toolkit

Scenario · Content and search

This page reads like a medical journal

“Patients say they don't understand this page.”

Practice websites get written by people who understand the procedure, for people who are trying to decide whether to book one — and the result routinely reads at a difficulty level built for the wrong audience. This measures exactly how hard a page is to read, rewrites it toward an agreed easier target while deliberately preserving every fact, number and call to action, and re-measures afterward so the improvement is a number rather than someone's impression; a clinician still has to confirm the simpler version is still medically accurate before it goes live.

What to ask for

Content qualityContent-quality audit — reads a page's visible text and grades whether the copy itself is any good: readability (Flesch–Kincaid grade and reading ease), thin content by word count, H1/title duplication, and leftover placeholder or lorem-ipsum text.Readability rewriterReadability rewrite — takes web copy that reads too hard and rewrites it to a target reading grade while keeping the meaning.Before/after checkBefore/after verification for a change to a live page — the safety net for the small edit that doesn't justify a staging round-trip.Spelling checkSite-wide spelling sweep — fetches a page (or every URL in a list), extracts the visible text, and flags suspected misspellings with a hunspell dictionary (nspell + dictionary-en).

See it work

A real run of Readability rewriter:

REVIEW — grade 31.2 → 12.1 (target 8)
  [warn] reading grade 31.2 (Very Difficult) — hard to read; aim for grade 6–12

  rewrite: ./out/rewrite.html
  report:  ./out/readability-rewriter-readability-rewriter.html
  json:    ./out/readability-rewriter.json

The captured report, exactly as a run hands it to a client —open the full report ↗

Practice sites are written by people who know the procedure, for people who do not. The result reads at a college-plus level on pages meant for someone deciding whether to book. This is measurable, and the fix is measurable too — which is unusual for a copy problem.

  1. Grade the page as it stands. The content-quality check reports the reading grade and reading ease, along with thin content and heading problems. A number is what makes this arguable with whoever wrote it.
  2. Agree the target grade. Patient-facing copy generally wants grade 8–10; a page for referring physicians legitimately does not. Rewriting the wrong page to a lower grade is a downgrade.
  3. Snapshot the page.
  4. Run the readability rewrite. It rewrites to the target grade while preserving facts, numbers, names, claims and calls to action, then re-grades so the improvement is measured rather than asserted. It deliberately hands you the rewrite instead of publishing it — a whole-body rewrite has too much blast radius to apply unread.
  5. [manual] Have a clinician read it, then publish it. This is the step that cannot be skipped on medical copy: the rewrite is a draft by construction, and the person who can tell you it is still accurate is not in this procedure. Publish and purge.
  6. Verify. Re-grade the live page to confirm the improvement landed, spell-check the new copy, and before/after check the page so a rewrite that dropped a heading or a booking link is caught.

Measured, not asserted

The reason this scenario is worth writing down is that “make it simpler” is otherwise unfalsifiable. Grade before, target agreed, grade after, on the live page. A rewrite that did not move the number did not do the job, regardless of how it reads to the person who wrote it.

What this does not cover

Clinical accuracy after the rewrite. A readability rewrite is instructed to preserve every fact, number, name, claim and call to action, and it is still a machine rewriting medical copy — a clinician has to read it before it goes live. Nothing here judges whether the simpler version is still true, and on treatment copy that is the only question that matters.

← All scenarios