You marked up your site with schema. You told search engines exactly who your brand is, what you sell and what you know about.
That does not mean Google agrees.
The gap between the entities you declare and the entities Google’s natural language processing actually recognizes is where a lot of content strategies quietly leak visibility. Search Engine Land is putting a spotlight on it in the next SMX Now session on Sept. 16 at 1 p.m. ET, featuring Ray Martinez, VP of SEO at Archer Education.
What an entity gap actually is
Search engines do not read your pages as strings of keywords anymore. They map things: people, products, places, concepts, and the relationships between them.
Your structured data is your claim. Google’s knowledge systems are the verdict. If Google’s NLP does not associate your brand with the entities you care about, you can rank for long-tail queries all day and still be invisible in the places that matter most now: knowledge panels, AI Overviews, and citations inside AI assistants.
Martinez’s session covers how to measure that difference instead of guessing at it.
The audit stack he’ll walk through
According to the session preview, the workflow combines three things:
- Schema.org markup from your existing site, as the source of what you claim to be
- The Google Cloud Natural Language API, to see which entities machines actually extract from your content
- An agentic coding tool such as Antigravity, Claude Code or Codex, to run and repeat the process at scale
The output turns your schema into a queryable knowledge graph. You then compare that graph against competitor content to surface three things: topics rivals cover, gaps everyone is missing, and entities Google has not yet connected to your brand.
That third category is the interesting one. It is the list of things you believe you are known for but demonstrably are not.
Why this matters right now
Traditional gap analysis has been keyword-shaped for a decade. You export competitor rankings, filter for terms you do not have, brief a writer, move on.
That approach breaks when the retrieval layer is an LLM. AI systems assemble answers by pulling entities and relationships, not by matching exact-match phrases. If your brand is not a recognized node connected to the concepts in a query, you are not in the consideration set at all.
An entity audit reframes the question from “what keywords are we missing?” to “what does the machine think we are an authority on?” For agencies, that is a client deliverable that no one else on the pitch list is bringing yet.
Turning findings into content
The session also covers the action side. The playbook Martinez outlines: build content around under-recognized entities, reinforce those connections with structured data and your own internal data, and make pages more retrievable and citable across both classic search and AI engines.
A practical way to sequence it:
- Claim — extract every entity your schema and copy assert
- Verify — run pages through NLP entity extraction and score confidence and salience
- Compare — benchmark against two or three direct competitors
- Close — publish depth on the weak entities, then link internally so the relationship is obvious
- Track — rerun the audit quarterly to see if machine understanding improves
That last step is the real prize. Most AI-visibility work today is anecdotal screenshots. A repeatable entity audit gives you a measurable baseline for discoverability that you can put in a board deck.
The takeaway
If your SEO reporting still ends at rankings and sessions, you are measuring the old surface. Start measuring whether machines understand what your brand does. The teams that build that muscle first will own the citations everyone else is fighting over in 2026.
Source: Search Engine Land
