Beat Specs Before Prose
Every scene in the novel has an emotional job. I map that job before I write a word of prose.
I had written twelve pages of a chapter and none of it was working. The dialogue was fine. The descriptions landed. The sentences read well in isolation. But the chapter felt flat, and I couldn’t figure out why until I stopped reading forward and started asking a different question: what is the reader supposed to feel at the end of this scene, and does everything before it build toward that feeling?
The answer was no. The prose was serving the plot. It was moving characters through events in the right sequence. But the emotional trajectory was incidental. Nobody had specified what the scene was supposed to do to the reader.
This is the problem that beat specs solve.
I’m writing a novel called New City. I’ve never written a novel before. I have years of design engineering and brand systems and governance protocols, and exactly zero published fiction. What I do have is a methodology for specifying what a system should produce before building the system. And I decided to find out whether that transfers.
A beat spec is what it sounds like. Before I write prose for any scene, I write a spec that defines the emotional beat. What should the reader feel entering this scene? What should shift during it? Where should they land emotionally by the end? The beat spec is the accommodation design for the reader’s experience. It defines the job the scene needs to do, separate from the plot mechanics.
The distinction matters. Plot is what happens. Beat is what it does to you. A character walks into a room and discovers something. That’s plot. The reader feels a slow tightening of dread across the three pages leading up to that discovery, then a specific kind of relief when the thing discovered is not what they feared. That’s beat. Same plot event. Completely different reader experience depending on whether you specified the beat or let it happen accidentally.
I map each scene across three dimensions. Story: what happens, the literal events. Narrative: how it’s told, the POV, the pacing, the information the reader has versus what the characters have. Reader: what the person holding the book feels, moment by moment. The beat spec lives in that third dimension. The prose serves all three, but when they conflict, the reader dimension wins.
This came from watching what worked and what didn’t in early drafts. The chapters that worked had a clear emotional trajectory even when I hadn’t consciously planned one. Something in the writing had intuitively landed on the right beat. The chapters that felt flat had good prose and correct plot and no emotional architecture. I was decorating a room with no floor plan.
The beat spec became the floor plan.
Here’s what one looks like in practice. I’m working on a scene where two characters meet for the first time. The plot requirement is simple: they need to establish a relationship that will carry weight later. The beat spec says something different. It says: the reader enters this scene mildly curious, maybe slightly impatient because the previous chapter ended on tension. The first two pages need to reward that impatience by giving the new character an immediately distinctive voice. By the midpoint, the reader should feel a subtle pull, almost recognition, something about this character that echoes someone from earlier in the book. The scene closes with a specific line that reframes the previous chapter’s tension. The reader leaves feeling like the story just got bigger.
None of that is plot. All of it is architecture.
The framework is fractal. The novel has three acts. Each act has four parts. Each part has four beats. The beats are the atomic unit, the smallest thing I specify before writing. And the spec at each level defines the same three dimensions: story, narrative, reader. The act-level spec says what the reader should feel across fifty pages. The beat-level spec says what they should feel across five.
I built this the way I build anything. I identified what was failing (flat chapters with good prose), I diagnosed the root cause (no emotional specification), and I built a structure that addresses the gap without dictating the creative output. The beat spec tells me the emotional job. It doesn’t tell me which words to use. The prose still has to do the work. But it knows what work it’s doing.
The closest analogy I have is album sequencing. A great album isn’t just great songs in a playlist. The sequence creates a trajectory. Track three exists in the context of tracks one and two. The energy of track seven is designed to follow the valley of track six. DJs know this instinctively. The next record isn’t chosen because it’s good. It’s chosen because of what the room needs at that exact moment in the set.
Prose works the same way. A sentence that’s beautiful in isolation can be wrong for the beat. A paragraph that feels rough on its own can be exactly right because it’s creating the tension the next paragraph needs to release. The beat spec gives me permission to write an ugly paragraph when ugly is what the reader needs to feel. Without it, I would polish every paragraph to the same smoothness and end up with the literary equivalent of a playlist on shuffle.
I don’t know yet if this produces a good novel. I know it produces better chapters than I was writing without it. The twelve pages I rewrote after building the beat spec for that scene came out different. The same events, the same characters, mostly the same dialogue. But the chapter did something to me when I read it back. It had direction. The prose was serving a job I’d specified, and the reader (me, in this case, cold-reading my own work the next morning) could feel it.
The methodology transfers. I’m still finding out how far.