Building a Research Dossier by Hand: A Six-Step Process

Most researchers assemble their dossier the way they clean out a desk drawer: the night before, in a rush, everything into the folder that will fit. Then they wonder why the finished document reads like an inventory and argues like a shrug.

A research dossier is an argument, not an archive. It has to persuade a committee of strangers, many of them outside your subfield, that the work mattered and that your role in it was real. It also has to convince them the next few years are worth betting on. No citation dashboard can make that argument for you. That is why the most durable way to build one is still by hand.

By hand does not mean abandoning the tools. It means keeping the decisions that give a dossier its credibility — what to include, and what each included item proves — with the only person qualified to make them. The six steps below come from how promotion committees, funders, reviewers, and hiring panels actually read these documents, not from how the software would prefer to export them.

• Step one: define the scope. Read the rules before you collect a single file.
• Step two: collect evidence as it happens. Keep a running file from your first week, not your last month.
• Step three: extract, do not just store. Record the claim each item makes and the role you actually played.
• Step four: synthesize by theme, not by output type. Write the through line the auto-export cannot see.
• Step five: verify every claim. Reviewers do check, and one wrong date poisons the rest.
• Step six: compile into one clean document. Structure and naming are arguments too.

The dashboard catalogs; the dossier argues

The temptation to outsource the whole thing is real. Google Scholar, Scopus, Web of Science, and PubMed can each generate a publication list and citation counts in seconds, and a new generation of AI tools will summarise a pile of sources into a tidy brief. What none of them can do is tell a committee why your work should change its mind about you.

The reform movement now governing research assessment has, quietly, made the manual case more important, not less. The San Francisco Declaration on Research Assessment, drafted at an American Society for Cell Biology meeting in December 2012 and signed by more than 23,000 individuals and organisations across 160 countries, tells institutions to judge papers on their own merits rather than the journal that printed them. The Leiden Manifesto followed in 2015 with ten principles for using metrics responsibly.

There is a history here worth keeping straight. The journal impact factor grew out of Eugene Garfield's work at the Institute for Scientific Information, where it was designed in the mid-20th century to help librarians decide which journals to buy. It was never built to grade a single paper, let alone a person. That it ended up standing in for both is a reason to write the parts of a dossier that numbers cannot cover, not a reason to defer to the numbers.

Start with the rules, not the record

Step one sounds administrative, and it is, which is exactly why people skip it. Before you gather anything, write down the boundary: the question your dossier must answer, the criteria it will be measured against — and the page limits it must respect. For an academic research dossier, that boundary lives in your faculty handbook and your department's written guidance. Read both first. Aaron Franklin, associate dean for faculty affairs in Duke's Pratt School of Engineering, calls the third-year review the most important document a junior faculty member receives, because it travels with the dossier and reviewers check whether you responded to its advice. Duke's Office for Faculty Advancement keeps that counsel public.

Step two is a habit, not a task. Keep a running evidence file from your first week on the job: acceptance letters, award notifications, grant decisions both funded and unfunded, conference programs, and records of every student or research assistant you supervised, including where they went afterward. That last detail matters more than it seems; career outcomes are a proxy for the quality of your mentorship. When the editors of University Affairs interviewed fifteen recently tenured academics in the United States, Canada, and Australia, the consistent advice was to download media coverage and invitations rather than trust links to survive, and to save nomination letters alongside award certificates because they explain why you were chosen. Their guide to documenting your work belongs beside the folder. For postdocs, the same file does double duty as the groundwork for a research role that lasts.

The two steps most people skip

Steps three and four are where a dossier stops being a folder and starts being a case. Most candidates do a version of step two. Very few do these well.

Step three is extraction. For each output, write down the claim it makes and the role you actually played. In collaborative fields this is not optional. Jennifer Nash, a distinguished professor of gender, sexuality and feminist studies at Duke, puts it plainly: be clear about the nature of your participation in collaborations, and about how much of the effort was yours. The CRediT contributor-role taxonomy exists for precisely this reason, with named roles such as conceptualization and formal analysis that let a committee see past an author list. If you do not answer the question inside the dossier, someone else will answer it for you, or leave it unanswered, which is worse.

Step four is synthesis by theme, not by output type. Annotate rather than list. Rank your most significant works and say why each sits where it does. The trap is the annotated CV: comprehensive, accurate, and carrying no single idea a reader could leave with. Research on promotion files has flagged exactly this danger — that reviewers will average a wide range of accomplishments into a bland middle ground. Selectivity is the antidote. Leaving strong-but-off-topic work out feels wrong in the moment and reads as confidence in review.

Verify, then compile

Step five sounds like proofreading and is really credibility. Check that every claim in your narrative is attributed to the right source, that dates and titles match your CV exactly, and that your statements do not contradict the record. Committees audit. One wrong date will quietly discount the rest.

Step six is compilation, and it is part of the argument. A dossier with clear headers, bookmarks, descriptive filenames, and searchable text tells reviewers you respect their time; a folder full of files named scan001.pdf tells them the opposite. The University at Buffalo keeps its research statement to three pages and its teaching portfolio to roughly twenty, with the service statement capped at two — read your own institution's numbers as carefully as you read the evidence. The CV is the spine every other document is checked against, so give it a separate, exacting pass; a winning academic CV follows its own logic, and that logic is worth reviewing on its own terms.

What the reform movement actually changes

The push to judge research on its own merits has a name now, and it is older than most people think. DORA's original signatories were editors and publishers trying to stop a journal metric from standing in for peer judgment. The Leiden Manifesto arrived in 2015. In Europe, the Coalition for Advancing Research Assessment, formed in 2022, now coordinates reform across universities and funders. DORA's own site carries the declaration in full.

The practical result is the narrative CV, sometimes called a résumé for researchers after the Royal Society's template, which asks researchers to describe their contributions under four headings: generating knowledge, developing other people, serving the wider research community, and engaging with broader society. The Luxembourg National Research Fund has used one since 2021, and its survey data are the clearest public picture of how the format lands. Applicant satisfaction rose from 48 percent positive in 2021 to 64 percent in 2023. The cost is time: 54.7 percent of applicants say the narrative takes more than a fifth longer to write than a standard CV, and about seventy percent of reviewers in 2023 reported checking ORCID or another source to verify what applicants claimed. The fund's three-year analysis is worth reading in full.

Does the reform make assessment fairer? The early evidence is sobering. A study by the Research on Research Institute across three European funders found reviewers still reach for journal prestige when the decision tightens, and that whether they accept the new formats depends on whether they understand the rationale rather than just the procedure. Treat the narrative turn as a useful tool, not a solved problem. The argument still has to be yours, and it still has to be written by hand.

The one habit that outlasts the deadline

Everything above compounds only if the evidence exists when you need it. The running file is the single habit that survives a job change, a rejected grant, a postponed promotion, and a lost hard drive. Give it a spine: an ORCID iD, the persistent identifier issued to more than six million researchers since the service launched in 2012, plus a stable profile that funders and publishers can read without you in the room. ORCID disambiguates your name and links the work, which is the one manual step genuinely worth automating.

The committee can find your publication list anywhere. The dossier exists for everything the list cannot say: how the work fits together, which parts were yours — and why the next chapter deserves their attention. None of that exports cleanly. Build the file the way you would want your best reader to read it, by hand, one honest claim at a time, with the receipts already in place.

Photo by Markus Winkler on Unsplash