Non-fiction readers are generous. They will follow you into unfamiliar territory, trust your judgement and keep turning pages — right up to the moment they spot a date that is wrong, a quote that has been quietly polished into something the speaker never said, or a statistic that turns out to describe Ohio rather than Oldham. Getting the research right is not a chore that sits alongside the writing. It is the writing.
What follows is the working method I would hand to any UK writer preparing a non-fiction manuscript, whether it is a business book, a history, a memoir with reported sections or a popular science title.
Keep a source log from day one
The single most useful habit is dull: write down where every claim came from while the tab is still open. A spreadsheet is ideal. Give it columns for the claim, the source, the type of source, the date you accessed it, the page or paragraph, and how confident you are.
That last column matters more than it looks. Six months later, "confident" and "roughly remember reading this somewhere" need to be distinguishable. When a copy-editor queries a figure at proof stage, you want the answer in ten seconds, not an afternoon of searching a browser history you have since cleared.
Also record the exact wording of definitions. If you write that recycling rates have fallen, you need to know whether the figure you are using covers household waste, municipal waste or all waste, because those numbers tell different stories.
How to judge a source
Three tests filter most material quickly. Proximity: how close is this to the original event, data or document? Transparency: does it show its working — methodology, sample, funding, corrections policy? Currency: is it still true, and has anything superseded it?
Treat secondary sources as signposts rather than destinations. A newspaper report about a study is useful for finding the study; cite the study. If you cannot reach the original, say so in your notes and dial down the confidence of the claim.
Red flags worth taking seriously:
- No named author, no date, no contact details.
- "Studies show" with no study attached.
- A figure repeated across dozens of sites, all traceable to one unattributed blog post.
- A report from a body with a financial stake in its own conclusion, presented as neutral research.
- A single vivid example used to stand in for a trend.
- Quotes that appear identical in several publications, word for word, with different dates.
None of these makes a source unusable. They make it something you must corroborate.
Interviews, quotes and permissions
Agree the ground rules before the recorder goes on, and do not assume you both mean the same thing by "off the record" — conventions differ between journalism, academia and public relations. Say plainly what you will do with the material and how the person will be described.
Ask for consent to record, and confirm it in an email afterwards. A short message setting out who you are, what the book is, how quotes will be used and whether the person will be named is worth more than any reassurance given verbally. If you promise to check quotes for accuracy, be clear that this is a fact-check, not a rewrite.
Tidying grammar is fine. Changing meaning is not. If a speaker's phrasing is confusing but their point is clear, paraphrase and attribute the sense rather than inventing a sentence for them.
Where you are storing recordings, transcripts and personal details, UK data protection law applies. Keep files secure, keep them only as long as you need them, and be careful with anything sensitive or identifying. If your book involves vulnerable people, medical detail or allegations about named individuals, take professional legal advice before you sign a contract, not after.
Checking numbers properly
Go to the original release. Press coverage of statistics regularly drops the caveats that make them meaningful.
Ask, every time: who was asked, how many of them, when, and what exactly was the question? Watch for percentages built on tiny bases, and for the difference between a percentage and a percentage point — a rise from 4 per cent to 6 per cent is two percentage points, and a 50 per cent increase, which sounds far more dramatic.
Check whether an average is a mean or a median; for incomes, house prices and waiting times they can diverge sharply. Check units and currency. If you are comparing figures across years, note whether they are adjusted for inflation, and say so in the text.
Recalculate a few sums yourself. It is startling how often a total does not match its parts. And if a number is load-bearing for your argument, describe its limits in the prose. A reader who trusts you on the caveats will trust you on the conclusion. For anything with financial or legal consequences, point readers to a qualified adviser.
Copyright basics for UK writers
Copyright arises automatically when a work is created; there is no register to check. For literary works it generally lasts for the author's life plus 70 years, after which the work passes into the public domain. Photographs and unpublished material have historically followed different rules, so verify the position for anything you plan to reproduce.
UK law contains limited exceptions, often grouped under the heading of fair dealing — for criticism and review, for quotation, and for news reporting. They are narrower than many writers assume. They generally require sufficient acknowledgement and use no more than is necessary, and they do not permit you to reproduce a substantial extract simply because you are discussing it.
So:
- Short quotations used to illustrate a point are usually manageable; longer extracts need permission from the rights holder.
- A chart or diagram from a report is normally protected even when the underlying data is freely available. Redraw it from the data and credit the source.
- Much UK government material is available under the Open Government Licence, but check each item — some is excluded.
- Song lyrics, poetry and letters are notoriously expensive to clear. Quote sparingly or not at all.
- Ask early. Publishers often have preferred wording and a permissions budget, and chasing a rights holder at proof stage is nobody's idea of fun.
This is general guidance, not legal advice. For a commercial book, or anything contested, speak to a solicitor or your publisher's rights team.
Build fact-checking into the draft
Do not save it all for the end. Mark claims as you write — a simple [CHECK] in the margin — and keep a running list of names, titles, dates and place names to verify in one pass.
When the manuscript is finished, read it once looking only at the claims, out of order. Then read the quotes against the recordings. Then check that a figure appearing in three chapters is the same figure each time, drawn from the same source.
A short pre-submission checklist
- Every factual claim has a source logged, with a date and page.
- Statistics trace back to the original release, not a summary of it.
- Quotes match the recording or transcript.
- Names, job titles at the time of the event, and dates are correct.
- Interview consent is confirmed in writing.
- Anyone criticised has had a fair opportunity to respond.
- Third-party text, images and charts are either yours, cleared, or within an exception you can name.
- Sensitive material has been read by someone with the relevant legal or professional expertise.
Do this well and the research disappears from view, which is exactly what it should do. The reader simply believes you — and that belief is the most valuable thing a non-fiction writer owns.
Photo: DariuszSankowski / Pixabay
