Entity Extractor

LLM-based search does not rank keywords, it retrieves and reasons over entities: the people, products, organizations, technologies, and concepts a page is genuinely about. This tool reads your pasted content and surfaces the entities it detects, so you can see your page the way a retrieval system roughly does.

It combines three passes: a curated dictionary of SEO and AI-search terms, proper-noun detection for named entities, and acronym detection. Each entity gets a salience score that weights how often it appears and how early, then everything is ranked.

Use it to check that a page about a topic actually mentions the supporting entities it should, and export the result as a mentions JSON-LD block or a CSV. It runs entirely in your browser.

Counting how often a word appears tells you what you could see by reading the page. It also rewards the wrong thing: a term repeated fifteen times in a footer outranks the subject of the article.

What matters is where an entity appears. A term in the title, in a heading, or in the opening third is doing structural work, because that is where a page establishes what it is about. Assistants retrieve passages rather than pages, and a passage is retrievable when it names its subject in the part that sets the subject up. So position is weighted here, not just counted.

Paste a competing page into the second field and the tool reports the entities it names that yours does not.

That is the concrete version of advice usually given as a score. “Your content is not specific enough” is not actionable. A list of the eleven things a page on this subject mentions and yours never does is a brief you can write against.