Home/Features/Content Optimization
Content OptimizationRefine

Five scores, and the weight behind each one

The draft is scored three seconds after you stop typing, on five things that can be counted: whether it has an H1, whether it has subheadings, how long it is, how hard it is to read, and how the focus keyword sits in it. Every weight is on screen.

quietly.sh/contentproduct exhibit
Quality score
83
H1 present0.20100
Subheadings0.15100
Length against 600 words0.2071
Flesch-Kincaid readability0.1560
Focus keyword usage0.00not set

weights renormalized over 0.70

A dimension that did not run drops to weight zero and the rest renormalize, so the overall stays an average of what was measured.

What it does

A score you can take apart

Title carries 0.20 of it, headings 0.15, length against a 600-word target 0.20, readability 0.15, and keyword usage 0.30 when you have set a focus keyword. Leave the keyword blank and that dimension is reported as not run while the rest renormalize, so the overall stays an average of what was measured.

Readability from the actual arithmetic

Words per sentence and syllables per word, run through Flesch-Kincaid to a grade level, then mapped to a score. It is the same formula a copy desk uses, computed on your text with no model in the loop, so it returns the same answer every time you ask.

Tone read back to you, and the grammar caught

How professional, casual, academic and technical the draft reads, with the dominant register named. Alongside it, the grammar and phrasing problems with a proposed replacement for each. This half needs the model, and when it is unavailable the panel says so and the measured scores keep working.

Two scores we took out on purpose

There used to be an originality percentage and an AI-detection score. The first asked a model to judge plagiarism having searched nothing, and both fell back to a hardcoded number whenever the call failed, so they rendered with total confidence on a check that never ran. A measurement of nothing is worse than a blank, so they are gone.

How it works

  1. 1

    Set one focus keyword

    One phrase, the one the piece is for. Keyword usage is then scored against how it actually sits in the text, and a SERP preview shows the title and description as a result page would render them.

  2. 2

    Write, and read the panel

    Scoring runs three seconds after you stop typing, once the draft passes a hundred characters. Every save keeps a version you can open and restore, so a rewrite you regret is one click back.

  3. 3

    Publish where the site lives

    Send the piece to WordPress, to a GitHub repo, or to a site you built here, as a draft or as published. No score gates the button. The panel tells you what it found and you decide when it is finished.

Write against numbers that mean something

Seven-day trial. Paste a draft in and read where the five scores land before you change a word.