How to Read a Coverage Report: A Worked Example

Published by StoryNotes. The worked example below uses our own public sample reports, which means you can open them in another tab and check every claim in this article against the actual page.

The report lands. Twenty-odd pages, six sections, a letter grade on each one, and somewhere in the middle a sentence about your protagonist that you will still be thinking about on Thursday.

Most writers then do the same thing, which is read it start to finish and begin rewriting from whichever note hurt most. That is an understandable response and a bad method, because a coverage report is not written in priority order. It is written in the order the script gets analyzed, which has nothing to do with the order the problems should be fixed. The note that stings and the note that matters are not the same note often enough that rewriting from pain is close to rewriting at random.

TL;DR: Read every concern line in the report first, in one pass, before reading any section in full. Cluster them and look for the repeat — a problem named in three sections is one problem, and that is your rewrite. Use grades to decide where to look, never as a verdict. Open the cited page before you agree or disagree with a note. Then decide, in writing, what you are not going to fix, and rerun the analysis after the rewrite to confirm the note is actually gone.

Structure varies by provider, but a full report is generally a six-stage document, running from genre and tone through character and plot structure to theme and scene-level craft, with each section carrying its own grade, a set of flagged strengths and concerns, an analysis body, and two or three suggested rewrites.

Four things in there behave differently and should be read differently:

The concern lines are the part almost nobody reads as a set, and they are the part that repays it most. Everything below is a method for doing that.

One concrete fix: Before you read a word of the analysis body, scroll the whole report and copy every concern line into a single flat list. That list, not the report, is the document you are going to work from.

Because the sections are not independent, and the report does not tell you which ones are downstream of the others.

A screenplay is a system. When a protagonist's goal goes soft in act two, that single fact shows up as a pacing note, a character note, and a dialogue note — three sections, three vocabularies, one cause. Read in written order, those arrive as three separate problems and you fix three separate things, none of which is the thing. Read as a set, the repeat is visible immediately.

This is also why the volume of a report is misleading. Eleven flagged concerns can be four real problems, and a writer who counts notes rather than clustering them ends up believing the draft is in worse shape than it is.

One concrete fix: Never estimate the health of a draft by the number of notes. Count clusters instead, after you have grouped them.

Here is what that looks like on a report you can actually open.

The public sample report for Charade — the 1963 Stanley Donen thriller, readable in full here — grades out well. Six sections, five of them at A- or A, one at B. On a top-to-bottom read, it looks like a strong draft with one soft section, and the obvious move is to go fix the B.

Now read only the flagged concerns, across the whole report, in one pass. There are three:

• From the genre and tone section: "False Bartholomew's CIA identity obligation goes unearned — reveal arrives late."
• From the premise section: "Carson Dyle's backstory revealed late; antagonist pressure arrives slowly."
• From the scene and dialogue section: "Bartholomew exposition scenes run long; value-shift arrives late."

Three sections. Three different craft vocabularies. Three notes that, read in their own sections, sound like three unrelated problems — a genre obligation, an antagonist build, a scene-level pacing issue.

Read as a set, they are one sentence: information in this script arrives later than it needs to.

One concrete fix: Do the pass with the analysis bodies collapsed, or with the page scrolled fast enough that you cannot read them. The reasoning is persuasive and it will pull you into the first section you stop on.

A note that appears once is an observation. A note that appears in three sections is a structural fact about the draft, and it is a structural fact whether or not any individual phrasing of it is right.

That last part is the useful bit. You can disagree with the genre and tone note about the CIA reveal — reasonable people would — and the cluster survives, because two other sections arrived at the same shape independently while looking at different things. Convergence across sections is worth more than the confidence of any single note, for the same reason a symptom reported by three instruments is worth more than a loud reading on one.

The practical test: after you have your flat list, try to write one sentence that would make more than one line on it go away. If such a sentence exists, that sentence is your rewrite. If you can only write one that addresses a single line, you have page-level fixes rather than a structural one, which is a genuinely good outcome and a much shorter draft.

One concrete fix: Write the cluster sentence at the top of a blank page before you open the script. If you cannot write it in one sentence, you have not found the cluster yet.

Now go back to Charade's grades, because they demonstrate the trap precisely.

The lowest-graded section is genre and tone, at B. Every other section is A- or A. A writer sorting by grade goes straight to the B — and the B section contributes exactly one line to the cluster, no more than the two A- sections do. Sorted by grade, the rewrite that matters is invisible, because it lives in the sections that scored well.

Grades are answering a different question than the one you have. They describe where this draft sits on one dimension. They do not describe how much of your next three weeks a given section deserves, and the two come apart constantly — a section can grade well because the fundamentals are strong and still contain the single most actionable note in the document.

The other direction is worth naming too, because it costs writers more time. A low grade frequently reflects an obligation the draft has not met yet rather than something broken. That is not always a rewrite. Sometimes it is a decision you already made on purpose, and the correct response to it is a sentence in your notes explaining why, not a week of work.

One concrete fix: Read the grades once, for shape, then cover them. Do your triage off the cluster list with the grades out of sight.

The page citations are what separate a coverage note from an opinion, and they are underused — most writers read the note, form a reaction, and never click through.

A well-formed note points somewhere specific. In Charade's plot structure section, one of the suggested rewrites reads: give Reggie an active investigative beat in act two — around p.93, in Reggie's room, have her be the one who notices the agenda is missing from the police list, since she has been carrying that bag since the morgue. That is checkable in about ninety seconds. Open page 93, read the scene, see whether Adam does the noticing. Either the note is describing the page or it is not.

Which matters most for the notes you want to argue with. Disagreement with a note comes in two kinds and they look identical from the outside:

1. The note is mistaken about the script. It describes a beat that is not there, attributes an action to the wrong character, or misses a setup that is on the page. Discard it. Do not spend a paragraph refuting it in your head.
2. The note is describing something real in language you would not have chosen. "The antagonist's pressure arrives slowly" may be a clumsy name for a choice you made deliberately about withholding — but the withholding is real, and the note is reporting its downstream effect. This kind reads exactly like the first kind from the couch, and stops reading like it on the page.

The two are only separable on the page, which is the argument for opening it as a rule rather than as a courtesy.

One concrete fix: For every note you instinctively reject, open its page and read the surrounding two pages before the rejection counts. Notes that survive that test go on the list. Notes that do not, you never think about again.

Write it down. This is the step writers skip and the one that determines whether the next draft is better or merely different.

A full report surfaces more than one draft can absorb. If you address eleven notes at once, you get a draft where you cannot attribute any change to any decision, and if it reads worse — which happens, because notes interact — you have no way to find out which fix did it. Rewrites are experiments, and an experiment with eleven variables returns no information.

So the output of reading a report should be short:

• The cluster. One root cause, stated in a sentence, with the pages it touches.
• Mechanical fixes. Page-level things that are unambiguous and cost an hour total — a scene that runs long, a name introduced twice, a setup that never pays.
• Parked. Everything else, with one line each on why it is parked. "Deliberate." "Downstream of the cluster, recheck after." "Note is wrong, checked p.44."

That third list is the valuable one on the next pass. When the second report comes back, the parked list tells you instantly whether a note is new information or something you already ruled on — and stops you relitigating the same decision every draft.

One concrete fix: Cap the next draft at one cluster plus the mechanical list. If the report surfaced two unrelated clusters, that is two drafts.

The rewrite is a hypothesis: if information arriving late is the problem, moving the Dyle backstory earlier should resolve the pacing note, the antagonist note, and the scene-length note together. Hypotheses get tested.

Run the revised draft through the same analysis and read the concern lines again — the same fast pass, the same flat list. Three outcomes, and they mean different things:

• The cluster is gone. The fix held. Move to the next cluster, or to a different kind of read.
• The same note reappears in different words. The rewrite treated a symptom. The cause is upstream of where you worked, which is useful and slightly annoying information.
• New notes appear where there were none. Normal, and not a failure. Fixing a structural problem changes what is visible behind it — a report on a draft with a soft act two is partly reporting on the fog, and clearing the fog reveals the actual terrain.

Rereading is the specific thing a structural pass does that no human reader can, and it has nothing to do with quality. A reader who has already read your script cannot un-read it. Their second pass is contaminated by the first, they will unconsciously grade the delta rather than the draft, and asking a friend to read version four is asking for a comparison. The same analysis applied to draft one and draft four is applying the same rubric both times, which is what makes the comparison mean something.

One concrete fix: Keep the flat concern list from every draft in one file, in columns. Three drafts in, the notes that never leave are the ones you have been avoiding.

One thing, worth being plain about, because it determines what you do next.

A structural report reads the script as a mechanism and reports on the mechanism. Where the engine stalls, which page the goal stops being legible on, whether a genre obligation gets set up and never paid, whether a character's function changes without the script noticing — the Night of the Living Dead sample flags exactly that last one, that Barbara's function collapses into passivity after the opening pages, which is true, checkable, and precisely the kind of finding this instrument is for.

What it is not is a reaction. It will not tell you whether an audience will be moved, whether the voice is one somebody would want to spend two hours with, or whether the thing is good. Those are real questions and they have a different instrument: readers, ideally more than one, briefed properly. That is a whole method of its own, and the comparison of beta readers, peer swaps, and structural coverage covers how to run it.

Use the report for what it reports on. A structural map with page numbers is enough to rewrite from, on its own, at any stage of a draft — and if you are still trying to work out why the passes keep coming before you have a report in hand, the diagnostic guide is the piece that starts one step earlier.

One concrete fix: After you finish the cluster list, write down the one question the report did not answer. That sentence tells you who to hand the next draft to.

1. Copy every concern line into a flat list before reading any section in full.
2. Cluster the list. Write the one sentence that makes more than one line go away.
3. Read the grades once for shape, then cover them and triage off the cluster.
4. Open the cited page for every note you want to reject.
5. Write down the cluster, the mechanical fixes, and — explicitly — what you are parking and why.
6. Rewrite one cluster. Rerun. Check the note is gone.

The report is not a grade on your script and it is not a to-do list. It is a set of readings from six instruments pointed at the same object, and the job is to work out what one thing they are all describing.

Both sample reports in this article are public and complete: Charade and Night of the Living Dead. Read them the way this article describes rather than top to bottom, on a script whose problems you can already see, and the method takes about ten minutes to learn. Then it is yours for every report you ever get, from anyone.

Coverage for writers, not development execs — $20 a script, five minutes, no subscription.