that asks for validation rather than research.

Why keep research and clearance separate in the brief?

Locating a file, describing what it shows, and judging whether it answers the question is research. Whether it can be reproduced, licensed, or used as final footage is clearance, usually answered by someone with standing. Keeping them apart prevents an assumption from doing the work of permission.

What should accompany each meaningful selection in a return?

A short rationale saying what the item answers and what it does not; original source context such as where it came from, who made it, when, and whether it was staged; and visible uncertainty. Failed routes and unanswered questions belong in the return when they change a decision.

How should a thin or incomplete return be handled?

Ask the researcher to explain the mismatch rather than filling a quota. The honest responses are that the finding supports a treatment choice, that it narrows the next search with specific missing pieces named, or that it shows a need for separately scoped development such as a test shoot or previz. Research should not continue until someone is satisfied.

Further reading

• Do You Need a Treatment Writer, Designer, Visual Researcher—or a Clearer Idea? — Before commissioning research, read this to test whether the treatment's real gap is visual evidence or an idea that has not yet been chosen.
• Does Your Film Need an Archivist, a Historical Adviser, or More Reporting? — When the assignment mixes locating material, checking provenance and judging portrayal, read this to split those tasks and define the shape of the return before you hire.