Somewhere in every field there is a report everyone quotes. It established a number, named a gap, or made a comparison that became the shared reference point for the whole argument. Such reports are not usually the largest or best funded in their field. They are the ones built so that a stranger, on deadline, with no time to read, can lift one finding and stand on it. That is a craft with learnable parts.
Begin with the finding, not the topic
Topics do not get cited; findings do. "Housing in Europe" is a topic. "In three European capitals, rents paid by new tenants rose faster in the past four years than in the previous twelve combined" is a finding, and the difference is what makes one report quotable and the other wallpaper.
The working method is to write the finding as a sentence before any drafting, then test it. Is it specific, with numbers and a defined period? Is it checkable, meaning a determined reader could trace it to sources? Is it new, meaning you cannot find it stated anywhere else? A finding that passes all three becomes the spine of the report; every section either builds it, supports it, or pressures it. Sections that do none of those things are decoration, and decoration is what readers skip.
Assemble evidence the way a skeptic will attack it
Assume your hardest reader is a rival analyst with an afternoon to spend destroying your finding. That reader checks three things first: where the numbers came from, whether the arithmetic holds, and what you left out. Building for them means sourcing every load-bearing number to its primary origin, filing, registry, official dataset, named publication, and stating your calculation steps at least once in full so the reader can watch the machinery turn.
It also means hunting your own counter-evidence while you still have time to change the finding. The strongest reports contain their own strongest objection and answer it, because the authors found it first. Reports that never mention inconvenient data are not hiding it from anyone; the skeptics find it, and then the whole report is reduced to a cautionary example. This is not just honesty, it is self-interest, aligned.
Write the quotable sentence, deliberately
Citations happen at sentence level, and the sentences that get lifted have a common shape: one clause of finding, one clause of scale or contrast, no hedging a reader has to wade through, and no jargon standing between the fact and the reader. "The region's five largest employers added more debt in eighteen months than in the previous six years" can be printed as-is with attribution. "Debt accumulation dynamics exhibited marked acceleration during the observed period" cannot, and no editor will try.
The discipline is to write these sentences on purpose. Each section of the report should contain at least one sentence that could survive extraction, because extraction is how reports travel: into briefings, into other articles, into the answer an AI system gives when someone asks the question your report addressed. Systems that quote sources favour text that is precise, self-contained and clearly attributed. Write for the lift.
Charts are citations waiting to happen
The charts that get reused share properties worth copying: they answer exactly one question, they carry their source and units on the image itself rather than in a caption nobody screenshots, and they are simple enough to be read at briefing-room size. A chart with three competing messages gets ignored; a chart with one message gets stolen, with credit, which is the goal. Where your data is checkable, consider publishing the underlying table. Reports that let readers verify become the reference; reports that ask to be believed become the footnote.
Structure for the reader with ninety seconds
Assume three readers: the one with ninety seconds, the one with ten minutes, and the one who reads everything. The ninety-second reader needs the finding, the strongest supporting number, and the biggest caveat, which means those three items belong at the top in that order, in prose, not buried in a summary written last. The ten-minute reader needs the method and the evidence, which is the report's middle. The full reader needs the appendix and the sources, which is the end. Reports written in this order get read at all three depths; reports written as an unbroken ascent get read at none.
Why openness compounds
Reports get cited when they are useful, and usefulness is a form of generosity: the report gives away its finding cleanly, explains where it came from, admits what it cannot show, and asks for nothing except to be checked. The possessive reports, the ones that bury sources, inflate certainty and guard their data, get quoted less even when they are right, because quoting them costs the quoter something. Openness is not the opposite of authority. Over time, it is where the authority comes from.