How to Finish a Novel: A System for Writers Who Have Started Several

I have been there with four manuscripts and I finished the last five, and the difference was not talent, or a better idea, or more free time. It was that at some point I stopped relying on wanting to write and started relying on a system that did not care whether I wanted to. This is that system. It has five parts: a definition of finished, a budget, rules for the middle, a way to handle the crisis of faith, and a routine that gets the day's words out regardless.

Part 1: Define finished

Most abandoned novels were never going to be finished because "finished" was never defined. The book in the writer's head was perfect, and every draft chapter was a failure against it. That is not a standard; it is a trap.

Define the finish line as a first draft: a complete sequence of scenes from opening to ending, with every scene present in at least rough form, at approximately the target length. Not good. Complete. A first draft is allowed to have placeholder scenes ("[they argue about the letter here, she leaves]"), bad dialogue, and a subplot you already know you will cut. It is not allowed to have gaps.

Write the definition down, with a number:

FINISHED = every scene in the outline exists in prose, in order,
           from the opening image to the last line,
           at 70,000 words or more.
           Quality is not a criterion at this stage.

The number matters because it makes the end visible. Every day you can say how far you are. Vague goals produce vague progress.

Part 2: The budget

Novels are finished by arithmetic. An 80,000-word draft at 500 words a day, five days a week, takes 32 weeks. At 1,000 words a day it takes 16. At 250 words it takes 64, which is still a finished draft in fifteen months, which is faster than most people's third abandoned attempt.

The mistake is choosing the number from a good day. Choose it from a bad week. The right daily target is the one you can hit when you are tired, unenthusiastic and short on time, because those are most days. A target you hit nine days in ten builds a habit; a target you hit one day in three builds guilt.

Some schedules that work:

Track it. A spreadsheet with a date column and a word-count column is enough. The tracking is not for motivation; it is so that you cannot lie to yourself about whether you are writing. Two weeks of blank cells is information.

One rule about the budget: overshooting today does not reduce tomorrow's target. The habit is the daily act, not the total.

Part 3: Rules for the middle

The first third of a novel writes itself on novelty. The last third writes itself on momentum. The middle is where books die, and it needs rules.

Rule 1: No revising until the end. Reread yesterday's last page to find the thread. Fix a name if you notice it. Otherwise, forward only. Every chapter you revise before the ending exists will be revised again once you know the ending, and the time you spend on it is time the draft is not moving. Worse, revising the opening is the most comfortable way to avoid the middle, which is why so many drafts have a polished first act and nothing after it.

Rule 2: Notes instead of fixes. When you realise in chapter 14 that the sister needs to have been introduced in chapter 3, do not go back. Write a note in a revision file:

REVISION NOTES
- Ch 3: introduce Dana at the funeral. She and Mara argue about the house.
- Ch 9: Mara's key should be Dana's key (she stole it in ch 3 now).
- Continue as if both are done.

Then write chapter 14 as though chapter 3 already says what it will say. The draft will be inconsistent. First drafts are allowed to be inconsistent; the revision checklist exists for exactly this.

Rule 3: Placeholder scenes are legal. If a scene will not come, write a paragraph describing what happens in it, in square brackets, and move to the next. Some placeholders stay placeholders until revision; many turn out to be scenes you did not need.

Rule 4: When stuck, skip. If a scene will not come and you cannot even placeholder it, write the next scene you can see. Draft order does not have to match reading order. The outline holds the sequence; you can fill it in any order.

Rule 5: End each session mid-thread. Stop when you know what happens next, not when you have run out. Write the first line of the next scene before you close the file. Tomorrow starts in motion.

Part 4: The crisis of faith

Somewhere between 30 and 60 percent, you will decide the book is bad. Not this scene; the book. The premise is thin, the characters are cardboard, someone has done it better, and the sensible thing is to stop and start the other idea, the good one, that has been growing in the back of your mind since chapter 8.

This will happen. It happens to writers with twenty published novels. Treat it as weather, not as information. Some things that are true at this moment:

• You are comparing a rough draft of this book to a fantasy of the other book. The other book will have the same middle.
• The reason the draft looks bad is that you can now see it. In chapter 2 you could not. Seeing the problems is the skill; it does not mean the problems are fatal.
• Nobody can judge a novel from its middle, including you. The ending changes what the middle means.
• The new idea is attractive precisely because it has no draft yet.

The rule is procedural: no judgement until the draft is complete. Write the doubt in the revision notes ("Ch 19: I think the whole heist subplot is dead weight, check after the end") and keep going. You will be right about some of the doubts and wrong about most, and you will only find out which by finishing.

If the doubt is too loud to write through, lower the daily target to the minimum and keep the streak. A hundred words a day through a bad month is a finished draft three weeks later than planned. Zero words a day through a bad month is usually the end.

Part 5: The routine

The system runs on a routine that does not depend on inspiration. Mine, which is not special:

1. Same time, same place. The brain learns that this chair at this hour means words. It takes about three weeks.
2. Open the file before anything else. Not email, not the news, not the outline. The manuscript, at yesterday's last line.
3. Reread the last page. Only the last page. This finds the thread and is the one form of rereading allowed.
4. Write to the target. Badly if necessary. The point is the words, not their quality. The delete key is disabled by policy until the target is hit.
5. Log the count. Date, words, one line about what happened in the scene.
6. Write the first line of tomorrow. Then stop, even if you could continue. Especially if you could continue.
7. Once a week, look at the outline. Not to revise it, but to check that the draft is still heading somewhere. Adjust the outline, not the draft.

Twenty to forty minutes for 500 words, once the habit is set. Most people can find that. Most people who say they cannot find that are spending it on the thinking-about-writing that the routine is designed to replace.

Where AI fits, and where it does not

If you draft with AI assistance, the system does not change; the tool just fits into it. Where it helps:

• Step 3, finding the thread: a two-line summary of yesterday's scene from the model is a faster re-entry than rereading.
• Rule 4, when stuck: ask for three ways the scene could go, pick one, write it yourself.
• Placeholders: a scaffold for a scene you cannot see, to be rewritten in revision.
• The crisis of faith: ask the model to summarise the plot so far in five lines. It is oddly reassuring to see that there is a plot.

Where it does not help: it cannot want the book finished. The daily act is yours. Writers who expect the tool to carry the middle end up with 30,000 words of scaffold and the same crisis of faith, now about a draft that does not sound like them. Our AI novel workflow is explicit about which stages the tool can take and which it cannot.

After the finish line

When the draft is complete, three things:

1. Write the date and the count somewhere you will see it. You finished a novel. Most people who start do not.
2. Put it away for at least two weeks. Start the next book's premise, or read.
3. Then, and only then, open the revision notes and the checklist.

The second novel is easier, not because the writing is easier, but because you now know that the middle is survivable. That knowledge is the only thing that distinguishes writers who finish from writers who start.

Try it in SparkLore

SparkLore's project view is built around the daily act: a word-count target per session, a running total against your defined finish line, the outline beside the manuscript so you can skip to any scene, and a chapter summary generated when you close a scene so that tomorrow's re-entry takes ten seconds. The revision notes file lives in the project too. If you want to test the "when stuck, skip" move before creating a project, the free AI story continuation tool will take your last page and draft three directions for the next beat, which is usually enough to get the day's words moving. Create a project and write the finish-line definition first.