What is a good readability score?

60 or above for a page aimed at a general reader. That band sits at standard on the Flesch table and reads like a sentence of 15 to 20 words. Documentation written for specialists lives lower, around 30 to 50, where the vocabulary is the point. Under 30 the reader has to go back over sentences to follow them.

What is the Flesch-Kincaid grade level?

The US school year a reader needs to follow the text on one pass. Grade 8 means an eighth grader reads it without stopping, and most web copy lands between grade 6 and grade 9. This checker reports it for English, because the formula was fitted on English syllables.

How does the heading checker read my page?

It reads Markdown heading lines and h1 to h6 tags in the order they appear, then prints the outline with each problem marked. It flags a page with no H1, a page with two, a level that jumps from H2 to H4, a heading with no text, and two headings that say the same thing.

Does it work in Spanish?

Yes, Spanish drafts are scored with Fernández Huerta instead of Flesch. Syllables are counted with the Spanish rules for diphthongs and hiatus, and the long-sentence limit moves to 30 words. Leave the selector on auto and the language comes from the text.

Does my text leave the browser?

The score on this page is computed in your browser, so a draft you paste here stays on your machine. The JSON endpoint behind the same check exists for agents: a call to it sends the text to the server, which scores it and returns the report without storing it.

Readability checker

Paste a draft and see how hard it is to read. The check gives you the Flesch score with its band, the Flesch-Kincaid grade, the sentences that run long, and the heading outline down to the H1.

Free, no sign-up

Your draft

Language

Run it from your agent

The check on this page is also a JSON endpoint and a tool on a public MCP server. Neither needs an account or a key.

JSON endpoint

curl -X POST https://seodraft.app/api/tools/readability-checker \
  -H 'content-type: application/json' \
  -d '{"text":"# How we price\n\nWe charge $5 a month. The trial lasts 7 days and starts when you sign up, which is long enough to write a week of drafts and see what the rules say about them.\n\n## What you get\n\nEvery draft is checked before it ships.","language":"en"}'

Free MCP server

Add this URL to Claude, ChatGPT, Cursor or any MCP client. It exposes every free tool, with no sign-in.

https://seodraft.app/tools/mcp  ·  check_readability

This page for agents

The same guide and the API contract, as Markdown. /tools/readability-checker.md

What a readable draft looks like

A readable draft keeps one idea per sentence and one claim per paragraph. On the Flesch scale that lands above 60 for a page written for a general reader, which reads like a sentence of 15 to 20 words built from words the reader already knows.

Sentence length is what moves the score most. A sentence over 25 words usually carries a second idea, and the reader holds the first one while hunting for the verb of the second. Cutting it where the second idea starts raises the score, and the reader keeps both ideas.

The heading outline is the other half of a page that reads well. One H1 names the page and the H2s under it name its sections. Levels that run in order let a screen reader, a search engine and a person skimming on a phone follow the same structure.

How this readability checker scores your text

English text gets Flesch reading ease and the Flesch-Kincaid grade level. Reading ease is 206.835 minus 1.015 times the words per sentence minus 84.6 times the syllables per word, so the two levers are sentence length and word length. The grade level turns the same two numbers into the US school year a reader needs.

Spanish text gets Fernández Huerta, the Spanish adaptation of Flesch: 206.84 minus 0.60 times the syllables per 100 words minus 1.02 times the sentences per 100 words. Leave the language on auto and the checker picks the formula by counting which language's function words the text uses.

Syllables come from a documented heuristic. English counts vowel groups, drops the silent trailing e and the silent -ed, and adds back the beat of a consonant plus -le. Spanish counts vowel nuclei, keeps a diphthong as one beat and splits a hiatus in two.

Only prose is measured. Fenced code, Markdown syntax, image markup and URLs come out first, every list item counts as its own block, and a heading is read as a heading instead of a sentence.

What each finding means

Every finding names the change it asks for. hard-to-read appears under a score of 50. long-sentences lists up to five sentences over 25 words in English or 30 in Spanish. long-paragraphs counts blocks over 120 words, which is where a paragraph turns into a wall.

passive-heavy appears when more than a tenth of an English text reads as passive, spotted by a form of be next to a past participle. It is a heuristic, so read the examples before rewriting them.

The heading findings read the outline. no-h1 means a text that has headings and none at level 1, and multiple-h1 means two or more. skipped-heading-level catches an H2 followed by an H4. empty-heading is a heading line with no words, and duplicate-heading is two sections with the same title, which usually means both cover the same ground.

too-short is context rather than a defect: under 100 words the formulas swing on a single sentence, so the score is a hint at that length.

What the score leaves out

A readability score measures the length of sentences and words. Whether the text says anything stays outside it, so a page of short sentences about nothing scores above 70 and an editor still sends it back.

Syllable counting is a heuristic, so a name, an acronym or a loanword can come out a beat off. Across a page the effect is small. On one sentence it shows, which is worth remembering before rewriting a line because the number moved.

Google publishes no readability threshold, and a score is not a ranking factor. The reason to raise it is the reader who gets the answer on the first pass, and the AI assistant that quotes a sentence it can lift whole.

Questions

What is a good readability score?: 60 or above for a page aimed at a general reader. That band sits at standard on the Flesch table and reads like a sentence of 15 to 20 words. Documentation written for specialists lives lower, around 30 to 50, where the vocabulary is the point. Under 30 the reader has to go back over sentences to follow them.
What is the Flesch-Kincaid grade level?: The US school year a reader needs to follow the text on one pass. Grade 8 means an eighth grader reads it without stopping, and most web copy lands between grade 6 and grade 9. This checker reports it for English, because the formula was fitted on English syllables.
How does the heading checker read my page?: It reads Markdown heading lines and h1 to h6 tags in the order they appear, then prints the outline with each problem marked. It flags a page with no H1, a page with two, a level that jumps from H2 to H4, a heading with no text, and two headings that say the same thing.
Does it work in Spanish?: Yes, Spanish drafts are scored with Fernández Huerta instead of Flesch. Syllables are counted with the Spanish rules for diphthongs and hiatus, and the long-sentence limit moves to 30 words. Leave the selector on auto and the language comes from the text.
Does my text leave the browser?: The score on this page is computed in your browser, so a draft you paste here stays on your machine. The JSON endpoint behind the same check exists for agents: a call to it sends the text to the server, which scores it and returns the report without storing it.

The full process runs in seodraft

Inside seodraft the same prose layer runs as a gate on every draft: the answer in the first sentence under each H2 (answer-first), paragraph length and sentence variety, next to the brief the agent writes before the body and the evidence bank the article draws from.

Create account Create account Connect your agent