By Ryan Richardson · Published 8 October 2026
A filled row, for a hypothetical chapter on pricing: promise, 'the reader stops guessing at a number and tests one instead'; before, 'picks a price off a feeling'; after, 'can name the test that would prove the feeling wrong'; handed over, 'one worked test design'; carrying line, 'the next chapter is what to do once the test disagrees with the feeling.' A row that only has the chapter title and a vague topic description isn't a full row yet.
Build the grid somewhere you cannot draft, if you can manage it. Drafting is more fun than deciding, and inside a drafting tool you'll have written good words by the afternoon, standing on a structure nobody checked.
Interrogate the whole grid before any chapter gets assigned, not chapter by chapter as each one comes back. A chapter in the wrong place is a five-minute fix in a grid; once prose around it depends on where it sits, moving it becomes a two-day rewrite.
Once a first draft exists, run the grid again in reverse: summarise each chapter as written, not as planned, and compare the two. That pass catches drift the original outline never finds, because a writer, human or AI, reads what it meant to write more easily than what's actually on the page.
Set a threshold before you look, the same way you'd set a scoring floor before a score. If more than a third of chapters no longer match their stated promise, the manuscript has drifted structurally, and the fix is structural, not editorial. Rewriting sentences in a chapter whose promise has already moved just makes the wrong promise read more smoothly.
| Claim | Value | Source |
|---|---|---|
| Matrix difficulty/coverage for this step | Moderate / Often skipped | The Sixty Steps matrix |
| Drift threshold that signals a structural, not editorial, fix is needed | more than one-third of chapters no longer match their stated promise | The Sixty Steps manuscript |