Module 04 · Lesson 2 of 4
Entity Naming and Descriptive Headings
An entity is a distinct thing a system can identify: a person, an organisation, a product, a place, a concept. Search and AI systems read pages partly by the entities they name and how those entities relate. Writers, taught to avoid repetition, systematically sabotage this. Machines are not bored by clarity.
Say the thing
| Circling (weak) | Naming (strong) |
|---|---|
| "this innovative feature" | "Google AI Overviews" |
| "the platform" | "Google Search Console" |
| "this powerful approach" | "Generative Engine Optimisation" |
| "our solution" | "the crawl budget audit" |
The left column reads like marketing copy from 2014. The right column reads like something a retrieval system can use. Repetition of the clearest term is not a style failure; it is passage independence doing its job. Vary sentence structure instead of varying the entity's name.
Consistency across pages
Pick the canonical name for everything you discuss and use it everywhere: product names, feature names, your own company. "AISEOCourse" on the homepage and "AI SEO Course dot co" in the footer and "our learning platform" in blog posts reads as three vague entities instead of one clear one. If a genuine alias exists (an acronym people use), introduce it once in parentheses, then use one form consistently. Authors count as entities too: real bylines with consistent names and short bios give every page a person to attach to, which matters more as AI systems weigh credibility.
Headings that carry alone
Test every heading: does it still mean something quoted with nothing under it? "How AI Overviews choose sources" passes. "The important bit" fails. "More on this" fails worse. Headings are the hooks retrieval grabs first and the map scanning readers navigate by; both audiences want the same thing. Question headings are fine when the section genuinely answers that question; otherwise prefer descriptive statements, which age better and avoid promising a directness the section does not deliver.
Where structured data fits
Structured data is a clarity tool, not a citation lever: it states facts about entities in a machine-readable form, which helps classic search features and general understanding. Google is explicit that its AI features need no special schema, and nobody serious has demonstrated markup that causes citations. Use schema where it truthfully describes your page (an article's author, an organisation's site, a course that actually exists), keep it accurate, and do not expect it to substitute for naming things well in prose.
- List the page's headings alone. Mark each one pass or fail on the quoted-alone test. Fix the failures by stating the section's actual subject.
- Highlight every pronoun and generic noun ("it", "this feature", "the tool"). Replace each with the entity's name where the sentence allows.
- Check the page's key entity appears, named identically, in the title, the H1, the first paragraph and at least one heading.
Fifteen minutes per page, and it changes how the whole page reads to both audiences.
- Run the heading audit on the page you rewrote in Workbench 4.1.
- Rewrite every failing heading to state its subject plainly.
- Replace five generic references with explicit entity names.
- Write down your canonical names (brand, product, key concepts) in one list, and keep it as the site's naming reference.
Is repeating the entity's name bad style?
Does schema markup cause AI citations?
What is the quoted-alone heading test?
Sources used in this lesson
Google Search Central: AI features and your website
Google Search Central: structured data