Documentary research is two jobs
at once. Hold both on one board.

Date every event, and source every date: One card per event with the date, what happened, and where that comes from. A chronology assembled from memory and secondary articles feels solid until a commissioner or a lawyer asks a simple question about one line in it. Attach the record to the event and the question is already answered.

Put both versions up when accounts disagree: Two people remembering the same week differently is normal, not a failure. Keep both cards, connect them, and note what differs: who was present, how long after the fact they spoke, what they had at stake. Very often that difference is the most interesting thing you will find all month.

Mark what is verified and what is still claimed: Label each event as documented, corroborated by two accounts, or single-source and unverified. When you start structuring, that labelling tells you instantly whether the beat you want to open on is something you can actually assert or something you have to attribute.

Keep the timeline separate from the story order: Events happened in one order and the film will tell them in another. Holding both on the same board, as two arrangements of the same cards, is what stops a clever structure from quietly misrepresenting the sequence it was built from.

Separate the contributor from the source: Some people belong on camera and some are the reason you knew where to look. A pre-interview is largely how you find out which is which. Keep both kinds on the board with that distinction written on the card, because treating a nervous expert source as a lead contributor costs you a shoot day and their trust.

Track access honestly, including the silence: Approached, responded, warm, pre-interviewed, agreed, declined, gone quiet. Access is the real currency of documentary and it moves slowly, so a board that shows the current state of every approach is worth more than a director's optimistic summary of it.

Look for who has never been asked: The person every previous piece interviewed will give you the answers every previous piece got. Keep a running region for the people adjacent to the story who nobody has approached: the junior colleague, the neighbour, the clerk, the family that stopped talking to press in 2009. That is usually where a new film is.

Note what each person can actually speak to: Write the specific moments a contributor was present for on their card and connect it to those points on the chronology. Then the holes are visible as stretches of timeline with nobody next to them, which is a much more useful brief for the next month than a general wish for more interviews.

Log the rights holder at the moment you find the asset: Write down the archive, the reference number, and who controls it while you are looking at it. Rediscovering the provenance of a clip six months later, when the edit is locked and someone needs a licence, is one of the most avoidable and most common time sinks in nonfiction production.

Put an estimated licence cost on the card: Archive pricing can be the difference between a sequence and no sequence, and it varies enormously by source and by how you intend to use it. A rough figure on each card, marked as a guess where it is one, means the film's structure is being built against a budget instead of against a hope.

Keep records with the claim they support: Court files, inspection reports, letters, minutes, datasets, and the local paper that covered it at the time. Attach the document to the chronology event it evidences rather than filing it by type, so the paperwork is sitting next to the moment it proves when someone questions that moment.

Flag the sequences that depend on one asset: If a whole act rests on footage from a single holder, that is a structural risk, not an archive detail. Marking it on the board during research gives you months to find an alternative, which is the difference between a redesign and an emergency.

It works from the board as context: Because the AI can see the whole canvas, you can ask it to group your research cards by which part of the chronology they support, to draft candidate act structures from what is there, or to point at cards whose dates or claims look like they conflict. It reasons over what you wrote, so the quality follows your notes.

It does not search archives or find footage: Storyflow has no archive search, no rights or licensing database, and no connection to any footage library. It does not tell you what exists, who owns it, or what it costs. You do that work with the archives themselves and with a researcher, and the board is where you record what you learned.

It does not transcribe, and it does not handle clearances: There is no interview transcription, no release form management, and no legal clearance workflow here. Keep using your transcription service and your production's paperwork system. Storyflow holds the thinking, not the compliance.

Restructure without losing the evidence: Move a card into a different act and everything on it comes along, including the source, the verification label, the rights holder, and the attached file. Trying a completely different shape for the film costs a few drags rather than a rebuild, which is what makes it cheap to keep re-evaluating.