Knowledge panels for service businesses: a 2026 playbook.

A knowledge panel is the boxed information card that appears on the right side of a Google search result when the query is about an entity Google understands. Search "Anthropic" and you get the company panel. Search a well-known doctor or author and you get a person panel. Search a small business and, in most cases, you get nothing — because Google hasn't yet decided that the business is a distinct enough entity to warrant one.

For a long time, knowledge panels were a vanity metric — nice to have, not strategically critical. That stopped being true around 2023, and is decisively no longer true in 2026. The reason: knowledge panels are one of the cleanest signals to AI search engines that an entity is real, distinct, and worth surfacing in generative answers. ChatGPT, Claude, Gemini and Perplexity weight entity confidence heavily when they decide what to recommend. Businesses with knowledge panels are surfaced in AI answers. Businesses without them, even with strong organic SEO, frequently aren't.

This playbook is the sequence I've used to build knowledge panels for service businesses across all four agencies.

1. The entity foundation

Google's Knowledge Graph is built on entities — distinct, identifiable things in the world that have a stable identity, a clear set of attributes, and corroboration across multiple sources. Building a knowledge panel is the process of getting your business recognised as one of those entities.

The foundation is a clear, consistent identity surfaced in the same way across every place the business appears online. Same legal name. Same trading name. Same address. Same phone number. Same founding year. Same founders. Same category. Inconsistencies — a different abbreviation here, a different address format there, a missing apartment number on one citation — are what stop the entity from forming cleanly in the Knowledge Graph.

2. Schema markup, done properly

Schema.org markup is the structured-data layer that tells Google explicitly what the entity is. For a service business, the relevant types are usually LocalBusiness (or one of its more specific subtypes: MedicalBusiness, Plumber, LegalService), Organization, and Person for individual founders or practitioners.

The fields that matter most for entity recognition:

• name, legalName, alternateName — the exact identity
• url, logo, image — the canonical visual identity
• address, telephone, geo — the physical identity
• founder, foundingDate, numberOfEmployees — the institutional identity
• sameAs — an array of canonical URLs for the entity across other authoritative sources. This is the single most important field, and most businesses leave it empty.

3. The sameAs strategy

The sameAs array is how you tell Google "this entity is also represented at all of these other places, and they all corroborate the identity I'm asserting." For a knowledge panel to form, Google needs to see a meaningful number of authoritative sources independently confirming the entity exists.

For a typical service business, the high-leverage sameAs targets are:

• The business's verified Google Business Profile (every business should have this; many don't claim it properly)
• Industry-specific directories that Google trusts (for healthcare in Australia: HealthEngine, HotDoc, the relevant AHPRA register; for trades: the relevant industry association)
• Wikidata — the structured-data sibling of Wikipedia. A clean, accurate Wikidata entry is one of the fastest ways to lift entity confidence
• LinkedIn company page (verified)
• Crunchbase, when applicable
• Industry publication mentions, when the business has been credibly covered

4. Wikidata is the lever most operators ignore

Wikidata is the open-source knowledge base that feeds a significant portion of what Google's Knowledge Graph knows. Unlike Wikipedia, Wikidata accepts entries for organisations and people that don't meet Wikipedia's stricter notability bar. Done correctly, a Wikidata entry can lift an entity from "Google has heard of this business" to "Google has a clean, structured record it can confidently surface."

The trick is doing it correctly. Wikidata entries that get rejected or reverted usually fail because they cite only the entity's own website rather than independent sources. The fix is to build the Wikidata entry around what other sources say about the entity — third-party industry directories, news mentions, regulatory registers, verified profiles on other platforms.

5. Corroboration through independent sources

Beyond the structured-data layer, knowledge panels form when independent sources corroborate the entity's identity in language that's consistent with what the entity is asserting about itself.

This is where most "knowledge panel campaigns" stall, because it's where the work gets harder than copy-pasting schema markup. The corroboration needs to be:

• Independent — not the entity's own website, not pages the entity controls
• Structured — written in a way that mirrors the entity's identity (same name, same role, same category)
• Authoritative — sources Google already trusts
• Reasonably distributed — multiple sources, not just one; ideally across categories (a directory, a news mention, an industry association, a profile on a partner platform)

6. The AI search bonus

Once the knowledge panel forms, the secondary effect — the part that's underweighted in most SEO playbooks — is the lift in AI-search recommendation. ChatGPT, Claude, Gemini and Perplexity all draw on the same underlying entity confidence signals that Google uses, and a service business with a knowledge panel is meaningfully more likely to be named when those models answer "who's the best [category] in [city]" or similar high-intent queries.

For most service businesses in 2026, AI-search referrals are still a small fraction of total traffic — but the fraction is growing fast, and it's growing especially fast at the high-intent end of the funnel where the user is asking for a specific recommendation rather than browsing options.

What to expect, and how long it takes

For a clean entity foundation that Google has the right signals for, knowledge panels typically form three to nine months after the supporting infrastructure is in place. For an entity with a confused identity history — multiple business names, inconsistent citations, weak third-party presence — it can take twelve to eighteen months and require active cleanup of the old confusion before the new identity stabilises.

Knowledge panels in 2026 are less about ranking better in Google and more about being recommendable inside the AI search engines that are quietly absorbing the high-intent end of the funnel.

This is the work I'd do for any service business serious about being findable in the next five years — not as a one-off SEO project, but as the structural foundation that everything else compounds on top of.