What Are the Interviews Doing in Your Documentary?
What Are the Interviews Doing in Your Documentary?
An interview approach starts with what a person can contribute to the film, not how many impressive names fit on a contributor page. Ask whose knowledge you need, what the encounter might change, and whether the person speaking needs to be seen.
A witness can describe an event they experienced. A researcher can interpret a record. A participant can revisit a decision, disagree with another account, or help the filmmaker discover that the original question was wrong. Those are different functions. They call for different relationships between conversation, image, and the rest of the inquiry.
“Intimate interviews with key figures” settles none of them. Neither does “no talking heads.” A face on screen is not a failure of imagination, and a voice over beautiful archive is not automatically a better use of testimony. The choice needs a reason that belongs to this film.
Replace the contributor label with a knowledge question
Imagine a documentary about the last years of a neighborhood theater. A retired projectionist, Jonah Vale, has agreed within this fictional scenario to a preliminary research conversation. The filmmaker has a damaged maintenance log. Several pages are blank, and the next surviving entry records a replacement projector.
The people, theater, log, and possible encounters in this example are invented. No participant has been interviewed, and none of the reactions described below has occurred.
A contributor list might identify Jonah as “the theater's legendary projectionist.” That tells us how the pitch wants us to regard him. It does not tell us what he can establish.
A more useful starting question is: What might explain the gap in the maintenance log, and what did Jonah personally encounter during that period?
That question contains two tasks. Jonah may remember how staff recorded maintenance, who used the book, and whether there were other logs. Those recollections could help explain the object. He may also remember the replacement projector arriving. Neither possibility entitles the filmmaker to treat him as the author of every entry or a witness to every managerial decision.
Separate the claims before choosing the form. “I wrote that entry,” “someone told me why the machine was replaced,” and “looking at it now, I think this is what happened” have different evidentiary positions. The film can contain all three, but it should not smooth them into the same kind of knowledge.
For the pitch, replace the status label with a function:
Jonah can explain the projection booth's working routines and help us distinguish his own entries from material recorded by other staff. We will ask what he recalls about the missing period and what additional records may survive.
That description is less grand than “the man who knows every secret.” It gives the interview an actual job.
Decide what would change because this conversation exists
An interview may provide necessary orientation. That is a legitimate function. A viewer cannot understand why the damaged log matters without knowing what staff used it for.
But supplying background is not the only possibility. The encounter may change the investigation. Jonah could identify a second book, correct the filmmaker's reading of a technical abbreviation, or say he was not working there during the relevant period. Each is a possible result, not a promised revelation.
The difference matters when you write the treatment. An informational passage can explain a practice already established through research. An investigative encounter invites an answer that may redirect the film. A scene about remembering can remain important even if it does not settle the historical question.
Do not force these into one promise. “Jonah will finally reveal what happened” makes discovery an obligation placed on a participant. “We will ask Jonah to examine the surviving log and identify what he can and cannot account for” defines a feasible encounter while leaving its result open.
The question also tests whether the interview belongs in the finished film. Suppose the preliminary conversation establishes only that another staff member kept the relevant records. Jonah has done something valuable: he has sent the filmmaker toward a better source. That does not create a duty to include his voice in the cut.
Conversely, a conversation may be central even when its factual content could be summarized in a sentence. If the film's concern is how a person now relates to their earlier work, the encounter itself—not just its informational yield—may be the material.
Three approaches to the same damaged log
Keep the research question constant and change the proposed relationship between the person, the object, and the audience. These are alternatives to consider, not a plan to collect three versions of the same scene regardless of need.
Let the examination become an on-camera encounter
In the first approach, the filmmaker invites Jonah to examine the surviving log on camera, subject to his agreement and the object's handling conditions. The film can show which page he is discussing, what prompts a correction, and how the filmmaker asks for clarification.
The point is not to wait for a photogenic emotional response. It is to make the process of working through the object part of what the viewer can follow.
A proposed passage might read:
At a table with the surviving maintenance log, Jonah will be invited to identify entries he recognizes and explain how the book was used. Our questions will remain in the exchange where they are necessary to understand his answers. The encounter may clarify the gap or establish the limits of what he can tell us.
This route suits a film in which the present act of looking matters. It makes the filmmaker's prompts available rather than presenting the resulting account as an unprompted speech.
It does not make a recollection independently verified. A pause cannot tell you whether an answer is true, and a close-up cannot explain why somebody hesitates. If Jonah cannot identify an entry, the responsible result may be a scene that preserves that uncertainty—not an edit that borrows a confident answer from a different question.
The log also has limits. A blank page does not announce its own cause. Showing it beside Jonah's face does not establish that he knows why it is blank.
Separate the retrospective voice from the historical material
In the second approach, the interview supplies a present-day recollection heard with clearly identified historical records. Jonah's face need not be the image carrying the passage.
This can serve a film organized around surviving traces of the theater's working life. His account can identify a routine, dispute an interpretation, or make a distinction the document alone cannot convey. The image need not merely illustrate each noun he says.
For example, an explicitly hypothetical account might distinguish routine notes from urgent repairs. The film could put that explanation in relation to dated entries rather than simply cutting to a picture of a projector. The useful relationship would be between the scope of the record and the scope of the memory.
The danger is making the two appear to corroborate each other when they do not. A photograph from a different year cannot silently become a record of the incident being described. A voice recalling “that winter” should not acquire an exact date merely because the editor places it over a dated page.
A pitch passage can make the intended separation clear:
Retrospective interviews will be heard in relation to dated theater records. Contributors' memories will remain attributed, and the film will distinguish material from the relevant period from later or illustrative material. Where an account and a record differ, that difference will remain a question rather than disappear into the montage.
This approach gains the ability to build a passage around the records. It gives up some of the visible circumstances of speaking. Whether that is a useful exchange depends on the inquiry; removing the face is not, by itself, an aesthetic achievement.
Keep the interview in research and follow its consequences
In the third approach, Jonah helps the filmmaker understand the log, but his interview is not proposed as screen material. Its contribution is a next research action.
The filmmaker might learn how to identify the person responsible for the book, discover that the entries use a convention they had misread, or determine that the apparent gap does not cover the period they thought. Any claim the film later makes still needs appropriate support. An off-screen conversation does not become anonymous permission to assert something as fact.
For our example, the immediate pitch could say:
Preliminary interviews will establish how the maintenance records were kept and who can account for the missing period. These conversations will guide the next stage of research; an on-screen encounter will be proposed only if its function becomes clear.
This is the most defensible starting point when the filmmaker does not yet know whether Jonah can address the central question. It is a decision about the current state of the project, not an argument that research interviews should always remain invisible.
The film may change course after the conversation. Keep room for that change. “Not planned for the cut” describes an editorial intention; it does not replace an agreement about recording and use.
Two filmmakers, two distinct uses of the encounter
Asif Kapadia describes an archive-led approach in a 2019 POV interview about Diego Maradona. His initial plan to film the subject changed. He later recorded conversations without a camera, but says the film's construction would follow the archival research rather than be built around those interviews.[^1] This is his account of the method, not a rule about documentary form or a shot-by-shot analysis of the film.
A contrasting example is available in the official Fresh Air transcript of Sarah Polley's 2013 interview. An excerpt from Stories We Tell includes Michael Polley reading an email, Sarah asking him to repeat a line, and his response to being directed. Polley then explains that keeping the filmmaking process in the film mattered to a work about storytelling.[^2]
At the level documented by that broadcast transcript, the interruption makes the production of an account part of the material. We can discuss that exchange without inventing its framing or claiming to have inspected the entire film.
The useful comparison is not invisible filmmaker versus honest filmmaker. One account describes retrospective interviews working within an archive-led construction; the other provides a documented instance of direction retained in the work. For the projectionist example, the choice is whether the present encounter contributes something the film needs, or whether the voice has a different job alongside the records.
Write a contributor brief that asks instead of predicts
Once the function is clear, the working brief should keep the inquiry and the production intention together. It need not be an exhaustive question script.
For Jonah, specify the knowledge sought: projection-booth routines, authorship of entries, and what he personally recalls about the relevant period. Identify the proposed material: a preliminary conversation about the log, followed by a possible filmed examination if that serves the inquiry and is agreed. State what would count as useful redirection: another record, another witness, a correction, or a limit on his knowledge.
Leave out the emotional result. “He will confront his guilt,” “she will finally forgive,” and “they will reveal the truth” do not become sound production intentions because they are written in a treatment. They assign an inner state or an outcome before the person has spoken.
Also separate editorial function from participation status. An intended contributor, someone willing to talk informally, and someone who has agreed to a particular recording and use are not the same status. The Oral History Association's guidance provides a useful reference for discussing purpose, recording, use, and access with a narrator; it is not a substitute for project-specific agreements or advice.[^3]
The same discipline applies when accounts disagree. Identify what each person had access to and what the disagreement concerns. Do not manufacture a confrontation merely because two statements differ. They may concern different days, different rooms, or different understandings of the question. The film's work may be to preserve a genuine disagreement—or discover that it was not one.
For the pitch, finish with a sentence that connects the contributor to the film's actual need:
We will ask the retired projectionist to examine the surviving maintenance log, so the film can distinguish what the records establish, what he remembers, and which questions require another source.
That sentence does not tell you what Jonah will say. It tells you why the film needs to ask.
Sources
[^1]: Jason Gorber, “Asif Kapadia on His ‘Archival Triology’ of ‘Diego Maradona,’ ‘Amy,’ and ‘Senna’”, POV, September 26, 2019. Questions about the changed interview process and recording method, especially the answer beginning “My plan for interviewing him quickly went out the window.” The published title's spelling is retained. Inspected September 19, 2026 (UTC).
[^2]: Terry Gross, “A Polley Family Secret, Deftly Pieced Together.”, Fresh Air, May 15, 2013. Official transcript: the Stories We Tell email-reading excerpt and the following question about directing the reading. This article uses the transcript and Polley's account, not an independently viewed film sequence. Inspected September 19, 2026 (UTC).
[^3]: Oral History Association, “Oral History Best Practices”, “Preparation” §5 and “Interviewing” §4. Inspected September 19, 2026 (UTC).
Frequently asked questions
What should replace a contributor label like 'the theater's legendary projectionist'?
Replace it with a knowledge or function question: what can this person contribute, what might the encounter change, and does the person need to be seen? For Jonah Vale, a useful function is explaining projection-booth routines and helping distinguish his own entries from material recorded by other staff, while asking what he recalls about the missing period and what records may survive.
Why separate 'I wrote that entry,' 'someone told me why,' and 'looking at it now, I think'?
They have different evidentiary positions. The film can contain all three, but it should not smooth them into the same kind of knowledge. Likewise, a recollection is not independently verified by a pause, close-up, or emotional response, and a blank page does not announce its own cause.
What are the three approaches to the same damaged maintenance log?
One invites Jonah to examine the log on camera, subject to agreement and handling conditions, making the process of working through the object visible. Another separates a retrospective voice from clearly identified historical records. A third keeps the interview in research and follows its consequences, proposing an on-screen encounter only if its function becomes clear.
How should a contributor brief avoid predicting an emotional result?
Specify the knowledge sought, the proposed material, and what would count as useful redirection, such as another record, another witness, a correction, or a limit on the person's knowledge. Leave out 'he will confront his guilt,' 'she will finally forgive,' or 'they will reveal the truth,' which assign an inner state or outcome before the person has spoken.
What do the Kapadia and Polley examples show about interviews?
Kapadia's account describes an archive-led approach where the initial plan to film the subject changed and conversations were recorded without a camera, with construction following archival research. The Fresh Air transcript excerpt from Stories We Tell includes Michael Polley reading an email, Sarah asking him to repeat a line, and his response to being directed, showing direction retained in the work. These are bounded accounts, not rules or full-film analyses.