How to write a choose your own adventure, from napkin map to first playable branch
By John S.7 min read
Most choose your own adventure drafts die somewhere around node 12. The opening is great, every choice gets its own path, and a week later the writer owns forty half-scenes and no endings. This is the order I work in instead: premise, size, map, endings, then one playable branch you can hand to a friend tonight.
Start with a premise that already has a fork in it
A CYOA premise is a situation with pressure in it, not a world. "A kingdom with three moons" gives the reader nothing to decide. "You are a courier crossing a city under curfew, and the sealed letter in your bag has your own name on it" hands them a decision before the first scene ends: open it, deliver it or get rid of it.
My test is simple. If I cannot write the first choice in one sentence, the premise is not ready. A quick second opinion helps too: type the premise into the CYOA generator and look at the questions it asks back. They show you what the opening leaves open, which is exactly where a reader will start poking.
Settle two conventions while you are here. Most CYOA is written in the second person and the present tense ("You turn the envelope over"), because the reader is the one acting. And decide whether the reader plays a blank slate or a defined character with a past. Both work. Switching halfway does not.
Decide the size before you draw anything
Branching grows faster than anyone expects. If every node offers two choices and no path ever rejoins another, each level doubles the work. The table shows what a pure tree costs, counting every scene including the endings.
| Choices per node | Choice points deep | Scenes to write | Endings |
|---|---|---|---|
| 2 | 3 | 15 | 8 |
| 2 | 5 | 63 | 32 |
| 3 | 3 | 40 | 27 |
| 3 | 5 | 364 | 243 |
Nobody writes 364 scenes for a first project. Set a budget instead: 20 to 30 nodes, four to six endings, 150 to 400 words a node. That lands somewhere between 3,000 and 12,000 words, a story you can actually finish. The way to stay inside the budget is to let paths fold back together, and that starts on the map.
Map the nodes on paper first
I draw the map before I write a single scene, usually as numbered boxes with arrows between them. Each node gets three things: a number, a one-line summary and its exits. Here is the courier story at napkin stage.
1 Curfew bell. The letter has your name on it.
> open it (2) > deliver it sealed (3) > burn it (4)
2 The letter warns you: the recipient wants you gone.
> go anyway (5) > run for the docks (6)
3 The recipient is expecting you. So is a guard.
> bluff (5) > hand it over and leave (7)
4 Ash in the grate, then a knock at the door.
> answer it (5) > out the window (6)
5 BOTTLENECK: the meeting in the counting house
6 ENDING: you sail at dawn and never learn why
7 ENDING: the guard thanks you and locks the doorNode 1 splits three ways. Nodes 2, 3 and 4 each offer a road into node 5 and a road out of the story. Node 5 is a bottleneck: a scene every surviving reader reaches, carrying different knowledge depending on the route they took. Bottlenecks are the cheapest way to keep a story wide without letting it explode.
There are other shapes, from the open map to the gauntlet, and each suits a different kind of story. I break them down in branching story structure.
Write the endings before the middle
Endings are destinations, and you cannot draw roads without them. Before writing node 2, write one line for every ending you plan. If two of those lines say roughly the same thing, merge them.
Give the endings different flavours. A useful mix for a first story is one full success, one bittersweet, one strange and one or two short failures. Five versions of "you die" is not variety.
A strong opinion of mine: a failure has to be traceable. The reader should be able to look back and say "I trusted the guard, and that was the mistake." A death that comes from a coin-flip choice with no information feels like the book cheated, and a reader who feels cheated stops trusting the next fork.
What makes a choice worth making?
Three kinds of choice drain the life out of a CYOA. The fake choice, where both options lead to the same text. The obvious choice, where one option is clearly right. And the blind choice, where the reader has nothing to go on, like a left tunnel and a right tunnel with no detail at all.
The fix is to make each option say something about the person choosing. At node 1, opening the letter is curiosity, delivering it sealed is duty, burning it is self-protection. None is correct. Each one tells the story who the reader is, and later scenes can answer to that.
Keep it to two or three options per node; four starts to read like a menu. Word the options as actions, not outcomes: "Burn the letter" rather than "Stay safe". Making those choices echo later is its own craft, covered in choices that matter.
Write the first playable branch
Do not write the story in node order. Write one complete path first, so you find out early whether the voice, the length of each node and the pacing work.
- Write node 1 in full, ending on the choice.
- Pick the path you find most interesting and write it all the way to an ending. On the courier map that might be 1, 2, 6.
- Write the bottleneck scene next, with a placeholder line for each route that can reach it.
- Fill in the remaining paths one at a time, checking each against the map.
- Only then go back and polish the opening, because now you know what it has to set up.
Paper and numbered sections are enough for all of this. If you want the story clickable, Twine is a free, open-source tool that needs no code for a simple story and saves your work as a web page you can share. Whichever you use, keep the map open beside the draft.
Test it on someone who has never seen the map
Hand the branch to a friend and watch, without explaining anything. Note where they pause, which option they pick and whether they flip back to try the other one. A reader flipping back is a good sign. A reader picking at random is not.
Ask two questions afterwards: did any choice feel fake, and did any ending feel unfair? Then count. If three testers in a row skip the same option at node 3, the wording of that option is the problem. Sharpen it before you add a single new node.
What to keep on a node sheet
Once a story passes about 15 nodes, the napkin stops being enough. I keep one row per node, and it saves me every time I come back to a draft after a week away.
- The node number and a one-line summary
- Where the reader is and who is with them
- What the reader knows by now, depending on the route
- Any item or fact the scene depends on
- Every exit, with the node it leads to
Questions
How long should a choose your own adventure story be?
For a first project, aim for 20 to 30 nodes of 150 to 400 words each. That gives a story of roughly 3,000 to 12,000 words that one reader can finish in a sitting and replay a few times.
Should a choose your own adventure be written in second person?
Second person present tense is the convention, and it suits the form because the reader is the one choosing. First person also works when the lead is a defined character with a voice of their own.
How many endings does a choose your own adventure need?
Four to six is plenty for a first story. Make them different in kind: one full success, one bittersweet, one strange and one or two short failures the reader can trace back to a choice.