Search engines stopped being keyword-matching machines years ago. They now try to understand things: businesses, people, places, products, and the relationships between them. That web of things is the knowledge graph, and entity SEO is the discipline of making sure your business exists in it, accurately and unambiguously. For a UK small business, this is no longer optional background theory. It decides whether AI-generated answers describe you correctly, whether Google's local features trust your details, and whether an assistant asked for "a good accountant in Sheffield" can even consider you.
What is entity SEO and how does it differ from keyword SEO?
The distinction matters because modern retrieval works in two stages. A system first resolves what the query is about, which entities are involved, then decides which documents or answers to surface. If your firm is a clearly resolved entity, every page you publish inherits that trust. If it is not, each page fights alone on keyword relevance.
Keyword work still matters. But a page titled "boiler servicing Leeds" from an unidentifiable website competes on thin evidence, while the same page from a business Google recognises as an established Leeds heating firm, with a verified profile, consistent citations and structured data, competes with its whole identity behind it. The comparison with adjacent disciplines is covered in more depth in AEO vs SEO vs GEO vs LLMO for UK small businesses.
What is Google's Knowledge Graph and why does it matter for a small business?
Those figures come from Google's 2020 post on the Knowledge Graph, which also describes knowledge panels as the boxes that summarise a topic with key facts, images and official links. The graph has only grown since, and Google has never published a comparable update, so treat five billion entities as a floor, not a current count.
The practical point for a small firm is not the scale, it is the mechanism. When Google shows your opening hours, links your reviews to the right premises, or lets an AI overview say "an independent bakery in Harrogate", it is reading graph facts, not rereading your homepage each time. Your job is to feed that graph clean, corroborated data. Most small businesses interact with it through Google Business Profile, structured data on their site, and third-party sources Google trusts, rather than through a branded panel of their own.
Why do entities matter more now that AI answers dominate search?
That 30% figure, along with the finding that 53% of UK adults say they frequently see AI summaries, comes from Ofcom's Online Nation 2025 report as summarised by Mediaworks. Whichever assistant a customer uses, the pattern is identical: the system retrieves facts about known entities and composes an answer.
The competitive backdrop is crowded. The Department for Business and Trade's Business Population Estimates count well over five million private sector businesses in the UK, the overwhelming majority of them small firms with 0 to 49 employees. An AI answer has room to name perhaps three of them for any given query. Being a well-defined entity is the qualifying round for that shortlist. The broader shift is examined in AI search vs traditional SEO for UK businesses.
How does Google decide whether your business is a recognised entity?
Think of it as a court assembling evidence. A single witness, your own website, is a claim. The same name, address, phone number and description repeated on your Companies House record, your Google Business Profile, a trade association listing and a local newspaper article is corroboration. UK businesses have a quiet advantage here: Companies House is an authoritative public register, and linking to your record from your site's structured data is free identity evidence many countries cannot match.
Confidence collapses when signals disagree. Trading under "TJ Electrical" on the website, "T J Electrical Services Ltd" on the profile and "TJ Electricals" in directories forces the machine to guess whether these are one business or three. The same applies to old addresses left on forgotten listings after a move. Ambiguity is the default state, and the graph only promotes entities it can pin down.
Which schema markup should a UK small business add first?
Schema markup is structured data, a small script that states facts in a format machines parse directly instead of inferring from prose. The order of value for a small firm is roughly: LocalBusiness (or a specific subtype such as Plumber, Dentist or Attorney) on the homepage, Person markup for the founder or named practitioners, Service markup for core offerings, and FAQ or Article markup on content pages last.
The sameAs property is the most underused line in the block. It explicitly tells search engines "this website, this company register entry and this profile are the same entity", doing the disambiguation work for them. Costs are honest here: the markup itself is free, most website builders accept a pasted JSON-LD snippet, and the real spend is an hour or two of care getting names and identifiers exactly consistent. If content production is the bottleneck rather than markup, the trade-offs are laid out in AI content writing services for UK small businesses.
How do you build entity signals beyond your own website?
A practical sequence for a UK firm. Claim and complete your Google Business Profile, since it is the single most direct feed into Google's understanding of a local business. Verify your Companies House details are current, because other data providers syndicate from it. Fix or remove stale directory listings before adding new ones, as one contradicting listing outweighs five agreeing ones. Then pursue the references only a real business can earn: chamber of commerce membership, trade association pages, supplier case studies, sponsorship mentions in local press.
Wikidata deserves a note of caution. It is a public, machine-readable database that search engines do read, and creating a well-sourced item is legitimate. But items about non-notable businesses with no independent sources get deleted, and manufactured notability tends to look exactly like what it is. Earn the coverage first, record it in Wikidata second.
How long does entity SEO take and what does it cost?
Be sceptical of anyone quoting a precise timeline, because the honest answer is that recrawling happens in days while reputation forms in months. The pattern practitioners see repeatedly: structured data changes reflect quickly in testing tools, business profile edits propagate within weeks, but the point where Google stops hedging about your brand, showing your site, profile and reviews as one coherent unit, arrives gradually and without announcement.
On money, the components themselves cost little. Schema markup is free to write, Google Business Profile is free, Companies House data is public, and most trade directories that matter are either free or priced as ordinary memberships you would hold anyway. Where firms do spend, it is on time: an owner's evenings, a staff member's Friday afternoons, or an agency retainer. That makes entity SEO unusually kind to small budgets, and unusually punishing to inconsistency, because the cheap inputs only compound if someone keeps them aligned.
How do you measure whether your entity work is paying off?
Search Console is the free backbone. Rising impressions for your brand name and "brand plus service" queries indicate the association is strengthening. The brand search itself is the qualitative check: a healthy result shows your site, your profile, your reviews and your social accounts stacked together, with no lookalike businesses interleaved.
For AI systems, run a monthly script of questions against the major assistants: "what does [business] do", "who runs it", "where is it based", "is it reputable". Log the answers. Wrong facts point at specific broken sources, and corrections you make upstream, on your site, profile or Companies House, typically flow through on the next crawl or model refresh. This measurement discipline is the unglamorous half of the work most firms skip, which is precisely why a structured AI visibility audit usually starts there.
What mistakes undo entity SEO work?
The most common error is drift. A business moves premises, changes number or rebrands, updates the website, and leaves a trail of contradicting listings behind. Every contradiction reopens the identity question the earlier work had settled. The second is aspirational markup: schema claiming awards, review scores or service areas that no independent source supports. Structured data is a claim, not a spell, and unsupported claims erode trust in the supported ones.
The third is vanity. A knowledge panel is a byproduct of being a well-evidenced entity, not a target to engineer, and time spent forcing one is better spent earning the coverage that would justify it. The last is stopping. Graph confidence decays as sources age and competitors accumulate their own evidence, so the firms that win treat entity hygiene like bookkeeping, boring, monthly and non-negotiable.
The quiet advantage of entity SEO is that it rewards exactly what small firms already have: a real identity, a real location, real customers and a public paper trail. The machines are trying to work out who you are. The whole discipline is simply answering them clearly, everywhere they look, in a voice that never contradicts itself.
Entity SEO and the Knowledge Graph for UK SMBs — FAQ
What is entity SEO in plain English?
Entity SEO is the practice of helping search engines and AI systems understand who your business is, rather than just which keywords your pages contain. An entity is a thing, a company, a person, a place, a service, that a machine can identify and connect to other things. Traditional SEO asks whether a page matches a query. Entity SEO asks whether the machine knows that Smith Plumbing Ltd is a plumbing company in Leeds, founded by a named person, with a verified address, consistent reviews and a clear set of services. When those facts are machine-readable and corroborated across the web, search engines can show your business confidently in results, map packs and AI-generated answers. The work is mostly unglamorous: structured data, consistent naming, profile hygiene and building corroborating references on independent sites.
Does a small UK business really need to care about the Knowledge Graph?
Yes, though not in the way a celebrity or FTSE company does. Most small firms will never get a branded knowledge panel with a photo carousel, and chasing one is usually the wrong goal. What matters is that the underlying graph of facts about your business is accurate and connected. Google's local results, its business profiles and its AI-generated summaries all draw on entity understanding. Ofcom's Online Nation 2025 research, as summarised by [Mediaworks](https://www.mediaworks.co.uk/insights/top-10-takeaways-ofcom-report-2025), found around 30% of Google searches now feature AI-generated overviews, and those overviews describe businesses as entities, not as keyword matches. If the machine's model of your firm is wrong, incomplete or confused with a similarly named company, you inherit that confusion in every AI answer that mentions your market.
What should I set up first if I have one afternoon?
Three jobs, in order. First, add Organization or LocalBusiness schema markup to your homepage, with your exact legal and trading name, address, phone number, and a sameAs array linking to your Companies House record, Google Business Profile and social accounts. Second, audit consistency: your name, address and phone number should be written identically on your website, Google Business Profile, and the two or three directories that matter in your trade. Third, write a genuinely descriptive About page that states in plain text what the business is, where it operates, who runs it and what it does, because both classic crawlers and AI systems lean heavily on that page when resolving an entity. None of this needs a developer beyond pasting a JSON-LD block, and all of it compounds over time.
How is entity SEO different from answer engine optimisation (AEO)?
They overlap but sit at different layers. Answer engine optimisation is about winning the answer itself: structuring content so ChatGPT, Google's AI features or Perplexity quote and recommend you when someone asks a question. Entity SEO is the layer underneath: making sure those systems know your business exists as a distinct, trustworthy thing before they would ever consider citing it. An AI assistant cannot recommend a firm it cannot confidently identify. In practice you do entity work first, establishing identity, then AEO work, earning citations and mentions. If you skip the entity layer, answer-focused content can still rank, but the assistant may attribute it vaguely or mix your firm up with a competitor. Treat entity SEO as the passport and AEO as the itinerary.
How long before entity work shows results?
Expect months, not days, and treat that as normal rather than a failure. Schema changes are typically recrawled within days or weeks, but graph-level effects, such as Google confidently associating your brand name with your services and location, build as corroborating signals accumulate across independent sources. A sensible rhythm is one setup sprint to fix markup, profiles and consistency, then monthly upkeep: new citations, fresh reviews, occasional press or association mentions. Measure progress by searching your exact brand name and watching whether Google presents your business coherently, whether AI assistants describe you accurately when asked, and whether branded impressions grow in Search Console. The absence of confusion is the early win, and richer brand results follow later. Consistency over a year beats intensity over a fortnight.
