Part one argued that the comparison grid scores the wrong axis. Part two priced what each restart takes from the book. This part is what I do instead, and it is short, because most of the work is in refusing to start with the feature list.
The test: open one thing and try to write
Before I look at any feature, I run one test on a candidate, and it takes a single session rather than a trial month.
Close everything. Open the one app. Now try to get from nothing to writing the next real sentence of your book, not a test paragraph but the actual next sentence, without opening a second thing.
You will need what you always need: where you left off, what this chapter is for, the name you gave the neighbour, whether the argument in chapter six has happened yet. The test is whether every one of those is reachable from inside the app you have open.
Time it, and note where you had to leave. Both numbers matter, but the second one matters more, because it names the thing that is not in there. Everybody's list is different: for one writer it is the research, for another it is the chapter order, for a third it is the last version they trusted.
The test only works on your real book, mid-project, on a day you actually have. A blank document in a new app always feels wonderful. That feeling is what has been selling us new apps for years, and it is not evidence about the twentieth chapter.
The four questions worth scoring
If the test surfaces a gap, these are the questions I use to decide whether a candidate closes it. They are about scope, which is stable, rather than preference, which is not.
1. Can the whole book live here? Not just the chapters, but everything the chapters lean on: what you looked up, what you decided and have to keep true, the drafts you might want back, and the people and places you invented. If any of that has to live somewhere else, you have not consolidated. You have added.
2. Is there one current version, always? The question is not whether the app has version history, because most do. It is whether the app makes "which one is real" a question you can stop asking. If you would still keep a safety copy elsewhere, notice that: the safety copy is the scatter starting again, and it starts for a reason worth taking seriously.
3. Can you get the book out without rebuilding it? A place your book lives has to be a place your book can leave. Check the export formats before you move in, not after. If getting a submission-ready manuscript out means reformatting it by hand in another program, the book does not really live there; it is being held there.
4. Does the help know the book, or only the page? Most writing assistance operates on the paragraph in front of it. That is genuinely useful for a sentence and useless for continuity, which is where long books actually break. The question is whether anything in the app has read the rest of your manuscript.
| What the grid usually scores | What decides whether you finish |
|---|---|
| Distraction-free mode, themes, typography | Whether the session can start without reassembly |
| Word-count goals and streaks | Whether continuity is cheap enough that you check it |
| Number of export formats | Whether the book can leave without being rebuilt |
| Whether it has an outliner | Whether the outline and the draft are the same object |
How to consolidate without losing a month
The reason people stay scattered is not that they disagree with any of this. It is that moving a part-written book looks like a fortnight of admin, and a fortnight of admin is a fortnight of not writing. So the move has to be small enough to survive contact with a normal week.
Move the spine, not the archive. The spine is the current manuscript, the chapter order, and the facts you check most often. Everything else, meaning old drafts, research you have not opened in six months and the cut scenes, stays exactly where it is. You are not migrating a library; you are giving the live book an address.
Do one chapter, then write a session in it. Move a single chapter, then have a real writing session there before moving anything else. One session tells you more than any comparison grid, and if the answer is no, you have lost an evening rather than a fortnight.
Bring the facts as you need them. Rather than transcribing every note up front, move each fact the first time the draft asks for it. Within two weeks the things you actually use have migrated themselves, and the things you never use have identified themselves as things you never use.
Set a date to close the old places. Not to delete them, just to stop writing in them. The scatter does not end when the new place is set up; it ends when the old ones become read-only.
- Run the one-thing test on your current setup and write down where you had to leave.
- Score your setup on the four questions. Be honest about question 2.
- Move one chapter into a single place and write one real session there.
- Pick the date the old places go read-only.
My own bias, declared
I should be direct about the fact that I work on a writing app, and about where it does and does not answer the four questions.
Write in a Click is built as one home for the entire book rather than as an editor with attachments. Planning, drafting and editing happen in the same workspace; Story Hub holds the world and the people in it beside the draft rather than in a separate file; version history means there is one current version instead of a folder of candidates. You can get the book out as PDF, EPUB, DOCX, RTF or plain text, so leaving does not mean rebuilding. And Robin, your co-author, has read the whole manuscript rather than the paragraph on screen, which is the difference question 4 is really asking about, because continuity is not a sentence-level problem. It is on the web, iPhone, iPad and Android, with offline drafting on mobile, so the phone is part of the same book rather than another place the book is.
Those are answers to questions 1 through 4, and you should check them the same way you would check anyone's: run the one-thing test on your own book and see where you have to leave.
Two things this will not do
Both worth saying plainly.
Consolidating is real work, and it is not free. If your book is nearly done, in final revisions with the end in sight, do not move it. Finish it where it is. Every cost in part two is charged over the remaining length of the project, and a project with two months left will not repay a migration.
And a single home does not write the book. It removes the reassembly, makes checking cheap, and keeps one version current. What it buys you is that your sessions are spent on the book rather than on finding the book. That is a real gain and a bounded one. Better to claim the bounded version than a larger one you will test against your own experience and find wanting.
The short version
Stop scoring apps on features you will stop noticing in a fortnight. Ask instead whether the whole book can live there, whether there is one current version, whether it can leave, and whether the help has read more than the current page. Then run one honest session before you commit to anything.
The search stops restarting when you change what you are searching for.
