Entity and brand authority building

What an entity is, in plain terms

An entity is a thing a system holds a record about, as opposed to a string of characters it matches. Your business name typed into a search box is a string. Your business understood as a specific organisation, founded in a particular year, based in a particular place, offering particular services, run by particular people, is an entity.

The difference shows up when a system has to make a judgement. If your brand is only a string, every mention of that string is ambiguous, and a system encountering a similar name has no way to tell whether it is you. If your brand is a recognised entity, mentions attach to a record, and each one reinforces the same understanding rather than scattering into noise.

This matters more now than it did, because AI answers require a system to decide whether it is safe to name a specific business in response to a question. A well-defined entity with corroborated facts is safe to name. A vague one is not, and the system hedges or picks someone else.

• A string is text to be matched, an entity is a record to be reasoned about
• Entity recognition lets scattered mentions accumulate into one understanding
• Ambiguity between similarly named businesses is resolved by entity signals
• Systems name entities they can verify and avoid ones they cannot

Consistency of name, description and contact facts

The unglamorous foundation is that your core facts must be identical everywhere they appear. One legal name and one trading name, used the same way. One address format. One phone number. One short description of what the business does, repeated verbatim rather than rewritten for each profile.

Most businesses fail this without realising. The company registration says one thing, the website footer another, the directory listings a third with an old suite number, and six social profiles carry six differently worded descriptions written by different people over eight years. Every inconsistency is a small reason for a system to hesitate.

We resolve it by agreeing one canonical record, writing it down, and then propagating it. That record covers the names, the address, the contact details, the founding date, the service list and the description at two or three lengths. From then on nothing gets written fresh. It gets copied from the record, which is the only way consistency survives staff changes.

• One canonical record covering names, address, contacts, founding date and services
• A description written once at short, medium and long lengths, then reused verbatim
• Legal name and trading name used consistently and distinguished where needed
• Every existing profile audited against the record and corrected
• The record documented, so future updates go to one place

The entity home concept

An entity home is the single page you designate as the authoritative description of the organisation, and that every other source points back to. In practice this is a properly built about page, not a paragraph of founder sentiment but a factual page that states what the business is, when it started, where it operates, who runs it, what it does, and how to contact it.

It works because it gives systems one place to resolve to. Your Organization markup lives there or references it. Your sameAs links out to every verified profile live there. Your social profiles and directory listings link back to it. What you end up with is a hub, and hubs are easy for a system to identify as the canonical source when facts elsewhere disagree.

The page has to be genuinely informative to do this job. An about page that says nothing checkable gives a system nothing to anchor on. Named people, real dates, specific locations, a clear service list and verifiable contact detail are what make it usable.

• One designated page as the factual description of the organisation
• Organization markup with sameAs links to every verified profile
• Inbound links from your profiles back to that page
• Named people with real roles, real dates and specific locations
• Checkable facts rather than positioning language

Wikidata and the profile estate

Wikidata is an open, structured database of entities that many systems read. Unlike an encyclopaedia article it holds statements rather than prose, and its inclusion criteria are about whether an item is a valid, identifiable thing with references, not about whether it is notable enough for an article. Where a business genuinely qualifies and can be referenced, a correct Wikidata item is a useful, durable anchor.

Around it sits the profile estate: the set of platforms that hold a record about your organisation. Company registries, industry bodies, review platforms, developer and business directories, professional networks, and the specific platforms your sector uses. Each verified profile carrying the same facts and linking back to the entity home is another piece of corroboration.

The aim is coverage of the places that matter to your sector rather than volume. Fifty abandoned directory profiles with mismatched details are worse than eight accurate ones on platforms anyone actually consults.

• A correctly referenced Wikidata item where the organisation genuinely qualifies
• Verified profiles on the platforms relevant to your sector
• Identical facts on every profile, linking back to the entity home
• Coverage judged by relevance, not by count
• Dead and duplicate profiles claimed and corrected or removed

Why we do not do Wikipedia article creation

We are asked for this often and we decline it, for two reasons that are worth stating plainly rather than hiding behind vagueness.

The first is disclosure. Wikipedia terms of use require anyone paid to contribute to disclose the employer, the client and the affiliation. Agencies that quietly write client articles are breaking those terms, and the practice is treated as a serious violation when it is discovered, which it frequently is. We are not willing to operate a service that depends on either breaking the rules or on the disclosure being ignored.

The second is that it usually does not work even when done honestly. Articles about businesses that do not meet the notability standard get nominated for deletion, and the deletion discussion becomes a permanent, publicly searchable record of a company trying to buy an encyclopaedia entry. The downside is worse than the upside, and the upside was speculative to begin with. If your organisation becomes genuinely notable, an article will tend to appear without you commissioning one. Until then, Wikidata, your entity home and a corroborated profile estate do the useful part of the job without the risk.

Why third-party corroboration beats self-description

You can write anything about yourself. Systems know that, and they weigh sources accordingly. A claim that only exists on your own site is an assertion. The same claim appearing in a trade publication, an industry body listing, a conference programme, a customer review platform and a partner site is a fact with multiple independent witnesses.

The evidence points the same way. A 2025 study found AI search systems systematically favour third-party authoritative sources over brand-owned pages and social content. That is a large part of why we sell digital PR, and why we do not promise that on-site work alone will get you named in AI answers. On-site work makes you legible. Off-site coverage makes you credible.

The practical programme therefore runs on both sides at once. Fix the canonical facts and the entity home so there is something clear to corroborate. Then earn mentions in the places your sector treats as authoritative, so the corroboration exists. Neither half does the job alone, and the second half is the slower and more valuable of the two.

• Independent sources carry more weight than your own description of yourself
• Coverage in places your sector already trusts, rather than generic placements
• On-site clarity and off-site corroboration built together, not sequentially
• No fabricated mentions, paid schemes or manufactured coverage, ever
• Progress measured by how consistently third parties describe you correctly