If you review a draft, you need to check whether it addresses the subject and audience in the brief. Content Optimizer compares your writing with categories selected from 11 taxonomies containing 25,622 categories, including 4,117 US cities. It returns the nearest categories and the phrases associated with your chosen category that the draft does not contain.
Choose the categories the page is meant to address
Select up to three categories, each from a different taxonomy. The editor refuses two categories from the same taxonomy. Each result reports one selected category's position within that taxonomy; two selections would require separate positions.
In practice you are stating three intentions: a topic category for the subject, an audience category for the intended reader, and a location or product category where relevant.
The campaign's taxonomy presets supply defaults. Your own choices are remembered in your browser for that campaign, and the presets can be restored.
Read the category list before editing
Each selected taxonomy returns its ten nearest categories. Your intended category is highlighted, and if it falls outside the ten the result gives its position separately.
The list stops at ten. A list of twenty-five proved unhelpful, and even ten can include categories with little relevance. After the first few matches, the remaining results may be the closest available without being useful. Focus on the first few categories rather than assuming tenth place is relevant.
An absent target category is useful evidence: the draft is closer to ten other categories than to the one you selected.
A worked example
Take a guide written for facilities managers choosing a maintenance contractor. You select an audience category for facilities management.
The report returns the contracting trade as nearest, several construction categories after it, and facilities management well down the list.
The cause is usually visible in the draft. The opening explains what contractors do, the certifications section describes their qualifications, and the pricing section explains how they quote. Each section addresses the supplier's work rather than the buyer's decision. The information is accurate but addresses the wrong audience.
That is a structural problem no term count reveals, because both audiences share a vocabulary.
Read the percentages correctly
The percentages measure similarity of meaning. They do not grade factual accuracy, completeness or likely search performance, and no engine is involved in producing them.
Compare the categories and their relative positions rather than aiming at a high number. A draft that reaches its intended category as the nearest match has answered the question the report asks.
Assess the missing phrases before adding them
The phrase list separates terms already present from terms associated with your category that are absent.
Read the second list as a description of what writing in that category usually involves, and ask which of them you have a genuine reason to discuss. Inserting terms to move the number produces worse writing every time, and the measurement cannot tell that it happened.
A workable test: if you cannot write a full sentence a reader would benefit from, the phrase does not belong.
The input limit and why long text is refused
The editor accepts 8,192 bytes and refuses longer text rather than trimming it. Most Latin characters take one byte, so that is roughly 1,200 to 1,400 words of ordinary English. Accented characters and other scripts take more, so text in those languages reaches the limit sooner than its character count suggests.
The refusal is deliberate. Scoring the first 8,192 bytes of a long article would return a number describing the opening and present it as the whole article. A page that starts well and wanders would look correct.
For a long piece, submit sections separately and read them as separate measurements. They do not combine into one article score, and averaging them yourself reintroduces the problem the refusal avoids. Checking sections separately is usually more useful anyway, because it identifies which part drifted.
At the editorial check, read the category results beside the draft and justify each revision through the reader’s unanswered question.
See how the comparison is produced, checking a draft before publication, or the US Cities taxonomy.