Knowledge graph optimization is the work of making your brand a clear, well-documented entity that Google can recognize, store and connect in its Knowledge Graph. You do it by publishing one authoritative "entity home" page, describing your business with Organization structured data, linking that page to consistent profiles elsewhere with sameAs, and earning independent sources that confirm the same facts.
This guide is for e-commerce and Shopify brands that want Google, and the AI answers built on top of Google, to describe them accurately. You'll learn how the Knowledge Graph gets its facts, how to check whether Google already knows your brand, an eight-step optimization process, the exact Organization properties worth adding to a store, and the mistakes that keep brands invisible or confused with someone else.
What is knowledge graph optimization?
Google's Knowledge Graph is a database of facts about people, places, organizations and things, and the relationships between them. Google launched it in May 2012 with the line "things, not strings": instead of matching the letters in a query, Search tries to understand which real thing you mean. When Google is confident a query refers to your brand, it can show a knowledge panel, answer factual questions about you, and connect your brand to your products, founders and category.
Knowledge graph optimization (sometimes called entity optimization or entity SEO) is the set of actions that make that confidence easier to reach. You can't submit your brand to the Knowledge Graph directly. Google builds entries automatically from what it finds, so your job is to make the facts about your business easy to find, consistent everywhere and confirmed by sources Google trusts.
Entities, attributes and relationships
- Entity: a uniquely identifiable thing, such as your company, your founder or a flagship product line.
- Attributes: facts about the entity, such as legal name, logo, founding date, website, headquarters and social profiles.
- Relationships: connections to other entities, such as "founded by," "brand of," "sells" or "located in."
A brand with a clear name, a stable set of attributes and well-documented relationships is easy for Google to resolve. A brand that shares its name with a city, a band and a software tool, and describes itself differently on every profile, is hard to resolve, so Google hedges or shows nothing.
How Google's Knowledge Graph gets its facts
Google has explained its sources in a few public places. At launch in 2012, the Knowledge Graph contained more than 500 million objects and more than 3.5 billion facts, rooted in public sources such as Freebase, Wikipedia and the CIA World Factbook. By May 2020, Google said it had amassed over 500 billion facts about five billion entities, and that it draws from hundreds of sources across the web. Wikipedia is commonly cited, but Google says it is not the only source.
Source: Google, Introducing the Knowledge Graph (May 2012) and A reintroduction to our Knowledge Graph and knowledge panels (May 2020); Wikidata Statistics (2026)
Google's Knowledge Panel help center describes three broad inputs: facts compiled from public sources across the web, licensed data for things like sports scores, stock prices and weather, and information sent directly by content owners, for example when an official representative claims a knowledge panel and suggests changes. Because of the volume of searches, Google says it relies mainly on automated systems, with people reviewing reports and removing problematic information.
For a brand, that means the inputs you can influence are:
- Your own website, especially the page that describes the organization and the structured data on it.
- Profiles you control, such as social accounts, marketplace storefronts and your Google Business Profile if you have a physical location.
- Open knowledge bases, such as Wikidata and, if you genuinely qualify, Wikipedia.
- Independent sources, such as press coverage, industry directories, retailer pages and review sites that mention you with consistent facts.
- Direct feedback, through a claimed knowledge panel once you have one.
Wikidata matters because it is open, structured and machine readable. Its statistics page currently lists more than 123 million items. Google has never published a list of the exact sources it uses for any given entity, so treat Wikidata as one strong corroborating source rather than a guaranteed switch.
Why the Knowledge Graph matters for AI search
Entity understanding used to be mostly about knowledge panels. It now sits under the AI answers your customers read. When Google introduced AI Mode in March 2025, it described the feature as tapping into real-time sources like the Knowledge Graph, information about the real world and shopping data. In other words, the same entity data that powers a knowledge panel is an input to Google's AI answers.
That has two practical effects for an online store:
- Accuracy. If Google's picture of your brand is thin or wrong, AI answers can repeat the error: an old founder, a discontinued product line, or a competitor's details mixed in with yours.
- Eligibility. When a shopper asks for the best brands in your category, the system has to recognize your brand as a distinct entity in that category before it can recommend you. Clear category relationships (what you sell, who you sell to) make that connection explicit.
Other AI assistants build their own understanding from the open web, so the same work helps there too. Consistent facts on your site, Wikidata and trusted third-party pages are what every system has to work with. For the content side of this, see our guide on how to appear in ChatGPT, and run your pages through the free AEO checker to see how readable they are to answer engines.
We benchmark your brand against your top three competitors on Google and AI answers, then list the prioritized entity and schema fixes and the revenue math.
How to check if Google knows your brand
Before you optimize, find out where you stand. There are four quick checks.
1. Search your brand name
Search your exact brand name, then your brand name plus your category (for example "Acme candles"). A knowledge panel on the right on desktop, or an entity card near the top on mobile, means Google has an entity it is confident about. Check every fact in it.
2. Query the Knowledge Graph Search API
Google's Knowledge Graph Search API lets you look up entities by name. It returns matching entities with schema.org types, a description, an official URL and an ID in the form kg:/m/0dl567 (the example in Google's documentation). You need a free API key from Google Cloud. Google notes the API returns individual matching entities rather than the full graph, is not suitable as a production-critical service, and is being migrated to Cloud Enterprise Knowledge Graph, so use it as a diagnostic, not a dashboard.
3. Search Wikidata
Search Wikidata for your brand. If an item exists, check its statements: official website, inception date, country, founder, social accounts and industry. If it doesn't exist, decide whether you meet Wikidata's notability policy (covered below).
4. Ask the AI answers
Ask Google AI Mode, ChatGPT and Perplexity "What is [brand]?" and "Who makes [your flagship product]?" Note what they get right, what they get wrong and which sources they cite. Those sources are the pages to correct or strengthen first.
How to optimize for the Knowledge Graph: 8 steps
Work through these in order. The early steps are on your own site and cost nothing but time. The later steps depend on the outside world confirming what you say.
- Pick one canonical name and description. Decide on your exact brand name, legal name (if different), a one-sentence description and a short "what we sell" line. Use them word for word everywhere.
- Build an entity home page. Use your About page (or homepage) as the single page that best describes the organization: who you are, what you sell, when and where you were founded, who runs it, and links to every official profile. Google recommends placing Organization markup on the home page or a single page describing the organization.
- Add Organization structured data. Mark up that page with JSON-LD using the most specific type. For stores, Google recommends the
OnlineStoresubtype. Include name, url, logo, description, sameAs and identifiers (see the table in the next section). Use our schema markup generator to draft it and the schema validator to check it. - Connect your profiles with sameAs. List the URLs of official profiles: Instagram, TikTok, YouTube, LinkedIn, Facebook, Pinterest, your Amazon storefront and your Wikidata item if one exists. Then link back from each profile to your site, so the connection runs both ways.
- Make every profile consistent. Use the same name, logo, description, founding year and website on every profile. Fix old bios, retired logos and outdated addresses. Inconsistency is the most common reason Google hedges.
- Create or improve your Wikidata item, if you qualify. Add statements with references to independent sources: instance of (business or brand), official website, inception, country, founder, industry and social media IDs. Only add facts you can source.
- Earn independent corroboration. Press coverage, retailer and stockist pages, trade association listings, award pages and podcast interviews all confirm your attributes from the outside. Ask partners to use your exact name and link to your site. Our guide to unlinked brand mentions shows how to find and convert existing mentions.
- Claim your knowledge panel and maintain it. Once a panel appears, claim it through Google's verification flow, suggest corrections with supporting URLs, and review every fact each quarter. Our Google knowledge panel guide covers claiming and editing step by step.
The table below ranks the signals by how much control you have and how directly they describe your entity.
| Signal | Where it lives | What it tells Google | Your control |
|---|---|---|---|
| Entity home page | Your About page or homepage | The official version of who you are | Full |
| Organization schema | JSON-LD on that page | Machine-readable name, logo, identifiers and profiles | Full |
| sameAs profiles | Social, marketplace and review profiles | Which accounts belong to the same entity | High |
| Wikidata item | wikidata.org | Structured, referenced facts in an open knowledge base | Medium (must meet policy) |
| Press and partner mentions | Third-party sites | Independent confirmation of your attributes | Low to medium |
| Claimed knowledge panel | Google Search | Direct feedback from a verified representative | Medium (after a panel exists) |
Organization schema for Shopify stores
Google's Organization documentation has no required properties. It recommends adding every property that applies, and notes that some help Google distinguish your organization from others while others can influence visual elements such as the logo in Search results and the knowledge panel. For e-commerce, Google recommends the OnlineStore subtype instead of the generic OnlineBusiness.
| Property | What to put in it | Why it helps |
|---|---|---|
name, alternateName, legalName | Brand name, common variants, registered company name | Matches your brand to search queries and to company records |
url, logo | Homepage URL; a logo at least 112x112px that works on a white background | Google uses the logo in Search results and knowledge panels |
description, foundingDate | Your one-sentence description; founding date in ISO 8601 | Stable attributes that should match every profile |
sameAs | URLs of every official profile and your Wikidata item | Ties separate accounts to one entity |
iso6523Code, vatID, taxID, naics | Identifiers such as a DUNS (prefix 0060), GLN (0088) or LEI (0199) | Unambiguous IDs that separate you from same-name companies |
hasMerchantReturnPolicy, hasShippingService | Store-wide return and shipping policies | Feeds the merchant details Google can show for stores |
contactPoint, address | Customer service contact; physical or mailing address | Confirms you are a real, reachable business |
Google calls a VAT ID "an important trust signal for users" because people can look businesses up in public VAT registries. If you have a DUNS number or LEI, Google prefers that you supply it through iso6523Code rather than the older duns and leiCode properties.
What Shopify does and doesn't do for you
Many Shopify themes output some Organization or WebSite markup, often just a name, logo and social links pulled from theme settings. That's a start, not a finished entity. Test your homepage and About page with the schema validator, then add the missing properties in a theme snippet or through a structured data app. Avoid duplicate, conflicting Organization blocks from a theme and an app at the same time. Our guides to Organization schema markup and schema markup for e-commerce include full examples.
Products, the Shopping Graph and your brand
Stores have a second graph to think about. Google's Shopping Graph is its dataset of products, sellers, prices, reviews and inventory, and it has grown quickly. Google said it held more than 24 billion listings in May 2021, 35 billion in February 2023 and more than 50 billion in May 2025, with more than 2 billion listings refreshed every hour.
View the data
| Value | |
|---|---|
| 2021 | 24 |
| 2023 | 35 |
| 2025 | 50 |
Source: Google, The Keyword blog: Shopping posts from May 2021, February 2023 and May 2025
According to Google, the Shopping Graph is fed by information retailers share directly through Google Merchant Center and by what retailers and brands post across the web. For a brand, the goal is that every product listing, on your site and on every retailer that stocks you, points back to the same brand entity.
- Use one brand value. Set the same
brandname in your Product markup and your Merchant Center feed that you use in your Organization markup. - Supply GTINs. Global Trade Item Numbers let Google match your product across retailers, so reviews and offers roll up to one product and one brand.
- Keep store policies consistent. Shipping and return policies in Organization markup, Merchant Center and your policy pages should say the same thing.
- Give retailers your facts. Send stockists your exact brand name, logo and description so their product pages reinforce, not dilute, your entity.
Our guide to product schema on Shopify covers the product side in detail.
Our free $100M AI Visibility Roadmap, built from 7,960 live pages and 197 A/B tests, shows how to get cited by ChatGPT, Google AI Overviews and Perplexity.
Wikidata and Wikipedia: what's realistic
Wikidata and Wikipedia are powerful corroborating sources, but they have rules, and breaking them can backfire.
Wikidata's notability policy accepts an item if it meets at least one of three criteria: it has a valid sitelink to a Wikimedia page such as a Wikipedia article; it refers to a clearly identifiable entity that can be described using serious and publicly available references; or it fulfills a structural need, for example to make statements in other items more useful. Many established brands meet the second criterion through press coverage and public business records. A brand new store with no independent coverage usually doesn't yet.
Wikipedia has a higher bar: significant coverage in reliable, independent sources. Its conflict of interest guideline strongly discourages writing about your own company. If you don't have that coverage, put the effort into earning it, not into an article that editors will delete.
| Source | Bar to get in | Best for |
|---|---|---|
| Your entity home and schema | None | Every brand, from day one |
| Wikidata | Clearly identifiable entity with serious, public references | Brands with some press or public business records |
| Wikipedia | Significant coverage in reliable, independent sources | Brands with sustained national or trade press |
| Industry directories and press | Varies by publication | Building the independent coverage the other sources need |
Common mistakes that confuse Google
Most entity problems come from contradictions, not from missing a clever tactic. These are the ones we see most often on e-commerce sites.
| Symptom | Likely cause | Fix |
|---|---|---|
| No knowledge panel for your brand name | Too few consistent sources, or a generic name shared with other entities | Build the entity home, add identifiers and sameAs, earn independent mentions |
| Panel shows the wrong logo or old details | Outdated profiles or logo markup still pointing at an old file | Update logo markup and profiles, then claim the panel and suggest edits |
| Google mixes you up with another company | Same name, no disambiguating facts | Add legalName, identifiers, founding date, location and category; use them everywhere |
| AI answers describe a discontinued product line | Old pages and retailer listings still live | Redirect or update old pages and ask retailers to refresh listings |
| Validator shows two conflicting Organization blocks | Theme markup plus an app or custom snippet | Keep one complete Organization block on the entity home |
One more mistake: rebranding without a plan. If you change your name, update the entity home, schema (keep the old name in alternateName), every profile and your Wikidata item in the same week, and redirect old URLs. A half-finished rebrand leaves two competing entities.
Where to start this week
You don't need a big project to make progress. In one week you can settle your canonical name and description, turn your About page into a real entity home, add complete OnlineStore markup with sameAs links and identifiers, and fix the three most visible inconsistent profiles. Then check your brand in the Knowledge Graph Search API and in AI answers, note what's wrong, and schedule a quarterly review.
The slower work, earning press and partner mentions and qualifying for Wikidata, is what turns a well-described brand into one Google is confident about. If you want help mapping how Google and AI answers see your brand today against your competitors, talk to our team or see our AI search optimization service.
Book a free 30-minute call. We'll benchmark you against your top three competitors on Google and AI answers and show the fixes to make first.