Entity Name Consistency

Entity Name Consistency checks two things: that your brand name is capitalised the same way everywhere, and that each H2 section opens with a real subject instead of a pronoun. Both signals help AI systems like ChatGPT, Bing, and Google resolve your site into a single, clear entity and quote your sections accurately. The check runs on English only — if a page's lang is not English, it gets skipped.

What does this parameter measure?

Two HTML/DOM signals that keep your entities legible to AI systems: consistent brand-name capitalisation, and H2 sections that stand on their own.

friendly4AI pulls multi-word capitalised phrases from your headings and paragraphs — two or more capitalised words in a row, with leading articles and prepositions stripped. It groups them by their lowercase form and flags any group that shows up in two or more distinct case forms, say "GEO Scanner" in one place and "Geo scanner" in another.

It then looks at the first sentence of each H2 section and flags any that open with an anaphoric pronoun: It, This, or They.

Why does it matter for AI-readiness?

AI knowledge graphs resolve your brand into one entity by matching its name across mentions. Spell that name three different ways and the graph can split it into separate entities, watering down the authority you have built.

Anaphoric openings break extraction too. When a model lifts out a section that begins "It also handles billing", the passage loses its meaning, because "It" no longer points to anything. Name the entity in the H2's first sentence and every passage stays self-contained and citable. This works alongside heading hierarchy and section length distribution, which govern how cleanly each section stands on its own.

How is it scored?

Under the v4.5 methodology, this Content Structure parameter is a penalty-based gradient. The processor starts at 100 and applies max(0, 100 - brandInconsistencies * 15 - pronounDriftInstances * 10).

• Pass (100): no brand-name case variants and no anaphoric H2 openings.
• Partial: one or two of either issue. Each brand inconsistency costs 15 points; each pronoun-drift section costs 10.
• Fail: penalties pile up into a confused entity graph — roughly three or more brand variants, three or more anaphoric openings, or some mix of the two.

Non-English pages return SKIPPED (non_english_language) instead of a score, because the detection rules are tuned for English capitalisation and English pronouns.

How do you fix common issues?

• Pick one canonical capitalisation for your brand name and use it everywhere, in headings and body copy alike.
• Rewrite any H2 opener starting with "It", "This", or "They" so the first sentence names the subject outright.
• Make each section's opening sentence readable cold, as if someone landed there straight from a search.
• Each brand variant costs 15 points and each anaphoric opening costs 10, so start with the most frequent offenders — that recovers the most score.
• Re-scan and check the brandVariants, brandInconsistencies, and pronounDriftInstances evidence fields.