Every book project starts with the same set of decisions, and on most projects nobody makes them on purpose.
Nobody sits down and chooses. The project begins, the interviews start, chapters accumulate, and by the time anybody notices that a decision was required the default has already taken it. Usually the default is whatever felt natural in week one, which is not the same as what the book needed.
The expensive part is that these decisions are structural. Changing one in month six means unpicking everything built on top of it, which is why so many projects reach a point where something is clearly wrong and nobody wants to say so, because the cost of naming it has become enormous.
Made deliberately at the start, each one takes an afternoon. Made by accident, each one costs a rewrite or gets lived with.
None of them are about writing. They are about what the thing is, who it is for, and what happens to it afterwards, which is why writers are frequently not the right people to raise them and why they get skipped.
Here is the list, in the order they have to be answered.
Answer these before you agree a price, because every one of them changes what the project costs and how long it takes.
What is the book for?
Not what it is about. What it has to do.
Sell copies. Reinforce a brand. Open speaking. Give clients something to take away. Establish a position in a field. Leave a record. Those are different objectives and they produce different books, and most people arrive with two or three of them held at the same time without noticing they conflict.
Pick one. Two if they support each other. Three is how a book ends up serving none of them.
This is the decision everything else depends on, and it is the one most likely to be left vague because vagueness feels like keeping options open.
Who is the reader?
One person, described specifically enough that somebody could disagree with the description.
A role, at a scale, facing a problem, at a moment. Operations leaders at companies between one and two hundred people who inherited a system nobody documented. That is a reader. Senior executives is a category.
This decision determines the vocabulary, the examples, the level of assumed knowledge and the length. Change it in month four and most of what exists has to be rewritten, because you made every one of those choices for somebody else.
What is the claim?
The one sentence a competent peer might dispute.
A book without one becomes a survey of what is already known, competently arranged. Those get written constantly and nobody remembers them, because there was nothing to remember.
The claim does not have to be startling. It has to be yours, supported by things you watched happen, and specific enough that somebody could hold the opposite view.
If this cannot be stated before writing begins, that is worth knowing in week one instead of month five.
How does a book get in front of the right readers?
This is the decision nobody makes and it determines whether any of the rest mattered.
Who receives a copy. How they hear about it. What you will do in the six months after publication and who will do it, given that the writer normally leaves at delivery.
It belongs at the start because the answer changes the book. A book destined for fifty named people is a different object from one going into retail, and both are legitimate.
Deciding it afterwards produces the most common failure in this business, which is a good book that nobody arranged to distribute.
What happens if book decisions are made late?
The project develops a soft spot that everybody works around.
A book with an unsettled reader acquires chapters aimed at different people, and each one is fine alone. The problem only appears when somebody reads them consecutively and cannot tell who the book thinks they are.
A book with an unstated claim accumulates well-written sections that do not build. Every chapter is defensible and the whole thing goes nowhere, which readers experience as losing interest around chapter four for reasons they cannot name.
Neither failure is visible in a single chapter, which is why they survive months of review by people looking at drafts one at a time.
Can any of them be changed mid-project?
One of them cheaply, and the rest expensively.
Distribution can be decided or revised at any point, because it affects what happens after instead of what gets written. Deciding in month eight to send fifty named copies costs nothing except the copies.
The reader and the claim are the expensive ones. Change either and most of your existing material belongs to a different book, and the honest response is usually to restart the outline instead of patch it.
Which is why those two get the most time at the start, and why a writer who moves straight to interviews without settling them is storing up a conversation nobody will want to have.
What is off limits?
Confidentiality, people, and the things you will not say.
Every experienced person has material they cannot use. Work under agreement. Clients who would be identifiable. A period involving somebody still in the industry. A story that would embarrass a family member.
Naming those at the start saves a specific and painful kind of waste, which is a strong chapter written from material that has to come out in the legal review.
It also frequently reveals that the constraint is smaller than assumed, because most examples survive generalisation.
Who decides when there is a disagreement?
Somebody has to, and it should be settled before the first one happens.
The client defines the project and the writer normally leads it, because the writer knows what breaks a book. That arrangement works and it has to be agreed instead of discovered.
Two people running a project produces stalemate, and stalemate in month seven with a manuscript half built is how books stop.
The version to avoid is nobody deciding, where every disagreement gets deferred and the book drifts toward whatever nobody objected to.
