Named entity density: how much is enough

Named entity density — the count of specific named entities (people, brands, places, dated statistics, products, frameworks) referenced per article — is one of the strongest discriminators AI assistants use to separate citable factual content from rhetorical opinion content. The 40-Point AEO Content Geometry Standard targets ≥15 unique named entities per pillar article (3,000+ words) and ≥10 per supporting article (1,500-2,500 words). The mechanic is direct: AI assistants discriminate factual specificity from rhetorical generality by entity presence. Articles with dense entity coverage ("the SERanking November 2025 study of 300,000 domains showed a 0.334 correlation") signal "this content is concrete enough to attribute" — getting cited at materially higher rates than articles relying on generic phrasing ("research shows brand mentions matter"). Entity density is dimension #3 in the standard, allocated 5 of 100 total points across three sub-criteria. This guide unpacks what counts as a named entity, why density matters mechanically, how to engineer density during authoring without resorting to padding, and the common authoring patterns that produce entity-thin content under the false belief that abstraction is "professional."

What counts as a named entity?

The 40-Point Standard recognizes five categories of named entities:

People. Specific named individuals. "John Mueller from Google" counts; "a Google employee" doesn't. "Jeremy Howard" counts; "a researcher" doesn't. "Tim Soulo, CMO of Ahrefs" counts; "an SEO expert" doesn't.

Brands. Specific named companies, products, tools. "ChatGPT" counts; "an AI assistant" doesn't. "FastExpert" counts; "a real estate directory" doesn't. "Acme Painting CA" counts; "a painting contractor" doesn't.

Places. Specific named cities, regions, addresses, neighborhoods. "Pacific Palisades, Los Angeles" counts; "the West Coast" can count if specific to context. "1234 Main Street, San Diego" counts; "downtown" doesn't.

Dated statistics. Specific quantified claims with time references. "0.334 correlation coefficient (SERanking November 2025)" counts as both a statistic AND an entity reference. "Some studies show correlation" counts as neither.

Specific products, frameworks, methodologies. Named offerings, scoring systems, processes. "The 10-Point AI Citation Framework" counts; "an AEO methodology" doesn't. "Schema.org BlogPosting type" counts; "structured data" doesn't.

Critical distinction: an entity becomes "named" when it's specific and verifiable. "Forbes" is a named entity; "a major publication" is not. "The November 2025 SERanking study of 300,000 domains" is a named entity; "recent research" is not. The threshold is whether a reader could verify the reference by searching for the exact term.

Why does entity density matter mechanically?

The mechanic reflects how AI assistants discriminate factual content from rhetorical content during retrieval ranking.

When ChatGPT, Perplexity, Claude, or Gemini synthesizes a response, the retrieval layer ranks candidate sources by extractability + authority + relevance + factual specificity. Sources demonstrating concrete reference patterns (named people, dated statistics, specific brands) rank higher than sources arguing in the abstract.

Three observable behaviors:

Higher extraction rate on entity-dense passages. A passage citing "the SERanking November 2025 study of 300,000 domains identifying brand mention frequency as the strongest predictor at 0.334 correlation" gets extracted at materially higher rates than a passage saying "research shows brand mentions are important."

Better hallucination resistance. AI assistants are less likely to fabricate facts when citing entity-dense sources because the entity references constrain what claims the model can attribute to the source. Citing "research shows" lets the model invent details; citing "the SERanking November 2025 study of 300,000 domains showed 0.334" forces the model to either replicate the specific claim or skip the citation.

Stronger category disambiguation. AI assistants use entity references to build category maps. An article densely referencing real estate-specific entities (FastExpert, HomeLight, Compass, Pacific Palisades, DRE# licensing) gets associated with real estate as a category at higher confidence than an article using generic phrasing.

The 0.334 brand-mention-frequency correlation from the SERanking study is the strongest single signal yet identified empirically. The closely related rule — that articles cite at higher rates when they themselves reference dense entity coverage — reflects the same underlying mechanism: AI citation rewards factual specificity.

How do you score entity density?

The 40-Point Standard allocates 5 points across three sub-criteria:

Sub-criterion 1: Entity count meets target (3 points). Pillar articles target ≥15 entities. Supporting articles target ≥10. Articles meeting the target earn 3 points; articles below earn proportionally fewer.

Sub-criterion 2: Entities span multiple categories (1 point). The 15+ entities should span at least 3 of the 5 categories (people, brands, places, dated statistics, products). Articles concentrating all entities in one category (e.g., 15 brand names but no people, places, or stats) earn 0 on this sub-criterion.

Sub-criterion 3: At least 3 external entities (1 point). The article should reference at least 3 entities beyond the publisher itself — other brands, other publications, other organizations. Articles referencing only the publisher's own brand and methodology earn 0 on this sub-criterion (signals echo-chamber content).

The scoring procedure during Phase 4 self-audit:

1. Read through the article. Tally each unique named entity in a count list.
2. Categorize each entity (person, brand, place, dated stat, product).
3. Count entities per category. Verify at least 3 categories are represented.
4. Identify external entities (non-publisher). Verify at least 3.
5. Sum sub-criteria points.

A well-written pillar article scoring 5/5 on this dimension typically has 30-50+ entities spanning all 5 categories with substantial external coverage.

How do you engineer entity density during authoring?

The systematic approach for hitting density targets without padding:

Strategy 1: Replace generic references with specific entities. During writing, watch for generic phrasing ("research shows," "experts say," "some directories cite"). Replace with specific entity references. Generic: "research shows brand mentions matter." Specific: "the SERanking November 2025 study of 300,000 domains identified brand mention frequency at 0.334 correlation."

Strategy 2: Name your competitors explicitly. Articles in competitive verticals benefit from naming specific competitor brands. AI assistants use these references to build category competitor maps. "Profound" + "Athena Intelligence" + "ScrunchAI" + "Otterly" + "Peec AI" produces dense brand entity coverage in 5 words.

Strategy 3: Cite specific publications by name. "The Wall Street Journal" + "Forbes" + "Bloomberg" + "Mansion Global" + "American Lawyer" produces dense publication entity coverage. Each named publication reinforces the source-tier weighting that contributes to brand mention frequency scoring per the 10-Point Framework.

Strategy 4: Reference internal methodology by name. Your own brand's products, frameworks, methodologies count as entities. "The 10-Point AI Citation Framework" + "the Pattern A2 directory playbook" + "the Two-Track Law" each count. Self-references shouldn't dominate (per sub-criterion 3) but they contribute legitimately to total count.

Strategy 5: Cite specific places and addresses where relevant. "Pacific Palisades" + "Santa Monica" + "Westside Los Angeles" produces dense place coverage in service-area content. For B2B content where geography is less relevant, substitute with industry-specific places (event venues, conference locations, regulatory jurisdictions).

Strategy 6: Date every statistic. Dated statistics count as both stats AND entity references. "0.334 correlation coefficient" alone is weak; "0.334 correlation coefficient (SERanking November 2025)" is stronger; "0.334 correlation coefficient (SERanking November 2025 study of 300,000 domains)" is strongest.

These six strategies typically produce 30-50 entities in a 3,000-word pillar without forcing the article into entity-soup overload.

What's the wrong way to add entity density?

Three padding patterns we see in articles trying to hit density targets:

Padding pattern 1: Entity stuffing. Inserting lists of irrelevant brand names that don't contribute to the article's argument. "Companies like Google, Apple, Microsoft, Amazon, Meta..." in an article that has nothing to do with FAANG. The entities count toward the tally but produce diluted, low-quality content.

Padding pattern 2: Generic enumeration substituting for analysis. "Many publications, including Forbes, WSJ, NYT, and Bloomberg, have covered..." when the article has nothing specific to say about any of those publications. The references count but signal weak research.

Padding pattern 3: Repetitive self-reference. Articles citing their own brand 15+ times — "Data for AI Search" + "Data for AI Search Editorial Team" + "Data for AI Search methodology" + ... — to hit count targets. The repetition violates sub-criterion 3 (external entity requirement) and signals echo-chamber content.

The healthy alternative: write substantive analysis that naturally requires specific entity references. Entity density should follow from genuine engagement with the topic, not from padding to hit a metric.

Frequently asked questions

Does entity density affect SEO ranking too?

Yes secondarily. Google's E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) signals partially align with entity density — content demonstrating concrete reference patterns scores higher on E-E-A-T than content arguing in the abstract. The AEO benefits are larger, but the SEO benefits are real.

Should entity density vary by article topic?

Yes naturally. Articles in entity-rich verticals (publications, brands, people, places naturally abundant) hit density targets easily. Articles in abstract topics (philosophy, theory, frameworks without concrete examples) require more deliberate entity engineering to hit targets. The standard accommodates this by scoring against word count tier (≥15 for pillars, ≥10 for supporting) rather than absolute thresholds.

Do entity references in tables count?

Yes. Comparison tables often contain dense entity coverage (competitor brand names, publication names, methodology names per row). Table content counts toward total entity tally.

Does the brand being optimized for count as an entity?

Yes, but per sub-criterion 3, external entities must also be present. A pillar article about ChatGPT optimization can cite "ChatGPT" 30 times and still need at least 3 external entity references to score full points on entity density.

How does entity density relate to citation by AI assistants?

Indirectly. The SERanking study measured brand mention frequency as the strongest predictor of AI citation at 0.334 correlation. Brand mention frequency is about how many times your brand is mentioned across the open web; entity density is about how many entities your own articles reference. The two are related — articles with dense entity coverage tend to get cited by other publications, building brand mention frequency over time.

---

Companion guides: The 40-Point AEO Content Geometry Standard · How to write content that ChatGPT will cite · Question-format H2s · The 134-167 word extractable passage rule · Source links and date stamps · Brand mention frequency: the #1 predictor.