AI BEACON · ENTITIES
What AI has filed you as,
and what it attaches to you.
A model cannot recommend a brand it has not understood. Entity mapping reads how the answers describe what your brand is, which category they place it in, and which products, markets and claims travel with your name.
Described as something you stopped selling.
Misclassification is quieter than a lost ranking, and it undermines every prompt at once.
An answer describes the brand around a product that was retired, and every prompt in that category inherits the mistake.
A similarly named business turns up in the answer's description, and prospects arrive with the wrong idea of what you do.
Each engine files you somewhere slightly different, which makes a category claim hard to argue for internally.
Collect the descriptions, then read them together.
Everything here comes from answer text, so it is what a buyer would be told rather than an inference.
Gather how answers describe you
Across your tracked prompts and every engine.
Pull out category and attributes
What the answers say the brand is, and what they attach to it.
Read the engines against each other
Agreement is a strong position. Disagreement is a job.
Fix what feeds the mistake
A stale page of your own is the most common cause, and the easiest to put right.
Correct wrong facts in answers →Your brand, as the answers hold it.
Four readings, all taken from what the answers say.
What the answers say you are.
The single most consequential line in any answer about your brand, because it decides which questions you are eligible for at all.
- In their wordsThe category each engine places you in, as written.
- Agreement across enginesWhether the answers have settled on the same description.
- Where it is too narrowA category that excludes half your market is as costly as a wrong one.
What travels with your name.
Products, integrations, markets, pricing claims and the odd invention. These are the details that turn up in answers whether or not you put them there.
- Products and featuresWhat the answers believe you offer.
- Integrations and marketsWhich tools and territories they associate with you.
- Claims about price and fitOften the most influential, and the least examined.
Where answers get you wrong.
Confusion with a similarly named company, a retired product or an old price point. Each one is traceable to something the model read.
- Mixed up with another brandRecorded where an answer blends two companies.
- Stale factsDetails that were true and are not now.
- Traced to a sourceThe page, thread or listing the mistake most likely came from.
Change what the models read.
You cannot edit a model. You can change the sources it keeps going back to, and that is the whole of the work here.
- Your own pages firstStale facts on your site are the most common and most fixable cause.
- Structured dataClear markup about the organisation helps the classification hold.
- Third-party listingsA directory or review entry that is wrong will keep being read.
Keywords describe demand. Entities describe you.
Entity work for semantic SEO and entity work for brand understanding overlap in method and differ in purpose, so they live on separate pages.
A typical stack
- –Positioning checked by reading one answer occasionally
- –A misclassification found months after it took hold
- –No record of which engine says what
- –Corrections made without knowing what fed the mistake
With entity mapping
- ✓The category each engine places you in, recorded
- ✓Attributes and claims attached to your name, listed
- ✓Confusions and stale facts logged as they appear
- ✓Each one traced back to a likely source
For entity work aimed at keyword and topic coverage rather than brand understanding, see entity mapping in Keyword Explorer.
The rest of AI Beacon, and where this one sits.
Entity mapping sits next to sentiment and diagnosis, and hands off to the pages that do the correcting.
What is entity mapping in AI answers?
Entity mapping records how AI answers define a brand: the category they place it in, the products, integrations, markets and claims they attach to its name, and the points where they get it wrong. In AI Beacon it is built from the text of answers to your tracked prompts across 7 AI engines plus Google AI Overview, AI Mode and Search, so disagreement between engines is visible, and each confusion is traced back to the source most likely to have produced it.
Key facts
- It reads how answers describe the brand, rather than inferring it
- The category each engine places you in is recorded separately
- Attributes include products, integrations, markets and claims about price and fit
- Confusions with similarly named companies and stale facts are logged
- Each mistake is traced to the page, thread or listing behind it
- Entity work for keyword and topic coverage lives in Keyword Explorer instead
Last reviewed 15 October 2026 by the SEORCE team.
On how models classify a brand.
Different question, same family of techniques. Keyword Explorer maps the entities in a topic so your content covers a subject properly. This page maps how AI answers understand your brand: what they say you are and what they attach to your name.
No. A model cannot be edited from outside. What you can change is what it reads, and the correction work usually starts with a stale page on your own site, then structured data, then any third-party listing that keeps getting quoted.
Because they read different sources. An engine leaning on a 2024 review will describe an older version of your product, and the source list for each answer shows which is doing that.
There is no reliable answer, and anyone offering one is guessing. What can be said is that OpenAI's documentation puts a robots.txt change at around 24 hours before it shows in search, and that an entity description, which is built from many sources, moves more slowly than that.
It helps and it is not a switch. Clear organisation markup gives a model something unambiguous to read, and Technical Suite validates and fixes schema at scale. Structured data coverage there is currently a small set of types and is being expanded.
Common enough to be worth watching, particularly where a name is shared with a company in an adjacent market. It is logged whenever an answer blends the two, with the answer text kept so you can see what was said.
See how AI defines your brand.
Track the questions your category asks, and read what the answers say your brand is.