This analyser scores a draft out of 100 for the things you can check before publishing: keyword placement, density, headings, length, sentence length and paragraph structure. It shows where each point came from and lists the edits that would recover the missing ones.
The score is about placement and structure, not about whether the page deserves to rank. A draft can score 90 and still be dull, wrong or aimed at the wrong query. Use the breakdown, not the number.
The eight checks and their weights
Title tag is worth 15: 10 for containing the keyword and 5 for a length between 30 and 60 characters, or 3 for up to 70. Meta description is worth 10: 6 for the keyword and 4 for a length between 120 and 160. Opening is worth 10 if the keyword appears in the first 100 words. Headings are worth 15: 10 for two or more headings, 6 for one, plus 5 if any heading contains the keyword.
Density is worth 15 and scores full marks between 0.5% and 2.5%, 8 points up to 4%, and nothing above that. Length is worth 15 against a guide of 600 words for an article, 350 for a product page and 250 for a category page, with partial credit at 60% and 30% of the guide. Readability is worth 10 when sentences average 10 to 22 words and long words stay under 25%. Structure is worth 10 for three or more substantial paragraphs and two or more headings.
Two details change the result more than people expect. Title and description matching is a plain substring check, so a hyphenated or joined variant can pass. Body matching is stricter: the keyword must appear as a sequence of whole words. And the H1 only counts as a heading if you paste it as one, either inside an <h1> tag or as a Markdown # line.
A worked example
A 264-word article with the keyword keyword density, a 39-character title and a 152-character description scored 90 out of 100 and was banded Healthy. The keyword appeared three times, a density of 1.14%.
| Check | Points | What was found |
|---|---|---|
| Title tag | 15/15 | Contains the keyword, 39 characters |
| Meta description | 10/10 | Contains the keyword, 152 characters |
| Opening | 10/10 | Keyword appears in the first 100 words |
| Headings | 15/15 | 3 headings, one contains the keyword |
| Density | 15/15 | 1.14% (3 uses in 264 words) |
| Length | 5/15 | 264 words against a 600-word guide |
| Readability | 10/10 | 13.2 words per sentence, 6.8% long words |
| Structure | 10/10 | 6 paragraphs and 3 headings |
The only weak check was length, and it was the right thing to flag. The draft covered its topic in 264 words and did not need another 336 words of padding. The tool cannot know that; it can only compare the count with its guide, which is why the fix list says length is a proxy rather than a rule.
How readability is measured
Sentences are split on full stops, question marks and exclamation marks. The average is the word count divided by the number of sentences with more than one word. Long words are those longer than seven characters, and the percentage is against every word in the body.
This is a rough measure. It does not understand clauses, lists or technical terms that are long but familiar to the audience. A legal page will always look worse than a children's story on this check, and that does not make it badly written.
What the score leaves out
It does not compare you with the pages already ranking, check factual accuracy, judge whether the page matches intent or read the internal links. It does not know whether the keyword is the right one. The most common way to get a high score with a bad page is to optimise for a phrase nobody searches.
It also does not see images, alt text, schema, canonical tags or page speed. Those are separate checks, and several of them matter more than a title that happens to contain an exact phrase.
Treat 80 or above as "the basics are in place", not as "ready". Read the draft once as a reader would, then fix the two or three findings that would actually change the experience.