EntityMap, entity SEO, and the file that wants to name you

Ask ChatGPT about the best kind of patio cover for Houston humidity and it may give you an answer built partly from a contractor’s own page, without ever naming the contractor. The advice survives the trip. The business does not.

That is the loss a new standard called EntityMap is built to stop, and it is worth understanding whether or not you ever publish the file.

EntityMap is new. The idea underneath it is not. Entity SEO is the older practice of getting search engines and AI tools to recognize your business as one specific thing instead of a scatter of names, pages, and listings. EntityMap is one proposal for handing that information over directly rather than hoping it gets inferred. Learning the proposal is a short way into the older practice, and the older practice is what pays off regardless of which proposal wins.

What EntityMap is

Most machine-readable files on a website describe pages. EntityMap describes entities, meaning things rather than pages: your business itself, the services it sells, the concepts it claims to know something about.

A site that adopts it publishes a file named entitymap.json at the root of its web address, written in JSON, a plain-text format built for machines to read. The specification, the written rulebook a file promises to follow, reached version 1.0 in April 2026 and defines 24 standard relationship types.

It requires the file to sit at the root of the domain, and recommends announcing it three ways so a program can find it:

• In robots.txt, the note at your site’s front door that tells visiting programs what they may read
• In a tag inside your page’s code
• As a plain link in your footer

Inside, each entity carries a name, a type, its relationships to other entities, and a sameAs link pointing at a public reference entry, usually Wikidata, an encyclopedia written for machines. That last field does the disambiguation work, telling a machine which pool company you mean out of every business sharing the name.

The problem it names is real

Two things happen to a small business in AI answers.

The first is fragmentation. Your business shows up as a legal name, a trading name, the nickname customers use, and a Google listing, with nothing telling a machine those are one company.

The second is what the standard’s authors call the ghost citation problem. An assistant reads your page, uses what it says, and credits nobody. EntityMap’s answer is to stamp the publisher’s name onto every chunk of the file, because AI systems break content apart and store the pieces separately. The specification calls that field

the mechanism by which attribution survives that extraction.

Both problems cost you the same thing. Work you already paid for ends up answering a customer’s question, and never points them back at the business that answered it.

Where the standard stands today

EntityMap is early, and that is the most useful thing to know about it.

It is published under an open license by Fred Laurent and Dixon Jones, the CTO and CEO of InLinks, an SEO software company, with their names on the specification. Standards frequently start this way, including parts of the labelling vocabulary the major search engines already agree on, schema.org.

Development is happening in public and under real scrutiny. Dan Brickley, a longtime steward of schema.org, filed an objection in May about how much of an entity description belongs in the file when it already exists elsewhere. The maintainers answered the next day and conceded part of it. A specification argued over in the open and revised inside a day is one being built seriously.

No search engine or AI company has published guidance saying it reads the file yet, and the standard’s own site tells publishers not to wait for that announcement. Adoption is scattered across the web with no register to check, so treat any precise count of adopting sites, high or low, as a guess.

What the last standard like this took

GoodRelations, a vocabulary for describing products and businesses, was started in 2001 and folded into schema.org in November 2012, and parts of it still live inside the markup ordinary sites use today. Publishers adopted it years before the search engines did.

That is the pattern worth noticing. Outside vocabularies become standards because sites used them while nobody official was reading yet, and the sites already carrying one on the day support arrives are the ones holding a head start over everybody who waited for the announcement.

Where this leaves your business

Tracking a specification through its arguments, its revisions, and whichever search or AI company announces support for it first is not your job. You have a company to run, and hours spent reading standards threads are hours not spent on the work that actually pays you.

It is our job. Watching proposals like this one, separating the two that will matter from the dozen that will not, and knowing when something crosses from interesting to worth implementing is the daily work of generative engine optimization, the craft of earning a mention from AI tools.

The groundwork underneath pays off whichever proposal wins. Name, address, phone, hours, and service list identical on your site, your Google listing, and every directory. Accurate labels in your page code, which is worth doing for reasons that have nothing to do with AI. A Wikidata entry linked if your business has one. Every one of those is what an EntityMap file would describe, and each already earns its keep in ordinary search today.

Our own site generates the machine-readable files we publish at every build, which is why adding another one is a build step here rather than a project. That is the position we want clients in as well, ready on the day the answer changes without having spent the year watching for it.

If someone has pitched you an entity map, or you want to know whether your business is ready for one, ask us and we will tell you where you stand.

Frequently asked questions

What is EntityMap?

EntityMap is a proposed open standard for publishing a machine-readable index of the things your business is about, rather than of your pages. Version 1.0 was published in April 2026 under an open license by the CEO and CTO of InLinks, an SEO software company. A site that adopts it serves a file called entitymap.json at its web address root.

What is entity SEO?

Entity SEO is the work of making search engines and AI tools treat your business as one definite thing rather than a scatter of names and pages. It means facts that agree everywhere, labels in your code that state what you are, and links to a public reference entry so machines know which business you mean.

Do AI assistants read entitymap.json?

Not automatically, as far as anyone has documented. No search engine or AI company has published guidance saying the programs it sends out to read websites go looking for the file on their own. That is a narrower statement than it sounds. An assistant handed the file, or reading it while working through your site, gets your business, your services, and the connections between them in a form built for machines rather than inferred from paragraphs, and that understanding is what a recommendation rests on.

Should a small business publish an EntityMap file?

It depends on what the file costs to produce and what shape your business facts are in. A file generated from content you already maintain is cheap to add and puts you among the sites that are ready if support arrives. A file assembled by hand from facts that disagree across your site and your listings only describes the disagreement. Sorting those facts out is the part that pays off either way.