Discussion starter: an entity SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Entity SEO
Wikidata, Crunchbase, LinkedIn and GBP disagree about your company. Which source has Google seemed to trust most?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
No single source wins, and Google doesn’t publish a ranking of them. Which one Google appears to follow changes with the fact and the surface. For local details in Maps and the local panel (name, address, phone, hours, category), your Business Profile is the input Google hosts and you verify, so fix it first. For entity facts in a knowledge panel (founding date, founders, headquarters, description), Google blends sources. The useful question there is not which source Google trusts most, but which source carries the wrong value you are seeing.
Google’s own description supports that framing. Its Knowledge Panel Help says facts in the Knowledge Graph “come from a variety of sources that compile factual information”: public sources, licensed data (its examples are sports scores, stock prices and weather forecasts) and content owners, including those who claim their knowledge panels. The page doesn’t name Wikidata, Crunchbase or LinkedIn. Any claim that one of them is Google’s most trusted source is practitioner inference, not documented behaviour.
So the practical answer is a method rather than a name: find which surface shows the wrong value, trace which source carries it, fix the upstream copy, and make your own site the reference every other source can be checked against.
Different Google surfaces read different inputs
Where the fact appears Typical facts What feeds it Fix first Local panel and Maps Name, address, phone, hours, category Your Business Profile, plus edits Google can apply from user suggestions The Business Profile, then the name, address and phone on your site Knowledge panel for the organisation Description, founded, founders, headquarters, logo, social profiles The Knowledge Graph: public sources, licensed data and content owners. When the description comes from Wikipedia, the panel shows the attribution. Whichever public source carries the wrong value, then panel feedback AI Overviews and AI Mode Answers to “who founded X” or “how many people work at X” Pages retrieved for that query, which can include your About page, LinkedIn, Crunchbase, press coverage and directories Every page that ranks for those questions, starting with your own Organic results for brand queries Snippets from your site and your profile pages Normal ranking and snippet selection The profiles you control, and your About page This is how the same company can show one headquarters in its local panel and another in an AI answer on the same day. The surfaces are not disagreeing about which source to trust; they are reading different inputs.
Where each of the four sources fits
Google Business Profile
Google’s own product, verified by the owner, and the main input for local surfaces. It suits businesses that meet customers in person or serve an area, so an online-only company may not qualify. Google can apply changes suggested by users, so review the profile for edits you didn’t make. The business name should match the name you trade under in the real world, which can differ from the legal name held elsewhere.
Wikidata
An open, structured database anyone can edit. Google shut down its own Freebase database in the mid-2010s and supported moving that data to Wikidata, which is why practitioners watch it closely. Changes need references, and an item has to meet Wikidata’s notability policy to stay. Treat it as a public reference, not a profile you own.
Crunchbase
A company database whose profiles companies can claim and edit. Its data is also licensed to other platforms, so an error can travel into sales databases and directories that then look like independent confirmation. Its company pages can rank for brand queries about funding and founders, which makes them likely pages for AI answers to retrieve.
LinkedIn
A self-managed company page. Company size is a range the page admin selects, while the employee number LinkedIn displays is based on members who list the company, so the two can disagree with each other and with your real headcount. Headquarters and founding year are admin fields, so this one is quick to correct.
Make your own site the reference point
None of the four is fully under your control. Your website is. Google’s Organization structured data documentation says adding the markup to your home page “can help Google better understand your organization’s administrative details and disambiguate your organization in search results”. The recommended properties cover many of the facts in dispute:
legalName,alternateName,foundingDate,numberOfEmployees,address, identifiers such asvatID,leiCodeandduns, andsameAslinks to your profiles. Google recommends placing it on the home page or a single page that describes the organisation, such as the about page, and notes that “Google does not guarantee that features that consume structured data will show up in search results”.Markup is one input, not an override. The plain-text About page does a different job: founding year, founders and current leadership, headquarters, headcount range, registered company name and number. That is the page Wikidata editors, journalists, data vendors and AI answers can cite, and the page you point to when you ask any other source for a correction.
Trace the wrong value before you fix anything
- List the disputed facts. For each one (legal name, trading name, founded, founders, CEO, headquarters, headcount, logo, description, social profiles), record the value in every source. A sheet with one row per fact and one column per source makes the pattern visible.
- Record what Google shows and where. Check the knowledge panel, the local panel, and AI answers to questions like “[brand] founder” or “[brand] headquarters”. Note any attribution link.
- Search the wrong value with the brand name. If the panel says you were founded in 2011, search the brand name with “2011” and see which pages carry it. A value that appears in one upstream source and in directories that copy it points to the source to fix first.
- Check the entity record. The Knowledge Graph Search API returns the entity ID, name, description and, where one exists, a detailed description with its source URL (Google’s example uses a Wikipedia article). As of September 2026 it is still documented, with a notice that Google is migrating it to Cloud Enterprise Knowledge Graph and a warning that it is “not suitable for use as a production-critical service”. If you cannot find an entity for your brand, the problem may be different: Google may not have established the company as an entity yet.
- Claim the knowledge panel if one exists, and suggest changes. Google lists content owners who claim panels among its sources of facts, and says it may remove information that’s demonstrably false or outdated, as shown by legal documents, expert consensus or other reputable primary sources.
The order I would fix them in
1. Agree the canonical facts internally: legal and trading names, what “founded” means (incorporation or launch), headquarters, headcount range↓2. Publish them on your About page and in Organization markup, withsameAslinks to every profile↓3. Correct the sources you control directly: Business Profile, LinkedIn, Crunchbase↓4. Correct Wikidata with references. For facts that changed (CEO, headquarters), add the new value with a start date and preferred rank, and give the old one an end date. For values that were never right, add the correct value and deprecate the wrong one with a reason rather than deleting it.↓5. Suggest changes through the claimed knowledge panel, citing the corrected public sources↓6. Recheck the panels, AI answers and the entity record monthlyThe order matters because each step gives the next one something to cite. Wikidata edits and panel feedback carry more weight with a public source behind them, and your About page is the one source you can update today. Corrections don’t show up on a published schedule, so recheck monthly rather than daily.
Hypothetical example:
A software company moved its headquarters from Melbourne to Sydney in 2023 and changed CEO in 2025. Wikidata still shows Melbourne and the former CEO, Crunchbase has the new CEO but the old headquarters, LinkedIn shows Sydney, and the knowledge panel shows Melbourne and the former CEO. That doesn’t prove the panel reads Wikidata, but Wikidata is the only source matching both wrong values, so it is the best lead. The team updates the About page and markup, adds Sydney and the new CEO to Wikidata with start dates, references and preferred rank, fixes the headquarters on Crunchbase, and suggests the change through the claimed panel, citing the About page and the company’s registry record.
What the field evidence does and doesn’t show
Practitioners report knowledge panels changing after Wikidata or Wikipedia corrections, but those reports are anecdotal, timings vary, and they don’t reveal a consistent hierarchy between sources. The more defensible reading is that Google looks for agreement: a fact stated consistently by several independent sources, including the company itself, is harder to displace than a fact only one database carries. That is an inference from how Google describes the Knowledge Graph, not a documented rule. It does have a practical advantage: work that builds agreement helps whichever source Google ends up using.
Watch for this:
Unreferenced edits that make Wikidata match your preferred story risk being reverted, and they can draw attention to an item that barely meets notability. Treat every edit as something a sceptical editor will check: add a reference, keep the history, and be open about your connection to the company.
If I had to pick a starting point with no other information: the Business Profile for anything local, and whichever source matches the wrong value Google shows for everything else. If no source matches, the job is consistency rather than a single fix: agree the facts internally, publish them on your site, align every profile to them, and give Google’s feedback channels a public source to check against.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.