A professional fact check on a nonfiction book costs more than editing it. Published rates from working fact-checkers start around $10,000 and run past $30,000 on heavily sourced work, and your publisher almost certainly won’t pay for it. That surprises nearly every author who hears it, and it leaves most of them with a question nobody answers: what do you actually do about accuracy when a full check is out of reach? Here is the honest answer, in tiers, starting with the part you already get for free.
Key facts
- Full professional check Commonly $10,000 to $30,000 for a book
- Published flat fees Around $12,000 for narrative nonfiction
- Hourly, 2026 medians $50 to $60 an hour
- Who pays Usually the author, not the publisher
- Against the cost of editing More than all four editing stages combined
- Included in a copy edit Flagging, not verifying
- Check first, if nothing else Names, numbers, dates, and quotations
- Legal exposure sits with You, in almost every contract
What is fact-checking, and how is it different from editing?
Fact-checking is independent verification of every checkable claim in your book against a source. It’s not proofreading, it’s not part of a copy edit, and it’s not something an editor does quietly while working on your sentences. It’s a separate discipline with its own practitioners, and it’s billed separately.
A fact-checker works claim by claim rather than page by page. They build a list of every assertion that can be verified, then verify each one: a date, a job title, the year a company was founded, whether the person you quote actually said it, whether the study you cite says what you say it says. On serious work, they go back to the primary source rather than to a secondary account of it.
The distinction that matters commercially is this. A copy editor flags what looks wrong. A fact-checker confirms what is right. Those are different jobs with different costs, and confusing them is why authors are surprised twice: once by what a copy edit doesn’t cover, and again by what a fact check costs.
What does it cost, and who pays for it?
Published rates from working book fact-checkers commonly start around $10,000 and run past $30,000 on heavily sourced work. One well-known checker states a floor of $12,000 for a narrative nonfiction book; an agency publishes $12,000 for a 70,000- to 80,000-word project; authors who have written about their own budgets report figures around $20,000. The 2026 industry survey puts fact-checking at $50 to $60 an hour, which reaches the same place once you count the hours a book takes.
Now the part that lands hardest. In trade publishing, the publisher usually doesn’t pay for it. Magazines have fact-checking departments; book publishers, with rare exceptions, don’t. The contract typically warrants that the work is accurate and puts the consequences on the author. So the cost falls to you, and so does the exposure.
Set that against what editing costs and the picture gets uncomfortable:
Table 1. Fact-checking against editing, 80,000-word nonfiction book.
| Service | Typical cost | What you get |
|---|---|---|
| Structural edit | $3,200 to $4,000 | The shape of the argument |
| Line edit | $2,800 to $4,000 | How it reads |
| Copy edit | $2,400 to $3,200 | Correctness, consistency, flagged doubts |
| Proofread | $1,600 to $2,400 | Errors on the finished pages |
| All four editing stages | $10,000 to $13,600 | A finished, correct book |
| Full independent fact-check | $10,000 to $30,000 | Verified claims and a source trail |
Editing figures are 2026 medians from the Editorial Freelancers Association rate chart. Fact-checking figures are the published rates of professional checkers and agencies. (Note that these are the industry’s average editing rates. Senior editors with in-house publishing backgrounds and, better still, verifiable multiple bestseller and award-winner credits should charge above these bands.)
So a full check can cost as much as the entire editing sequence, or three times as much. That’s why the honest version of this article is not “hire a fact-checker.” It’s “here’s what accuracy actually requires, and here’s how to get most of it for a fraction of that.”
What does your copy editor already do about facts?
They flag, and flagging is more valuable than authors expect. A competent copy editor reads with a running sense of what is checkable, and queries anything that looks wrong, internally contradictory, or improbable. What they don’t do is verify it because that’s a different job and it’s not in the fee.
What arrives in a copy editor’s query list, at no extra cost:
- Internal contradictions. The figure on page 40 disagrees with the table on page 190. This is the single most common factual problem in nonfiction and it’s entirely catchable without any research.
- Dates and arithmetic that don’t work. A person aged 34 in 1998 who was born in 1961. A percentage breakdown summing to 103.
- Claims that are obviously checkable and look off. A treaty in a year that doesn’t match, a song released after the scene it appears in.
- Quotations that don’t sound like the person. Not verified, but flagged for you to confirm.
- Names, titles, and spellings of real people and organizations. Regularized and queried where inconsistent.
That is a genuine layer of protection and it comes with a service you were buying anyway. But it’s not a fact-check. Treat it as the floor rather than the ceiling, and treat every query as a task rather than an annoyance: an unanswered query becomes a published error with a paper trail showing you were warned.
How do you get accuracy on a real budget?
By ranking your claims by consequence and buying verification only where the consequence is high. Nobody outside a magazine checks every sentence, and pretending otherwise is what makes authors give up on the whole question.
Table 2. What to do about accuracy at each budget, 80,000-word nonfiction book.
| Budget for accuracy | Do this |
|---|---|
| $0 | Answer every copy editing query properly. Self-check names, numbers, dates, quotations. Build a source list. |
| $500 to $1,500 | Pay a checker for a targeted pass on your highest-risk chapters and every claim about a living person. |
| $2,000 to $5,000 | A partial check: all quotations, all statistics, all named individuals, plus anything legally exposed. |
| $10,000 and up | A full independent check, claim by claim, against primary sources. |
The tier that does the most work is the free one, and it does more than authors believe. A systematic self-check of the four categories below catches the overwhelming majority of errors that reach print because most published errors aren’t exotic research failures. They are a misremembered middle initial, a number transposed while retyping, a date carried over from an earlier draft.
What actually gets nonfiction books into trouble?
Not obscure factual slips. Three things, and only one of them is what people worry about.
Claims about living people. This is the whole of your legal risk and most of your reputational risk. A wrong date embarrasses you; a wrong assertion about what a named person did can end in a lawyer’s letter. Every statement about an identifiable living person that is negative, contested, or drawn from one source deserves verification, and it deserves it before publication rather than after.
Quotations. The most frequently wrong category in published nonfiction, and the easiest to check. Misquotation happens through the ordinary process of taking notes, and quotes migrate: a line gets attributed to a more famous person each time it’s repeated, until the version everyone knows was never said by anyone. If you check nothing else, check every quotation against its original source.
Numbers. Statistics degrade as they are passed along. A figure from a 2009 study becomes “recent research shows” in a 2019 article and arrives in your manuscript with no date at all. Go to the study. Check what it measured, on whom, and when, and whether the sentence you have written is what it found.
What almost never causes trouble, despite the anxiety it generates: small background details in scene-setting, approximate descriptions of well-known events, and anything a reader would understand as impression rather than assertion. Spend your worry where the consequences are.
How do you check your own book?
Systematically and in passes, treating each category separately, which is how professionals do it and is far more effective than reading through hoping to notice things.
- Build a claims list first. Go through the manuscript once doing nothing but extracting checkable assertions into a spreadsheet, with the page number. Do not verify anything yet. Extracting and checking at once means you do both badly.
- Do one category at a time. All names in one pass. All dates in another. All numbers. All quotations. Category passes are far faster and far more accurate than page-by-page reading because you stay in one mental mode.
- Go to the primary source. Not to an article about the study, the study. Not to a quotation site, the book or the transcript. The secondary layer is where most errors are introduced.
- Mark each claim’s status. Verified, unverified, or softened. Anything still unverified at the end either gets cut, gets attributed (“according to X”), or gets rewritten as the impression it actually is.
- Check anything you are certain about. Certainty is where errors live. The facts you looked up are usually right; the ones you knew are the ones that turn out to be a version you absorbed in 1997.
Give this real time. On a well-sourced book, a thorough self-check runs to several weeks, which is a large ask of your time but still an order of magnitude cheaper than the professional version.
What is a source trail, and why should you build one?
A source trail is a document mapping every factual claim to where you got it, keyed to the page. It’s invisible to readers and it’s the most useful thing you can build, for three reasons that have nothing to do with scholarship.
It makes a paid check affordable. A fact-checker working without your sources has to find them, which is most of the cost. Hand over an annotated manuscript and the scope of that check can drop dramatically because you’ve converted a research job into a verification job.
It’s also your defense if a claim is challenged. Being able to show, quickly, where something came from changes a serious complaint into a correctable one. Not being able to is what turns a small error into a public problem.
And it makes the book updatable. When you revise for a second edition, or a figure changes, you can find the claim and its source in minutes instead of reconstructing a year of research.
Build it as you write, not afterwards. Reconstructing sources at the end is one of the more miserable experiences available to a nonfiction author, and it’s the reason many books quietly ship with claims whose origin nobody can now name.
Can AI fact-check a book?
It can help you build the claims list, but it can’t do the verification. That distinction needs stating precisely because this is the area where the gap between what the tools appear to do and what they reliably do is widest.
Genuinely useful: extracting checkable assertions from a chapter into a list, spotting internal contradictions across a long manuscript, flagging statistics that lack a date or a source, and drafting the questions you should be asking about a claim. That is real work and it saves real hours.
Not usable: the verification itself. A model will state confidently that a study says something, and the failure mode is not obvious ignorance but plausible, well-formed, specific wrongness, including citations to sources that don’t exist. On a task whose entire purpose is accuracy, a tool that is usually right isn’t fit for the job because you can’t tell which claims are the exceptions without checking them all, which is the work you were trying to avoid in the first place.
Use it to find what needs checking. Do the checking against sources you have opened yourself.
When should you hire a checker regardless of budget?
Four situations, and in these the cost is not the relevant number.
- You name living people in a negative or contested context. Investigation, family conflict in memoir, business disputes, anything where a person might reasonably object.
- Your book makes health, financial, or safety claims. Where a reader might act on what you wrote and be harmed.
- You are writing about a contested history. Where a hostile reader is actively looking for a mistake and one error will be used to dismiss the whole book.
- Your credibility is the product. If you are an expert publishing in your own field, a single checkable error costs more than the check would have.
In those cases, a partial check on the exposed material is far better than nothing, and it’s what most professionally published nonfiction actually gets. Ask for a scoped engagement: these chapters, these categories of claim, this many hours. A good fact-checker will tell you where your risk actually is, which is frequently not where you thought.
If your book needs research done rather than checked, that’s a separate service again, and I offer both as part of my research and fact-checking work.
Frequently asked questions
Doesn’t my publisher fact-check the book?
Almost certainly not. Magazines maintain fact-checking departments; book publishers, with very few exceptions, don’t; and standard contracts warrant accuracy on the author’s side. Ask your editor directly what verification the house performs, and expect the answer to be a legal read on the riskiest passages rather than a check.
Is fact-checking part of a copy edit?
No. A copy editor flags what looks wrong or contradicts something elsewhere in your manuscript, which is valuable but isn’t verification. Any editor offering a full fact check inside a copy editing fee is either misunderstanding the work or not planning to do it.
How long does a full fact check take?
On a well-sourced book, comparable to the writing of several chapters: weeks of full-time work. That’s why the fees look the way they do. It’s not a premium on a quick task, it’s a slow task priced honestly.
What if I can’t afford any of this?
Do the free tier properly and be transparent in the book. Answer every editorial query, self-check names, numbers, dates, and quotations in separate passes, attribute anything you couldn’t verify, and cut claims that won’t stand up. An author who has done that is in a far better position than one who paid for a partial check and skipped the queries.
Should I use a disclaimer?
A short note on method genuinely helps, and it’s standard practice in narrative nonfiction and memoir. Explain how you worked: which material came from documents, which from interviews, which from your own recollection, and where you have approximated. It buys you no legal protection and it substitutes for nothing, but a reader who knows how the book was assembled reads it more generously than one left to guess.
What about facts in fiction?
They matter more than novelists expect. Readers who know a subject are unforgiving about getting it wrong, and a single wrong detail in a specialist area breaks the spell for exactly the readers most likely to review you. Full checking is rarely justified; asking one knowledgeable reader to look at the relevant scenes usually is.
Who is responsible if an error gets published?
You are, in nearly every trade contract, and that’s the practical reason to take this seriously even where the risk feels remote. Read the warranties clause before you assume anyone else is carrying it.
Nonfiction with claims that need to hold up? Verification and research are separate from editing, they get priced separately, and the right scope depends entirely on where your exposure sits. Take a look at my research and fact-checking services, or write to me describing the book and what it asserts, and I’ll advise on your next steps.