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
The knowledge panel shows the wrong founder or headquarters. Which sources did Google update from after you fixed them, and how long did it take?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
Google doesn’t document which source feeds a given fact in a knowledge panel, and it publishes no refresh schedule for the upstream sources it reads. So any answer that names “the” source Google updated from is an inference from one case, and the timings people report after fixing Wikipedia, Wikidata or their own site are anecdotes. The one documented timeline is for feedback: Google says it reviews feedback from verified users within a few days and emails a resolution, though this can sometimes take longer, and it may not make a change it can’t confirm.
That shapes the order of work. Fix the public sources first, because they are what a reviewer can check, then submit the change from the verified (claimed) panel with links to them. If you also want to learn which source Google follows, stagger the fixes and log dates, accepting that this keeps the wrong fact live for longer. For founders in particular, adjust expectations about markup:
founderis not among the Organization properties Google documents, whilefoundingDateandaddressare.First confirm what is wrong, and where
- Knowledge panel or local panel? If the panel shows a map, opening hours and reviews, it is built from your Business Profile, and an address error gets fixed there, not in Wikidata.
- The right entity? The panel’s share link typically resolves to a search URL with a
kgmidparameter, the entity’s ID. If the wrong founder or city belongs to a similarly named company, a parent or a subsidiary, you have an entity confusion problem, not a data error, and editing values won’t fix it. - The wrong person, or the right name linked to the wrong person? Founder names in a panel can be links. If clicking one opens a different person’s panel, Google has joined the right name to the wrong person entity.
Where wrong founders and headquarters come from
What the panel shows Typical origin Where to look A co-founder missing, or an early investor or first CEO listed as founder A Wikipedia infobox, the Wikidata “founded by” statement (P112), a funding-round article, a database’s people list Wikipedia and Wikidata edit histories; early press coverage The registered office shown as headquarters Company registry data copied into business databases and directories The registry record and the directories that repeat it The old city after a move A stale infobox or Wikidata “headquarters location” (P159), old press releases, directories Search the brand name with the old city and see which pages carry it A parent’s or subsidiary’s city Entities merged or confused The kgmid, and whether the description and logo also belong to the other company A different person with the founder’s name Person entity mismatch Click the founder’s name in the panel A quick way to find the carrier is to search the wrong value next to the brand name. If one upstream source and several copies of it show the same wrong value, fix the upstream one first; the copies can follow.
How to learn which source Google updated from
Because nothing is documented per fact, the evidence has to be generated. A panel change after an edit is consistent with that source being used, not proof of it: Google may have refreshed from several sources at once, or the feedback review may have made the change. Sequencing and fingerprints give stronger evidence than timing alone.
- Baseline. Screenshot the panel (logged out, in the target country, on desktop and mobile), record the kgmid, and save the answers AI Overviews or AI Mode give to “who founded [brand]” and “where is [brand] headquartered”.
- Inventory. List every source that carries the fact, its current value, and whether it keeps a public edit history. Wikipedia and Wikidata do, with timestamps. Your own site has a deploy log. Business databases may not, so record the date you changed them.
- Sequence. Where you can afford to wait, change one source at a time with a gap of a few weeks between changes, and record the date each went live. The gap matters because refresh timing is unknown; without it, two changes can’t be told apart.
- Fingerprints. Note how each source formats the value: “Perth, Western Australia” or “Perth, WA”, founders with or without a middle initial, the order of co-founders. If the panel adopts one source’s formatting, that points to the source more reliably than timing does. Use differences that already exist. Planting markers in Wikipedia or Wikidata to trace Google would misuse a community resource and invite reverts.
- Recheck on a schedule. Weekly is enough. Log every observed change with its date.
- Submit feedback last. From the verified panel, describe what is wrong and include publicly accessible URLs that support the change, which is what Google’s feedback page asks for.
The trade-off:
Sequencing is how you learn which source Google follows, and it keeps a wrong fact live for longer. If the error does real harm (a disputed founder, a headquarters that matters for regulation or tenders), fix everything at once and accept that you won’t know which change did it.
What each fix involves
Your own site
State founders, founding date and headquarters in plain text on the About page, and list a registered office separately if it differs. In Organization markup, use
foundingDateand anaddressthat matches the headquarters: Google describesaddressas the physical or mailing address, if applicable, so a registered-agent mailing address there can contradict the headquarters you want shown. Adding schema.org’sfounderdoes no harm, but Google’s documentation doesn’t list it.Wikipedia
The conflict-of-interest guideline very strongly discourages paid editors from editing affected articles directly, requires them to disclose employer, client and affiliation, and asks for changes to be proposed on the article’s talk page through an edit request. The direct edits it allows are narrow (spelling and grammar, reverting unambiguous vandalism, adding reliable sources), so a founder correction goes through the talk page. The guideline stresses independent sources, so bring press coverage or official records, not only your About page.
Wikidata
Edit with references. For a fact that changed, such as a headquarters move, add the new value with a start date and preferred rank and give the old one an end date. For a value that was never right but still appears in a published source, add the correct one and deprecate the wrong one with a reason, rather than deleting it.
Databases and registries
Correct the profiles you control, such as Crunchbase and LinkedIn. A registry’s registered office is a legal record you can’t change without moving the office, so explain the difference on your About page instead. Directories that copied an old value update on their own schedule, if at all.
What to expect on timing
Route What Google documents What to plan for Feedback from the verified panel Reviewed within a few days, with an email on the outcome; can sometimes take longer The only timeline you can put in a plan Feedback from anyone, through the Feedback link in results Anyone can submit; no timeline given Extra signal, not something to wait on Your own site Nothing specific to knowledge panels Request indexing of the About page through URL Inspection; when the panel reflects it is unknown Wikipedia and Wikidata Nothing; edit requests wait for volunteer editors Editor response time first, then Google’s refresh Business databases and directories Nothing Copies can outlive the fix at the source For stakeholders, I would put it this way: days for the verified review once the evidence is public, an unknown tail for everything else, and no guarantee. Recheck weekly for the first month, then monthly. If feedback is declined, the reasons Google lists are inappropriate content or accuracy it can’t confirm, so for a founder or headquarters correction the answer is stronger public evidence, not the same request again.
AI answers need separate work. AI Overviews and AI Mode also draw on the web pages they retrieve, so a panel fix doesn’t correct an old funding article or directory page that still names the wrong founder. Ask publishers to correct factual errors, citing your About page, and keep that page the clearest statement of the facts anywhere on the web.
Hypothetical example:
A fintech’s panel lists an early angel investor as co-founder. Searching the investor’s name with “co-founder” and the brand finds it in a 2019 funding-round article, a business database profile and the Wikidata “founded by” statement, which cites that article. The team adds a founders line to the About page, corrects the database profile, and proposes a Wikidata correction citing a later interview in which both real founders describe the company’s start. They submit verified feedback the same week the Wikidata change is accepted, and the panel changes nine days later. They can’t tell which change did it, and they accepted that when they chose speed over sequencing. The 2019 article keeps surfacing in AI answers until its publisher adds a correction.
If nothing changes
If the public sources are fixed and verified feedback is declined or goes nowhere, work down this list. The panel may be tied to the wrong entity (check the kgmid, and whether the description belongs to someone else). A source you haven’t found may still carry the value (search the wrong value again, including in other languages if you operate abroad). A Wikipedia or Wikidata edit may have been reverted (check the histories). Or your own site may still contradict itself: an old press page, a footer address, a PDF annual report. Each of those has a fix, and none of them is solved by resubmitting the same request.
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.