Bringing a Scrivener project in, and what it shows you

A Scrivener project is already a deck of cards. That is not a figure of speech: the binder is a tree of documents, each document has a synopsis that sits on the front of an index card in the corkboard, each has a label, a status, keywords, notes, and a body. Writers who use it have been working the card method for years without a table, and the ones who leave it usually leave for one of two reasons — the corkboard shows the cards but cannot check them, and the manuscript is a compile of the binder and nothing more.

This guide is about what happens when that deck moves to software built to check cards as well as show them: what comes across, where each thing lands, and — the useful part — what the first hour on the other side shows you about the book that the corkboard could not.

What a binder knows

Before moving anything, it is worth seeing what you already have, because most Scrivener users have more structure than they think.

Open the binder and read it as a table of contents: that is your order. Open the corkboard on any folder: the synopses are scene cards, or the fronts of them — each is the one-line answer to what happens here? The labels are usually one of two things, point of view or plot thread, and either is a column of the plot grid in guide 19. The keywords are tags. The document notes are the things you told yourself while writing the scene. The Research folder is the sources.

And the body is the prose, with the footnotes and the inline annotations and the comments you left yourself in the margin.

All of that is on disk, in a folder, in plain formats. None of it is locked in. The question is only where each piece goes.

Where each piece lands

A project is a folder ending in .scriv. It can be brought in as the folder itself, or as a zip of it. What comes across, and where:

Nothing is split, because Scrivener already chose the split — the document is the unit, and the document becomes the card. Both Scrivener 3 and Scrivener 2 projects are read; they keep the files in different places, and both places are looked in.

Before anything is created, the count: so many documents and folders as so many cards; so many with notes; so many research files attached; so many in the Trash and left there. And under it, the check that matters: every character is accounted for — putting the cards back together gives the document exactly as it came in, to the character. If it would not, nothing is created, and the first place the two differ is shown, because losing a word is the one thing an import is not allowed to do, and a promise you cannot check is a claim.

The first hour: names, dates, quotations

Right after the cards arrive, a short pass over them: the names the text repeats, the dates it mentions, the passages in quotation marks. These twenty-three capitalized names appear five or more times. Which are characters? Tick the ones that are, and each gets a character card, linked from every scene it appears in. Untick the ones that are places, or the town, or the name of the boat, and give the places their own cards if you want them.

This is the step the corkboard never had. A Scrivener label can say Ada's POV; it cannot say who is in the room. After this pass, every scene knows its cast, and the plot grid and the ledger of guides 19 and 20 have something to draw.

What the first look shows

Here is where it gets interesting, because the cards that came across are complete as prose and nearly empty as scene cards. Each has a title, a synopsis, and a body. None has an answer to what does she want here?, what's in her way?, how does it end?, or what's different for her by the end? — because Scrivener never asked.

Run the pass over the order, and read what it says. It will say the same thing forty times, in different words: either nobody wants anything in this scene, or nothing is in their way — one of the two is blank; nothing changes for anyone in this scene — at least, nothing you've written down. This is not a judgment of the scenes. It is the software noticing that the cards have no answers yet, which is true, and it is a list of the forty things to do next.

So do them. Take the scenes in order, and for each, read the synopsis and the first and last paragraphs of the body, and answer the five questions from guide 17 in a line each. Ninety seconds a scene; an hour for forty. As you go, two things happen.

The scenes that cannot answer show themselves. You reach a scene and cannot say what she wants in it. Read the body again. Still nothing — it is atmosphere, or it is the writer working out what happens next, and it was carried through three drafts because it was well written. It is the scene the outline hid. Mark it, and move on; the cut pile is a decision for later.

The order starts to argue. Answering what's different for her by the end? for scene 11 — from trusting Ned to suspecting him — and then for scene 12 — from suspecting Ned to trusting him again — you notice that scene 12 was written before you moved scene 11 forward, and the two have swapped a change. In Scrivener you would have found this on the read-through. Here, it is two lines on two adjacent cards.

Then the threads

With the scene cards answered, name the threads, as in guide 19, and lay the beats. If your labels were plot threads, most of the laying is done: the label came across as a tag, and each tagged scene is a candidate cell. If your labels were points of view, they are already on the scene cards as whose eyes, and the threads are still yours to name.

The grid will show the gaps — the locket's twenty-three silent scenes — and the corkboard could not have, because the corkboard showed one folder at a time and never checked anything.

Two orders, one deck

Scrivener projects often have two binder roots: the manuscript, and a folder of scenes cut or not yet placed. Each root becomes its own order. The second one is a holding deck — a place scenes wait, in an order that means nothing, until they are placed in the first. When you compile, only the manuscript order compiles. Nothing from the second is lost, and nothing from it leaks in.

If you kept old drafts of chapters in the binder as siblings, they came across too. Retire the ones you are done with — a retired card keeps its text, goes to the Cut, and can be restored from it — rather than deleting them; the version you will want in March is always the one you deleted in January.

Doing this on the table

There is no paper version of an import. But there is a paper version of the first hour: print the synopses, one per card, and answer the five questions on each. Writers who do this with a Scrivener book they thought was finished tend to find the same four or five scenes the pass finds. The software's advantage is that it finds them every time the order changes, and says so in a sentence.

Doing this in software

Everything in this guide is what the software does, so the section is short. Two things to know that the table cannot offer.

The check runs from now on. Once the scene cards have answers, the pass reads them after every change — a scene moved, a field filled — and reports in plain sentences: this scene doesn't move any of your threads forward. Is it earning its place?; the thread about the locket goes quiet for twenty-three scenes; Ned is in the room for four scenes without doing anything the cards record. Each names a scene. None says whether it is good.

The manuscript compiles from the order. Move a scene; its prose moves. Fork the order to try the book with scene 17 after 23, compile both, and read them side by side. The binder was an outline you compiled; this is an outline that compiles itself, and the reverse outline — the thing every editor tells you to make and nobody makes — is what the deck is at every moment, because the cards are the outline and the prose is on the back of them.