Content analyzer
What is entity density and why does ENTITY_SPARSE reduce AI citability?
Seeklyn fires ENTITY_SPARSE when a passage of 50 or more words names fewer than 2 distinct entities — brands, people, products, places, and organizations. Entity count shows a ~0.66 correlation with AI citation, the strongest single observed predictor in Seeklyn's content rubric.
Seeklyn flags this in every content analysis — run your page free at /dashboard/content.
What is entity density?
Entity density is a measure of how many distinct named entities appear in a passage relative to its length. Seeklyn's ENTITY_SPARSE flag fires when a passage is 50 or more words long and Seeklyn's entitySet function returns fewer than 2 distinct entities. Short passages under 50 words are skipped — very brief chunks can legitimately name only 1 entity in context. A passage naming "Seeklyn" and "Google AI Overviews" clears the flag even if it names no other proper nouns.
Entity count shows a ~0.66 correlation with AI citation — the strongest single observed predictor in Seeklyn's 8-flag rubric. The implication is concrete: passages that name specific brands, people, products, and organizations are cited roughly twice as often as passages that rely on generic nouns like "the tool," "the platform," or "the company."
Why do named entities boost AI citations?
Answer engines use named entity anchors — brand names, people, product names — to match passages to user queries. A passage about "the tool" is nearly impossible to retrieve for a query like "how does Seeklyn score content" because the passage contains no query-matching token. A passage that names "Seeklyn," "ChatGPT," and "Google AI Overviews" matches a wider surface of entity-bearing queries and gets retrieved more often.
Beyond retrieval, entities add specificity that makes a passage more authoritative to the engine's ranking model. Saying "Perplexity surfaces attributed passages 40% more often than unattributed ones" is more citable than "AI engines prefer attributed content" — the former names a specific system. Entity anchors let the engine produce structured answers with identifiable subjects, the format synthesis favors.
How Seeklyn detects entity sparseness
ENTITY_SPARSE fires when the passage word count is ≥50 and the entity set size is <2. Seeklyn's `entitySet` function scans each passage for proper-noun tokens and returns the deduplicated set. Deduplication means naming "Seeklyn" five times counts as only 1 entity — variety of distinct names, not repetition of 1 name, is what the flag rewards. The deduplicated count is compared against the minimum threshold of 2 distinct entities.
Passages under 50 words are exempt because the flag is designed for full content passages, not bullet-point fragments. A 30-word subheading that names only one brand is not sparse; it is a different structural unit that Seeklyn does not score the same way as a prose paragraph.
How to add entities to a passage
Replace generic nouns with the specific proper name. "The tool analyzes your content" becomes "Seeklyn's content analyzer scores each passage against 8 AEO flags." "The platform" becomes "Perplexity" or "Google AI Overviews." "The study" becomes "Princeton's GEO research (KDD 2024)." Each substitution adds 1 entity to the set without increasing word count.
Aim to name at least 2 distinct entities per 50-word passage — the minimum threshold — and ideally 3–4 per 100-word passage. Product names, organization names, researchers, publications, and place names all count. Adding entities also reinforces BM25 retrieval: named entities frequently overlap with query terms, so passages with ≥2 entities tend to score higher on both the Retrieval and Citation stages of Seeklyn's 3-stage citation funnel.
Frequently asked questions
What counts as a named entity for Seeklyn?
Named entities include proper nouns: brand names (Google, Seeklyn, ChatGPT), people names, product names, place names, and organization names. Common nouns ("the tool," "the platform," "the company") do not count. The distinction is between a proper noun that identifies a specific thing and a generic noun that describes a category.
Does a 40-word passage fire ENTITY_SPARSE?
No — Seeklyn skips entity-density checks for passages under 50 words. The 50-word floor exists because very short passages may legitimately name only 1 entity in context and should not be penalized for brevity. ENTITY_SPARSE only fires on passages ≥50 words with fewer than 2 distinct entities.
Is entity density the same as keyword density?
No — entity density counts distinct proper-noun mentions; keyword density counts how often any token repeats. High keyword density (repetition) actually fires the KEYWORD_STUFFING inverted signal, which correlates negatively with citation. Increasing entity density by naming more specific proper nouns improves citability; increasing keyword density by repeating the same word hurts it.
How does entity density relate to BM25 retrieval?
Named entities often overlap with query terms, so passages with ≥2 entities tend to score higher on BM25 retrieval as well. Retrieval (query overlap) and citability (entity density) are separate scores in Seeklyn's citation funnel — but adding entities tends to lift both stages simultaneously, because a specific brand or product name is more likely to appear in a user's query than a generic noun.