Why Researchers Build a Dossier Before They Write a Single Word

A reporter I know spent the better part of six months gathering material for a long investigation and then confessed she couldn't write a word of it. The problem wasn't a shortage of evidence. It was that the evidence sat in a heap — interview transcripts, court filings, spreadsheets, photographs — with no order, no source labels, and no way to tell which fact supported which claim. The day she built a dossier, one working file that put each fact next to its source in chronological order, the block lifted.

A dossier in this sense is not the finished piece and not a simple folder of saved documents. It is a deliberately arranged collection of everything a researcher has gathered, annotated, and sequenced before drafting begins. The word itself entered English from French, where a dossier was a bundle of papers labeled on the spine. That description still fits: the label tells you what's inside, and the spine holds it together.

People who research for a living build this file first for one unglamorous reason. A draft written before the material is ordered tends to bend the evidence toward the story the writer already believes. The file reverses that. It makes the material lead and the argument follow.

The file is where the argument gets built, not stored

Memory is the weakest part of any long project. Multiply a researcher's facts past a few hundred and the question "where did I learn that?" starts producing shrugs. In a working file, the source sits next to the fact, so the answer is already written down. That single habit is what lets a story survive fact-checking, an editor's questions, or a lawyer's review months after the reporting was done.

Order matters more than newcomers expect. When facts sit in the sequence they actually happened, gaps and contradictions surface on their own: a tone shifts in a run of correspondence, or a name appears in two places where one version must be wrong. The chronology does the first round of analysis for you, before any prose exists.

The deeper reason is about honesty. Build the argument first and you will notice only the material that fits it. Build the file first and the argument has to come from what the evidence can actually carry. That is why the dossier is a thinking tool rather than a storage device — the building happens inside it.

What goes into a working file

Journalism trainers use the least romantic name for this object: the master file. The Knight Center for Journalism in the Americas walks investigators through its parts, and the core idea is that nothing goes in without its provenance attached.

Most working files hold a small set of elements.

• A chronology: events in the order they occurred, so the sequence of the story stays visible.
• A source list: names and contact details, plus a log of when each person was reached and what was agreed.
• A document index: every paper, filing, or file numbered and stored in that order, with a note on where the physical copy lives.
• Quotes and summaries kept apart: exact words from documents and interviews sit in one place, and the researcher's one-sentence summaries sit in another, so the two can never be confused.
• Notes and insights, clearly marked: questions, hunches, doubts, and observations get a tag that separates them from verified material.

The one-sentence summary is the quiet engine of the whole system. It is hard to summarize something you don't understand, so the discipline of condensing each document into a line forces the researcher to actually read it. The Centre for Investigative Journalism's reporting guide recommends reviewing the master file at least once a week and adding insights as they come, because those marginal notes are where the writing genuinely begins.

How working investigators assemble one

Robert Caro, the biographer of Robert Moses and Lyndon Johnson, turns the practice into something close to a life's work. His editor at Newsday told him early: "Turn every page. Never assume anything. Turn every goddamned page." For the Johnson books he conducted 522 interviews and typed up his notes the same night, keeping the pages in drawers he could reach back into when an outline needed a quote. His wife and research partner Ina Caro remains, in his words, "the whole team."

What the file surfaces can change a book. Reading Johnson's correspondence in order, Caro noticed that letters begging committee chairmen for "five minutes of your time" abruptly reversed around October 1940, and the chairmen began writing to Johnson instead. That shift sent him hunting for the reason, and the reason became part of the biography. The NPR review of Caro's "Working" describes the habits that made it possible: same-night transcription and files he could search by hand when a draft demanded a detail.

Investigative reporters build smaller versions of the same machine. Bob Woodward's teaching boils down to a line he repeats to reporters: never leave without the document, and organize and save everything. The British investigator Ray Fitzwalter read eighteen years of council minutes twice to crack the Dundee housing scandal, because the answer was buried in the sequence. More recently, San Francisco Chronicle reporter Susie Neilson traced the financial ties between a prominent geneticist and a motivational speaker by pulling business filings and running names through OpenCorporates, archiving pages to the Wayback Machine so nothing would vanish mid-investigation. The Open Notebook's account of her reporting shows the modern form of the same discipline: capture the document and keep it, then sort it before you try to say what it means.

What goes wrong without one

Skip the file and the failures arrive in predictable forms.

• Claims lose their sources, so nothing can be checked or defended later.
• Editors and lawyers get answers in days instead of seconds.
• Quotes filed without attribution resurface as the researcher's own ideas.
• A mountain of material produces no argument because it was never sequenced.

There is a subtler cost. Unorganized research invites accidental borrowing: a quote gets filed without its attribution, sits in the notes for months, and resurfaces later as the researcher's own thought. The file is what separates "I read this somewhere" from "this is mine and here is the proof."

Writers fall into the opposite failure too, building the file forever and never drafting. That is where the oldest advice still holds. Investigative partners Donald Barlett and James Steele, who spent careers turning documents into Pulitzer-winning series, told younger reporters to "write early and often." The dossier and the draft are not rivals. The file keeps the claims honest; the early draft keeps the file from becoming a comfortable place to hide.

From the file to the first draft

The transition looks different from the outside than it does in practice. A researcher doesn't "finish" the dossier and then start writing; the two keep trading places. Caro's method is the clearest version: a detailed outline comes first, and each point in the outline carries the quote or document that supports it, pulled from the file. The outline is the bridge, written in the file's language of evidence and sequence but shaped like the eventual draft.

Provisional arguments get tested the same way. Instead of deciding the thesis up front, the researcher states what the material "suggests" or "implies," then checks whether the next batch of evidence survives that statement. If it doesn't, the statement changes, not the evidence. This is the moment the dossier pays for itself: the argument can be revised without the writer having to defend it emotionally, because it was never the starting point.

The payoff shows up in work that holds up. Books and investigations built this way survive fact-checks and time, because every sentence can be traced backward to something real. The writer's job in the draft is selection and arrangement: what to leave out and in what order the rest should land. The dossier has already done the heavier lifting of making sure the material is true.

Start with a single spreadsheet. Columns for date, event, source, and a one-sentence summary. Every fact you gather from now on goes in with its source attached, in order, whether you're working on a book, a long investigation, a policy paper, or a company report. You don't need six months of material before you begin. You need the file before you write a single word, because the file is where the thinking happens and the writing is what comes after.

Photo by Joachim Schnürle on Unsplash