Two feature screenplays arrive at 105 pages. One reads in a sitting. The other loses its reader somewhere around page sixty, and the note comes back as it drags in the middle.
Nothing about the page count separates them. What separates them is distribution — where the scenes are, how long they run, how much of each stretch is action rather than dialogue, and how much the reader is being asked to hold at once. That is pacing, and unlike most craft vocabulary, quite a lot of it is measurable directly from the page.
The note is almost never about length
The reflex when a reader says a draft drags is to cut. It is the wrong reflex, because trimming evenly across a screenplay preserves the exact shape that produced the note. A hundred-and-five-page draft with a slack stretch becomes a ninety-eight-page draft with the same slack stretch, slightly compressed.
The useful question is narrower: which stretch, and what is it made of? Three things are countable in any stretch — how often scenes change, how much scene lengths vary, and how the action share compares to what published practice runs there. Any one of them alone is a style. All three at once is a distinctive structural signature, and one you can locate in your own draft in about a minute. Whether it is the stretch your reader meant is a judgement the two of you make, not one the corpus can settle: it holds screenplays, not reader notes, so it can say what published practice looks like and nothing about how anyone reacted. And the count cannot tell you why either — a page of exposition and a confrontation that is meant to run long measure the same.
Published screenplays do not pace evenly
We measured this rather than asserting it. Across a corpus of published feature screenplays, normalising each one to its own length and cutting it into ten equal stretches, the balance of action to dialogue moves in a consistent arc: heaviest on action at the open, dipping through the middle, and climbing again at the end.
Action share is highest at the opening and ending and lower through the middle. That is not a rule anyone wrote down — it is the shape the corpus reports. And it means the flat mental model behind most structure advice, where every part of a screenplay is made of the same material and only the beats change, does not match practice.
The screenplay pacing desk has the full curves, with interquartile bands so you can see how much variation sits behind each median.
Scene length is a distribution, not an average
The second measurable thing is rhythm. Ask how long a scene should be and you will get an average, which is the least useful summary available — because the whole point is that scenes are not the same length.
The number that carries information is concentration: how much of a screenplay its longest tenth of scenes carries. Across the corpus that runs from roughly a third at the low end to nearly half at the high end. Both are published practice. Both make a completely different read.
Three drafts, one word count
Each ribbon holds the same narrative words, cut into 40 scenes. The only thing that changes is how much of the draft its longest tenth of scenes carries — the range published across 2,077 screenplays. A word count cannot tell these three apart. A reader can, by the second act.
Even pulse
Longest tenth carries 32% · 10th percentile
The longest tenth carries a relatively small share of the draft, so scene lengths are comparatively even.
Typical
Longest tenth carries 38% · Median
The middle-of-corpus shape: mostly short and medium units with a handful of longer scenes.
Set-piece driven
Longest tenth carries 46% · 90th percentile
The longest tenth carries close to half the draft, concentrating much more narrative weight in a few scenes.
The three ribbons above hold the same narrative words. The only difference is concentration. A word count cannot tell them apart; a reader can, by the second act.
An even ribbon means scene length is tightly concentrated. That is worth knowing before a rewrite because it gives you a concrete question to take back to the pages: does the draft want a steady cadence, or do differences in tension, internal turns, or action-to-dialogue balance call for more contrast? The measurement can reveal the uniformity; the rewrite choice still belongs to the writer.
The cast keeps arriving
The third measurable thing surprises most writers. In published features, only about a fifth of the speaking cast has its first speaking scene inside the first tenth, and barely two-thirds by the midpoint. New voices keep entering long after the setup is over.
Drafts often do the opposite. A first act that introduces everyone is trying to be efficient, and the usual argument against it is that a cold reader is being asked to carry every name at once before any has earned the memory. That is a craft argument, not something the curve measures — what the curve shows is only that the arrival is earlier than published practice. If your cast-arrival curve sits well above the corpus band early, that is a concrete, cheap thing to change — a speaking part in the opening tenth that could arrive later, or stay silent.
What measurement cannot do
It cannot name a beat. Where the inciting incident sits, whether the midpoint actually turns, whether a twelve-page scene earns its length — those are readings a person makes from the text. No word-position measurement can make them, and a tool that claims to detect them is guessing with a confident interface.
This matters because the temptation runs the other way. It is easy to build something that reports "your inciting incident is late" from a page number, and it would be wrong most of the time. What a measurement can tell you is where your words, scenes, and voices actually fall — a map of where to look. Deciding what it means is the work, and it stays the work.
The pacing map draws your draft's curves over the corpus band, in your browser, without uploading the screenplay. It reports position, states when your draft sits outside the population it is being compared to, and never produces a verdict.
Where this sits next to page count
Page count is a reader signal — an early estimate of runtime and discipline, and worth checking against the length bands before you submit. Pacing is the other half of the same question. Length tells a reader how much of their evening you are asking for. Pacing shows how that length is distributed across the draft.