Most Family Histories Do Not Start With Writing

Almost every guide to writing a family history opens the same way. Clear a shelf. Buy a notebook. Begin at the beginning. It is fine advice for a book that does not exist yet, and it is the wrong advice for most people who sit down to write one.

So we measured what people actually bring us. BookFoundry holds 1,023 book projects. 303 of them, close to a third, did not begin with writing at all. They began with a document that already existed: a transcript, a folder of letters typed up years ago, a half-finished draft, a genealogy report, an interview somebody recorded in 2019 and has not opened since.

That changes what the first week of the work should be. It also produces the one number in this article we would ask you to remember.

A note on where these figures come from, because it limits them. They describe projects made on BookFoundry, which is not a random sample of everyone writing a family history: people who go looking for software are already a particular kind of author. Only 28 of the 1,023 projects state a family or life-story subject outright, far too few to say anything about family books specifically, so every figure below is drawn from the whole corpus and labelled that way. They are honest about what they measure and no wider than that.

The number: unstructured material finishes at 5 percent

Of the 291 imported documents where we can count chapters, 113 arrived as one undivided block. No sections, no breaks, a single continuous run of text. That is 39 percent of everything people bring.

Then we looked at which of those documents ever produced a finished file.

A document that arrives with some structure already in it finishes at roughly four times the rate of one that arrives as a block.

We want to be careful about what that does and does not prove. It is a correlation, not a mechanism. A document that already has chapters may simply belong to somebody who was further along in every respect, better organised, clearer about the book, more likely to finish whatever they touched. The chapters may be a symptom of that rather than a cause.

But the size of the gap is hard to ignore, and it matches what we see in support conversations. The single block is where family history projects go quiet.

What a block actually is

It is worth describing, because if you have one you will recognise it immediately.

Somebody kept adding to the same file. Every conversation with an aunt went to the bottom of it. A paragraph about the 1970s sits directly above a description of a house nobody in the family has lived in since 1991. There are three accounts of the same argument, written years apart, and they disagree. The file is 40,000 words and opening it produces a specific kind of dread, which is why it has not been opened in a long time.

The instinct at this point is to start rewriting from the top. That is almost always the wrong move, because rewriting is how you avoid deciding, and the deciding is the actual work.

What a block needs first is not better sentences. It is three decisions:

Whose book is it. A family history that tries to give every branch equal weight becomes a directory. Pick a spine, usually one person or one household, and let everyone else appear in relation to it.

What order. Chronological is the obvious choice and often the worst one, because a life is not interesting in the order it happened. By person, by place, or by four or five decisions that changed everything downstream will usually produce a book somebody reads to the end.

Which version. Where your material contradicts itself, and it will, you are not obliged to resolve it. "My mother remembered it this way and my uncle remembered it differently" is often the truest sentence available, and it is more interesting than either version alone.

Only after those three does anything get written.

The material comes in two sizes and almost nothing between

Of the 287 imported documents where we can measure a length, the distribution is not a bell curve. It is two clusters with a valley in the middle.

Thirty-four percent are under 1,000 words. A few pages: one interview, one story written down after a funeral, a first attempt at a timeline.

Thirty percent are 10,000 words or more. Most of a short book, already on a hard drive, usually accumulated over years.

The median is 2,987 words, and it is the least useful number here because very few documents are actually that size. The tenth percentile is 97 words. The ninetieth is 20,030. The largest single document anyone has brought us was 253,676 words.

Which cluster you are in decides your next month. Under 1,000 words, your problem is that you do not have enough yet, and the work is collection: more interviews, more scanning, more asking while people are still here to ask. Over 10,000, you have enough and the work is the three decisions above.

The mistake we see most often is people in the second cluster behaving as though they are in the first, gathering more material to avoid organising the material they have.

A quarter of these books are not in English

Of the 712 projects that record a language, 408 are English. 188, or 26 percent, are Arabic. German, French, Portuguese, Russian, Spanish and Chinese together account for 81.

We did not expect that distribution, and it is worth stating plainly because most writing advice on the internet quietly assumes an English-speaking author writing for English-speaking readers. A family history is the least likely book to fit that assumption. It is written for one specific family, and it belongs in the language that family actually talks in. If the interviews happened in Arabic, the book that comes out of them should be in Arabic. Translating it into English first, on the theory that a book ought to be in English, loses the thing you were trying to keep.

Where a book will be read by two generations who do not share a first language, two versions of the same book works better than one compromise between them.

What people do with it at the end

Of the 1,023 projects, 277 produced a finished file. What they chose says something about what a family history is for.

• 223 exported a PDF. This is the one you print and hand to somebody.
• 109 exported a DOCX, the working format, the one you send to a cousin who finds four mistakes.
• 88 exported an EPUB for Kindle. Apple, Google, Kobo and Nook trail well behind, at 37, 33, 24 and 22.

Most of these books are made to be given, not sold. That is worth knowing before you spend three weeks on keyword research for a book whose entire readership will fit at one table at a wedding. It also means the deadline that matters is usually a birthday, a reunion or a funeral, not a launch.

One more figure, for the people it applies to: 144 projects are part of a series. Splitting by branch or by generation, rather than forcing four generations into one volume, is a more common answer than we expected.

What we would tell you to do first

Not seven steps. Two.

Gather everything that already exists into one place. Not organised, just gathered. Most people find at this point that they have more than they thought, and that the real gap is one or two specific conversations nobody has had yet. Those conversations are the urgent part, because the material in somebody's memory has a deadline that a file on a hard drive does not.

Then impose a spine before you write a sentence. Whose book, what order, which version. If you do nothing else from this article, do not start at the top of the block and begin polishing.

The writing itself is the part that goes fastest, and it is the least of your problems.

Figures in this article come from 1,023 book projects created on the platform, measured in August 2026. They describe our own users rather than authors in general. No individual project, title or text was read: every number here is an aggregate.