How to Plan a Documentary: A Director's Step-by-Step Guide

What you walk away with

After the eight steps you will have:

• A one-sentence central question, sharper than the question you started with.
• A subject map showing which people are committed and which gatekeepers control access to them.
• A three-act treatment that names the dramatic moment in each act, not the topic of each act.
• An interview question set per subject in three layers: rapport, theme, and turning point.
• A visual treatment with a one-sentence visual rule that your editor and DP can argue against.
• A production schedule with realistic buffer between major phases.
• A pitch deck that funders, festivals, and distributors will actually open.
• A pre-shoot dry-run document that exposes what you would not have caught until shoot day.

These are not deliverables for the sake of deliverables. Each one is what a producer, funder, or programmer will ask for if your film is moving forward. Having them ready closes funding faster, makes shooting cheaper, and makes editing shorter. Having them missing is what kills documentaries between development and post.

How AI fits across the eight steps

Documentary planning is one of those processes where AI is genuinely useful in some steps and genuinely dangerous in others. The table below tells you which mode each step lives in. The detail comes inside each step.

Read the table once. Then forget about the table and read the steps. The detail in each step is where the real work happens.

1. Start with the central question, not the topic

My first documentary was about a community garden. That was the topic. About eight months into production, my editor asked me a question I could not answer in one sentence. The question was: what is this film actually about.

A topic is a subject area. "Homelessness in San Francisco" is a topic. "My grandmother" is a topic. "Urban beekeeping" is a topic. None of these tell you what the film is going to do. They tell you what the film will look at.

A central question is what your film is going to answer. "What does a city owe to people who do unpaid work that feeds it?" That is a film. It tells your editor which scenes belong, which interviews are essential, and which beautiful footage has no place. It tells your DP whether you need wide observational shots or close intimate ones. It tells your funder why this film matters more than the next one in their inbox.

The shift from topic to question feels like a small move on paper. It is the move that decides whether you have a film or a research project.

Common mistake: holding onto your question as a topic for too long. If the only way you can describe your film starts with "it is about," you have a topic. If you can describe it starting with "it asks" or "it is the question of whether," you have a film. Force yourself to write the question form on day one. The version you have on day one is a hypothesis. It will change. That is fine. Without a hypothesis to argue against, every later decision is arbitrary.

How to do this in Storyflow

1. Create a new project named after your working title. If you do not have a working title, name it "Untitled doc" and rename later.
2. On the canvas, drop a Note in the middle. Type your question draft in heading style so it is bigger than everything around it. This is the home of your central question for the rest of pre-production.
3. End your question with a question mark. If it does not end with a question mark, you wrote a topic.
4. Around the question Note, drop three small Notes labeled "Why now," "Why me," and "What is this not." Why now forces you to articulate what makes this a 2026 film and not a 2018 one. Why me forces you to name your unfair advantage with this story. What is this not forces you to cut adjacent stories that share the topic but are not this film.
5. Open the Tactics panel and apply the Documentary Brief Tactic. The Tactic adds prompts for theme, tone, audience, and three-act intent. Do not fill these in yet. They are placeholders for the next two steps.

AI prompt to try

Open AI Chat on the canvas. Paste this:

I am developing a documentary with this central question: "[paste your draft question]." Help me sharpen it. First, ask me 5 questions that would expose weakness in this question. Things like: is the question too broad to answer in 90 minutes? Is it actually two questions stitched together? Is it a thesis in disguise? Then, suggest 3 alternative phrasings that make the question more specific without narrowing the documentary's scope. Do not write a better question than mine. Stress-test the one I wrote.

The last sentence is the trick. Without it, the AI will try to be helpful by writing a "better" question, and you will spend three days arguing with yourself about whether to take its suggestions. The AI is good at finding weak points. It is not good at finding the question that obsesses you. Keep that role for yourself.

When to skip AI

The first version of your question. The version that comes from sitting with a story you cannot put down. The version that arrives on a long walk or in the shower or at three in the morning. AI can stress-test that version. AI cannot generate it. Asking AI for question suggestions before you have written your own gives you a question that is technically interesting and has nothing to do with why you wanted to make this film. Write your draft first. Then bring AI in.

2. Map your subjects, sources, and access

A documentary has three categories of people. Most first-time directors collapse them into two and lose six weeks because of it.

The categories are: subjects (the people who appear on camera), sources (people who give you context off camera, like historians, academics, family members of subjects), and gatekeepers (the people or institutions who control access to subjects, locations, or archives). The gatekeepers are the category most directors forget exists.

On my second film, I scheduled twelve shoot days assuming my primary subject was the gatekeeper for our key location. He was not. The location was on city-owned land controlled by a parks department that had its own approval timeline. I found this out four weeks before shoot day. We made it work, barely, but we shot the film in a different order than I had planned and the edit suffered for it.

Common mistake: building your shoot schedule before you have committed access from your primary subjects in writing. A subject who said yes in March and stops returning calls in June is the most expensive lesson in pre-production. Verbal yes is a maybe. Written yes is a yes. Tape of them saying yes on camera is a hard yes. Schedule against hard yeses only.

How to do this in Storyflow

1. On the canvas, create a section labeled "Subjects." Drop one Note per subject. Each Note holds: full name, role in the story (the function they serve, not their job title), status (interested / committed / declined), and contact channel.
2. Below it, create a section labeled "Sources." Same format, but for context interviews. These are the people who help you understand the world your film lives in but who may or may not appear on camera.
3. To the right, create a section labeled "Access." Drop Notes for gatekeepers: institutions, locations, archives, family estates, anyone who can grant or deny something you need. Each gatekeeper Note names what they control.
4. Draw a line from a subject Note to the gatekeeper Note that controls their access. A line means "this is conditional." No line means "I have direct access." When you look at the canvas, you should see at a glance how many of your subjects depend on a third party who has not yet said yes.
5. For each subject, attach a Document on the right panel. Title it "[Subject name] research." Inside, paste prior interviews of theirs you have read, public statements, articles, transcripts of any conversation you have had with them. This Document is the source material AI will use to draft interview questions in step 4. It is also the source material you will reread the morning of every interview.

AI prompt to try

Once a subject has a research Document with at least 1,000 words of source material in it, ask AI:

Read the attached research Document about [subject name]. Identify: 5 themes I should ask them about, 3 areas where their public statements seem to contradict each other or evolve over time, and 2 questions they have likely never been asked publicly that fit my documentary's central question.

@-mention the research Document in the AI chat so the model has the source material. The contradictions are the part of the output that matters most. They tell you where this subject has changed their mind, walked something back, or held two positions at once. Those are the moments where an interview gets honest.

Where AI is a trap: cold outreach

Do not use AI to draft the message you send asking a subject to participate. Subjects can sense AI-written outreach inside three seconds. The pattern is too clean, the empathy is too generic, the asks are too symmetric. You will lose access by sending it.

The outreach is a relationship document. The lift in your handwriting is what sells it. Write it yourself. Five short paragraphs: who you are, why this film, why this person specifically (cite a thing they said or wrote that moved you), what participation involves, what control they have over their material. Send it through the channel they actually use, not the one you wish they used. If they live on Instagram DMs, send an Instagram DM.

3. Build your story spine

You have a question and a subject map. The treatment is where you decide what shape the film takes when you string events together.

A treatment is not a script. Documentary treatments are not written in dialogue. They are written as the structural argument of the film. Three acts is enough for most first documentaries. Some of my films use a five-act structure with two midpoints, but that is a choice you make later. Three acts first.

Act One is the world before the story changes. Act Two is the change. Act Three is the world after. Each act has a turning point that forces the next one. If you cannot name the turning point in an act, the act is not yet structured.

The treatment is also where you make your biggest cuts. Most of the documentary footage you have not yet shot exists in your head as ideas you have not committed to. The treatment forces those ideas into a yes or a no. Most of them get a no.

Common mistake: writing a treatment that explains the topic instead of dramatizing the question. If your treatment reads like a Wikipedia article, your film will too. A treatment is a story that has not been shot yet. If your beats are facts, your beats are wrong. Beats are moments. Moments have a subject, a place, an action, and a reason that moment matters to the question.

How to do this in Storyflow

1. On the canvas, open the Documentary Treatment Tactic from the Tactics panel. The Tactic creates a structured prompt set covering hook, inciting incident, rising action, midpoint, climax, and resolution. Six beats.
2. Fill in each beat in your own words. Each beat is one to three sentences. Resist the urge to be comprehensive. A beat is what happens, not everything that happens.
3. As you fill in beats, draw lines linking them visually to subject Notes from step 2. The character who appears in your inciting incident should be connected to that beat by a line. This catches the gap where you have written a beat with no subject yet attached, which is one of the most common reasons treatments fall apart later.
4. Open AI Chat. @-mention the Documentary Treatment Tactic and the central question Note in the same message. Ask AI to read both and tell you whether your treatment dramatizes your central question or describes its topic. The output is rarely perfect. The friction is what is useful.

AI prompt to try

After you have all six beats drafted, paste:

Compare my treatment to my central question. For each act, tell me: which subject carries this act, which moment dramatizes the question, and where the act risks turning into exposition instead of story. Identify the single weakest beat and explain why.

The "single weakest beat" instruction is the part that does the most work. Without it, AI will give you balanced feedback across all six beats, which is useless because you already know all six are imperfect. Forcing it to pick one tells you where to revise first.

When to keep AI out

The first draft of your three-act spine. Write the bones yourself before AI ever sees them. AI is good at evaluating a draft and weak at conjuring one from a question. The first draft is also where your voice as a director starts to show up. AI flattens voice. It writes correct treatments that sound like everyone else's. Write the bones in your own hand. Then bring AI in to stress-test.

4. Develop your interview questions

A subject is going to tell you their story whether you ask the right questions or not. The right questions get you a film. The wrong questions get you eight hours of footage that has to be carved into a story in the edit, which doubles your timeline and triples your cost.

Plan three layers of questions per subject. The layers are not equal. The layers do different jobs.

Layer one is opening questions. Five of them. Low-stakes. About their morning, their work, their hands. The job of layer one is to get them used to talking to a camera and to remind them you are a person.

Layer two is thematic questions. Ten of them. These come from the research Document and your treatment. They are the questions that tie this subject to the question your film is asking.

Layer three is turning point questions. One or two. These are the questions you cannot un-ask. The questions that, if answered honestly, change the film. Plan them. Sequence them late. Have a plan for the silence after.

Common mistake: asking only thematic questions. A subject who has not been warmed up answers thematic questions with cliches because that is what their public-facing self is for. Ten minutes of opening questions is the price of fifty minutes of usable footage. Cutting opening questions to save time is the most expensive ten minutes you will ever save.

How to do this in Storyflow

1. For each subject Note from step 2, attach a Document on the right panel. Title it "[Subject name] interview questions."
2. In the Document, draft your three-layer question set yourself. Five opening, ten thematic, two turning point. Write the questions in the order you would ask them, not the order they came to you.
3. Save the Document. Walk away for an hour. Come back and read it out loud. Anywhere your voice catches or you feel embarrassed asking the question, that question is either too clinical or too pushy. Rewrite.
4. Now open AI Chat. @-mention the subject's research Document and the central question Note. Ask AI to suggest twenty thematic questions specific to this subject. Read all twenty. Add the strongest five to your Document. Discard the rest. AI overproduces. Curating is your job.

AI prompt to try

Below is my research Document about [subject name] and my documentary's central question. Write 20 interview questions specific to this subject. Ten should be questions any subject in this role might be asked. Ten should be questions that would only make sense to ask this specific person, drawing on details from their research Document. For each, briefly note what kind of answer you expect (fact, opinion, story, reflection).

The "what kind of answer you expect" instruction does two things. It makes the AI think harder about what each question is for. And it shows you the questions whose expected answer is "fact." Those are the weakest. Facts can be looked up off-camera and do not need to come from this subject.

When AI is dangerous

Trauma, marginalized communities, sensitive topics. AI will produce questions that are technically correct and emotionally dangerous. It does not have lived experience of the topic. It cannot tell you when a question is invasive in this specific context. The bar for these interviews is not "is this an okay question." The bar is "does asking this serve the subject."

For sensitive interviews, do not let AI draft the question set without a human review by someone with lived experience of the topic. Pay the sensitivity reader. Build the time for them to actually read the questions. This is not a place to optimize.

5. Plan your visual approach

A documentary lives or dies on its visual treatment. Two films with the same subject and same questions can be told in opposite cinematographic languages and feel like different films. The visual treatment decides which one yours is.

Three components. The mood board is the emotional and aesthetic reference. The shot list is the specific shots you intend to capture. The visual rule is the one-sentence statement of how this film will look that all other choices serve.

The visual rule is the part most first-time directors skip. They build a beautiful mood board and assume the look will emerge. It does not emerge. It gets argued out of existence by a tired DP at the end of a long shoot day.

A visual rule is a constraint. Examples from films I have worked on or watched closely:

• "Every interview is shot at 50mm, never wider, never longer."
• "Every interior shot is lit by a practical light visible in the frame."
• "The first act is hand-held. The second act is locked off. The third act returns to hand-held."
• "Every cutaway lasts at least four seconds before a cut."

The rule does not have to be poetic. It has to be a constraint your editor can argue against. "Cinematic" is not a rule. "Beautiful" is not a rule. The rule has a verb in it, and an editor can hold it up next to a shot and say yes or no.

Common mistake: mood board with no rule. A Pinterest dump dressed up as a treatment. Your DP gets a folder of pretty images and no instructions for how to apply them. They make their best guess on shoot day, you change your mind on day three, and the film looks like seven different films stitched together. The rule prevents this.

How to do this in Storyflow

1. On the canvas, create a section labeled "Visual treatment."
2. Inside, create a sub-board for the Mood Board. Add 20 to 40 reference images: stills from films with a similar tone, photography that matches the subject's world, color palettes that fit the emotional arc. The rule is variety: do not let one reference film dominate the board, or you will accidentally make a worse version of that film.
3. Above the mood board, write your one-sentence visual rule in a Note. Make it bold. Make it bigger than the surrounding Notes. Your DP should see it before they see the references.
4. To the right, create a section for the Shot List. Apply the Shot List Tactic. The Tactic structures shots by scene, type, and priority (essential, nice-to-have, aspirational).
5. Open AI Chat. @-mention the visual rule Note, your mood board, and the treatment from step 3. Ask AI to suggest 5 specific shots for each act of the treatment that would honor the visual rule and capture a key moment in that act.

AI prompt to try

Below is my visual rule for a documentary, my mood board, and my three-act treatment. For each act, suggest 5 specific shots that honor the visual rule and capture a key moment in that act. Format each shot as: shot description, framing, lens choice, lighting note, why this shot. Then flag any shot you suggest that bends the visual rule, and explain how.

The "bends the visual rule" instruction is what makes the output usable. AI will sometimes suggest shots that are great but break your rule. You want to see those flagged so you can decide whether to break the rule for that shot or rewrite the rule.

The director's eye

The final visual call is yours. AI can suggest a wide-angle shot held for eight seconds. Only you know whether your subject's face deserves a wide or a close-up in that beat. The shot list is a starting point. Overrule freely. The shots you remove are as important as the shots you keep, because every removed shot is an hour of camera time you give back to the things that matter.

6. Build your production schedule

Most documentaries do not break in production. They break because production was scheduled wrong. A first-time director schedules ten shoot days, films for fifteen, edits for six months instead of three, and runs out of money at the worst possible moment in post.

Two layers. The macro schedule is research, shoots, transcription, paper edit, rough cut, fine cut, color, sound, delivery. The micro schedule is shoot day plans, location logistics, crew calls, equipment lists. Build the macro first. The micro fills in once the macro is locked.

The number that matters most in the macro schedule is the buffer. Two weeks of buffer between major phases is the floor for a 90-minute film. Crew get sick. Subjects cancel. Hard drives fail. The footage you shot on day three is incoherent and you need a reshoot. None of this is rare. All of it costs you days. Schedule the days in advance and you absorb the cost. Do not, and the cost rolls into the next phase, and eventually the cost rolls off the edge.

Common mistake: scheduling shoots back-to-back. Documentary shoots reveal what you do not yet know. You need a gap of at least one week between major shoot blocks so you can transcribe, watch back, and adjust your remaining questions. Without that gap, you film the same material three times because you keep asking the same questions to subjects who could have given you new material if you had listened to the previous tape.

How to do this in Storyflow

1. On the canvas, create a section labeled "Production schedule."
2. Apply the AI Planner to this section. The Planner reads the active canvas (treatment length, shot list size, subject count) and drafts a phased timeline.
3. Open AI Chat. @-mention your treatment, your shot list, and your subject Notes. Ask AI to draft a macro schedule with realistic buffer between phases.
4. Edit the draft. AI does not know your primary subject is only available three weekends in August. You do. Confirm subject availability against the schedule, not the other way around.
5. Once your macro schedule is committed, create a Kanban inside the schedule section. Use the AI Kanban to track each phase from planned to in-progress to complete. The AI Kanban reads the active canvas and can flag when a phase has a dependency on something elsewhere on the canvas (an unbooked location, an unconfirmed subject) that has not yet been resolved.

AI prompt to try

Below is my documentary treatment, shot list, and subject availability notes. Draft a 12-month macro schedule covering research, shoots, transcription, paper edit, rough cut, fine cut, color, sound, and delivery. Assume a 90-minute final film and a 3-person crew. Flag every phase whose dependencies I have not confirmed yet. Treat 2 weeks as the minimum buffer between major phases.

The flagged dependencies are the most useful part of the output. They are the list of things you should reconfirm with subjects, locations, and crew this week, before they become broken commitments later. Print the list. Send a confirming email for each item. Treat the list as your producer's daily checklist for the next two weeks.

Reality always wins

A schedule is a hypothesis. The schedule that survives contact with reality is the one you built with slack inside it. The schedule that does not survive is the one that assumed everything would go right because everything had to go right for the film to come in on budget. Build slack into every phase. Account for the days you will lose to weather, illness, equipment failure, subject mood, your own mood. The point of the schedule is not to be a contract you sign with yourself. The point is to be a useful map for a journey that will not go exactly the way you mapped it.

Try Storyflow free and see how the AI Planner reads your treatment, shot list, and subjects to draft a schedule grounded in your specific project.

7. Build your pitch deck and funding strategy

Most documentary filmmakers do not enjoy the pitch part. I did not enjoy it for years. It is not optional. Your film will need money, festival access, distribution, audience attention. The deck is what gets it.

A documentary pitch deck has eight standard slides: logline, synopsis, central question, characters, visual treatment, director statement, budget summary, ask. The version for funders looks different from the version for festivals, but the spine is the same. You build the deck once and then make small variants for each audience.

Common mistake: writing the deck after the shoot. The deck is part of pre-production. The deck is what funds the shoot. Writing it after means you funded the shoot yourself, which is the most common reason documentaries die in post. Money runs out at the moment you most need it: when there is footage on hard drives but no film yet.

The deck is also where the central question earns its keep. Every slide either traces back to the question, or the slide does not belong in the deck. If you cannot draw a line from the slide to the question, cut the slide.

How to do this in Storyflow

1. On the canvas, create a section labeled "Pitch deck."
2. Apply the Pitch Deck Tactic from the Tactics panel. The Tactic creates a structured slide set with the eight standard sections.
3. Fill each slide using material you already have on the canvas. Central question from step 1. Characters from step 2. Treatment summary from step 3. Visual rule from step 5. Schedule and budget summary from step 6. Most of the deck is connecting work you have already done.
4. The director statement is the only original copy you write here. Treat it as personal. Why you, why now. Two hundred to three hundred words. Read it out loud. If it does not sound like you, rewrite. Funders give money to filmmakers. Not to topics. The director statement is where they decide.
5. Once the deck is drafted, open AI Chat. @-mention the Pitch Deck Tactic. Ask AI to write three logline variants and three synopsis variants. Pick the best one of each. Discard the rest.

AI prompt to try

Below is my documentary's central question, treatment, character map, and visual rule. Write 3 logline options, one sentence each, under 25 words. Write 3 synopsis options, one paragraph each, under 100 words. For each logline and synopsis, briefly note the audience or funder type that version is best aimed at: impact funder, festival programmer, broadcaster, distributor.

The audience-typing is the part of the output that matters. A logline that works for a festival programmer rarely works for an impact funder. Programmers want story and craft. Impact funders want change and reach. Having three loglines aimed at three audience types means your deck travels.

The director statement is yours alone

This is the one place in the entire planning process where AI should write nothing. Not a draft. Not an outline. Not a "starting point." Funders read the director statement to decide whether they trust you with their money for the next year and a half. They are looking for a real person on the other side of the page. AI can write a competent statement. It cannot write yours. The competent version is worse than the rough honest one in your own handwriting.

Two hundred words is enough. Why this story. Why you. What you are afraid of getting wrong. What you have at stake. Read it out loud to a friend. If you do not flinch at one line in it, it is not honest enough.

8. Pre-shoot dry-run

Every film I have made has been better because of a pre-shoot dry-run. It is the cheapest insurance you will buy in your career.

A dry-run is a one-day session held two to four weeks before your first shoot day. The whole production team meets. You walk through the entire shoot in your head, on the canvas. You do not pick up a camera. You read every interview question aloud. You walk every location on a map. You confirm every piece of equipment is accounted for and on its way. You identify everything that could go wrong and decide whether the plan handles it.

Common mistake: treating the dry-run as a producer's status meeting. The dry-run is a stress test, not a check-in. You are looking for the question that is not in your set, the location you do not have permission for, the lens that is not on your equipment list. If everyone leaves the dry-run feeling good, you did not push hard enough. The dry-run is an adversarial process where each person on the team uses their own role-knowledge to find the failures the team did not catch on its own.

The dry-run is also the moment your team performs together for the first time. Dry-run quality predicts shoot-day quality more reliably than any other indicator I have tracked. A team that finds twelve red Notes in a dry-run has a team that catches problems on shoot day. A team that finds three red Notes in a dry-run has a team that misses problems on shoot day.

How to do this in Storyflow

1. Open the project canvas. The whole production team meets here. If your team is on the Team plan, the canvas updates in real time as everyone edits. If you are on Pro, the team can comment asynchronously and you take live notes during the session.
2. Walk through every section in order: central question, subjects, treatment, interview questions, visual treatment, schedule, pitch deck.
3. For each section, ask: what is this depending on that has not yet been confirmed? Drop a red Note on every dependency that is open. The goal of the day is to find every red Note before shoot day.
4. After the dry-run, the producer (or director if no producer) takes ownership of every red Note. Each Note has a name and a deadline. Each Note closes before shoot day or the shoot day moves. There is no third option.
5. Save the dry-run notes as a Document. After the shoot, review the Document with the team to see what the dry-run caught versus what surprised you anyway. This becomes the institutional memory for the next film.

Where AI helps after the session

The session itself is human. AI cannot tell you that your equipment list is missing the wireless rig you assumed your DP was bringing. AI cannot tell you that the location agreement has not been countersigned. The team is the AI in this step.

AI is useful afterwards. Once the dry-run notes are saved as a Document, ask AI to read the Document alongside the rest of the canvas:

Below is my dry-run notes Document and my full project canvas. Compare the dry-run notes to the treatment, shot list, subject Notes, and schedule. Identify any subjects, scenes, beats, shots, or schedule phases that the dry-run did not stress-test. List them as questions I should bring back to the team in a follow-up.

The output catches gaps the team missed because the team got tired or ran short on time. Most of the gaps are minor. A few are not. The few that are not are the reason you do this step.

Why the session is human

The dry-run cannot be done by AI because the dry-run depends on each person bringing their unique role-knowledge to the room. The producer knows what is unconfirmed. The DP knows what equipment is uncertain. The director knows which interview question they are afraid to ask. AI does not have the context of being on this specific team for this specific film. The team is the model. AI is the post-session reviewer.

Common mistakes I see in first documentaries

These are the patterns I see again and again. None of them are fatal alone. Most documentaries that die in production die from three or four of these stacked.

Choosing a topic instead of asking a question. The single biggest one. A topic does not force decisions. A question does. If your film still feels like a topic at week six, go back to step 1. Spend a week with the question. The week is cheaper than the six months you will spend in the edit trying to find the question retroactively.

Not confirming access in writing. Subjects say yes early and disappear later. Build your schedule on subjects who have committed in writing or on tape, not on subjects who have said yes verbally on a phone call you have not transcribed.

Over-shooting because the treatment is weak. A weak treatment leads directors to film everything in case they need it later. Most documentary post problems are pre-production problems wearing post-production clothes. The fix is a stronger treatment, not more footage.

Skipping the visual rule. A mood board without a rule is a Pinterest dump. The film looks like seven different films because the DP made seven different guesses across seven different shoot days. The rule prevents this.

Building the pitch deck after the shoot. The deck funds the shoot. Writing it later means you funded the shoot yourself and now you are out of money for post. This is the most common reason documentaries die in post.

Treating the dry-run as a meeting. A dry-run that does not surface red Notes did not happen. Push until the team finds the failures. That is the entire point.

Letting AI write the parts that are personal. The director statement, the cold outreach, the trauma-aware question check. AI is correct in form and dangerous in content for these. Keep it out.

Not budgeting for transcription. Every interview gets transcribed. Plan it into the schedule. The paper edit that comes from transcripts is the document that turns hours of footage into a film. Skipping transcription costs you a month in the edit, every time.

How I used this process for the urban beekeepers documentary

Two years ago I started a documentary about urban beekeepers reshaping their city's relationship to its food supply. The first three weeks of pre-production was me sitting with the question until it stopped being "what is the urban beekeeping movement" and became "what does a city owe to people who do unpaid work that feeds it."

That single shift saved the film. Every subject I added, every scene I scouted, every question I prepared had to either answer that question or get cut. The treatment came together in two weeks because the question made the structure obvious. The pitch deck took three days because the question was the spine of every slide.

The dry-run found two things I would not have caught in another month. Our primary subject's apiary was on city-owned land that required permission from a parks department I had not yet contacted. We had three weeks. We barely made it. The second thing the dry-run found: my DP and I had different assumptions about the wireless audio rig. Easy fix in the dry-run. A four-figure problem on shoot day if it had surfaced there.

The shoot ran twelve days, two days under our scheduled fourteen. The treatment told us exactly what we needed and the question told us when we had it. Post took four months instead of the typical six. The film delivered ahead of schedule and under budget. The planning is what made it possible. None of it was complicated. All of it was deliberate.

The tools you will actually use

Documentary planning is a multi-tool process. Storyflow is the canvas where the whole thing lives, but you will use these alongside it.

Storyflow for the project canvas, treatment, mood board, shot list, schedule, and pitch deck. The free plan supports 3 projects and 10 AI generations a month, which is enough to evaluate the workflow on a real project. Pro at $14.99 per month annually unlocks unlimited AI and 200+ Blueprint Tactics including the Documentary Treatment, Shot List, and Pitch Deck Tactics referenced in this guide. Real-time team co-editing on the canvas is on the Team plan starting at $12.74 per user per month annually.

A transcription tool for interview transcripts. The dedicated tools handle this better than canvas tools do. See the best AI research tools 2026 for source-grounded options like NotebookLM that work for transcripts and for research.

A storyboard or shotboard tool for the visual treatment if you want frame-level fidelity. See the best storyboarding software 2026 for tools that handle frame sequencing better than canvas tools.

A pre-production logistics app for call sheets, location agreements, and equipment tracking that the canvas does not handle. See the best pre-production tools 2026 for the dedicated production-logistics options.

A research tool for subject background and source archives. Storyflow holds the synthesis on the canvas, but the heavy reading lives in dedicated tools. See AI second brain for documentary filmmakers for a research-first workflow.

You are ready

You have an eight-step planning process that takes a documentary from a question to a dry-run. You have a Storyflow workflow that ties every step together on one canvas. You know which steps benefit from AI and which steps to keep AI out of.

Start with step 1. The central question is the lever that moves every other step. Spend a week with it. Then move forward. The week you spend on step 1 is the most leveraged week of the entire production.

If you want to compare this canvas approach to traditional production-management tools, read the best pre-production tools 2026. If you want to see how documentary research specifically works on a canvas, read AI second brain for documentary filmmakers.

Try Storyflow free and walk through this guide on a real documentary project.

Author

By Justkay, Documentary Filmmaker and Founder of Storyflow. I have made three documentary films using a version of the process in this guide. Storyflow exists in part because the planning canvas I built for myself became the product I wanted to use on my next film.