Most DIY Knowledge Panel guides stop one level above the thing you actually need, which is the text. They tell you to add Person schema. They do not give you the Person schema. So here is the parts bin, in the order you will use it.
Everything below assumes you have already done the two steps that come first: decided your identity line, and stood up a personal site on your own name that is separate from your company site. If those are not true, the parts here have nothing to attach to.
01Part one: the Person schema
This goes in the head of your personal site's homepage, and nowhere else. One Person block per person. Replace every value. Delete any line you cannot honestly fill, because an empty or invented field is worse than a missing one.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Person",
"@id": "https://yourname.com/#person",
"name": "Your Full Name",
"url": "https://yourname.com/",
"image": "https://yourname.com/img/headshot.jpg",
"jobTitle": "Founder and Chief Executive",
"description": "Your Full Name is the Founder and CEO of Your Company, based in Your City.",
"worksFor": {
"@type": "Organization",
"@id": "https://yourcompany.com/#organization",
"name": "Your Company",
"url": "https://yourcompany.com/"
},
"address": {
"@type": "PostalAddress",
"addressLocality": "Your City",
"addressCountry": "IN"
},
"alumniOf": {
"@type": "CollegeOrUniversity",
"name": "Your University"
},
"sameAs": [
"https://www.linkedin.com/in/yourhandle/",
"https://www.youtube.com/@yourhandle",
"https://x.com/yourhandle",
"https://www.instagram.com/yourhandle/",
"https://publication.com/author/yourname/",
"https://books.google.com/books?id=YOURID"
]
}
</script><script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"@id": "https://yourcompany.com/#organization",
"name": "Your Company",
"url": "https://yourcompany.com/",
"logo": "https://yourcompany.com/img/logo.png",
"foundingDate": "2019-04-01",
"founder": {
"@type": "Person",
"@id": "https://yourname.com/#person",
"name": "Your Full Name",
"url": "https://yourname.com/"
},
"address": {
"@type": "PostalAddress",
"addressLocality": "Your City",
"addressCountry": "IN"
},
"sameAs": [
"https://www.linkedin.com/company/yourcompany/",
"https://x.com/yourcompany"
]
}
</script>What each field is doing
- name is the canonical string. Whatever you put here has to match every profile character for character.
- url is the entity home, the site this block lives on.
- image is the master headshot, the same file you use on every profile.
- jobTitle and worksFor connect the person to the organisation.
- address gives the geographic anchor. City and country are enough.
- alumniOf only if it is verifiable and relevant.
- sameAs is the array that does the real work. Rules for it are in part three.
The Organization block goes on the company site and points back with a founder field naming the same person and the same url. The two blocks referring to each other is the whole point: it is how a machine learns that a person and a company are two connected things rather than one confused thing.
02Part two: the bio fields, written once and reused everywhere
Source: Our own eight client records, read live from Google on 31 August 2026
Write these once, keep them in one file, and paste from that file forever. The reason is consistency, not laziness. Rewriting your bio each time is how you end up with four different job titles in public.
| Field | Length | Where it goes | Rule |
|---|---|---|---|
| Identity line | One sentence | First paragraph of the personal site, and the first line of every bio | Plain English, no adjectives. Name, role, organisation, city. |
| Short bio | About 75 words | X, Instagram, podcast guest forms, conference programmes | Third person. Same first eight words as the identity line. |
| Standard bio | About 150 words | LinkedIn about, guest post author boxes, press kits | Third person. One verifiable number if you have one. |
| Long bio | 300 to 400 words | The personal site about page, speaker packets | Third person. Chronological. Links to the records that prove each claim. |
| Headline | Under 120 characters | LinkedIn, X, YouTube channel | First eight to ten words identical across all three. |
| One-line CTA | Under 15 words | Every profile's link field | Points at your own domain, never a Linktree. |
Third person is not vanity. Bios written in the first person are hard for a machine to attribute, because the subject of the sentence is a pronoun. "Bhavik Sarkhedi is a personal branding consultant based in Ahmedabad" states an entity and a fact about it. "I help founders get seen" states neither.
03Part three: the sameAs rules
This is the array that does the actual entity work, and it is the one most people get wrong. Each URL in it is a claim: this profile is the same person as the one described here. Get it right and eight scattered weak signals become one strong one. Pad it and you dilute the set.
| Good sameAs entries | Weak or harmful entries |
|---|---|
| Your LinkedIn profile | A LinkedIn company page |
| Your YouTube channel | A single YouTube video |
| An author archive on a publication | One article you wrote |
| Your Wikidata item, if one exists | A Wikipedia article about somebody else |
| A Google Books author page | A bookseller product listing |
| Your Instagram and X profiles, if active | A dead account from 2016 |
| Your Crunchbase or IMDb entry, if genuine | A paid directory listing |
| A conference speaker page that names you | A generic attendee list |
- Include only profiles that are genuinely yours. Not your company's. Not a fan page.
- Include only live profiles. A dead account dilutes the set rather than adding to it.
- Use the canonical URL form for each platform, as the platform itself displays it, with no tracking parameters and no mobile subdomain.
- Order does not formally matter, but put the substantial ones first. It costs nothing to be tidy.
- Include author archive pages. If a publication gives you an author URL, that is one of the strongest entries available to you.
- Do not pad. Six real profiles beat twenty directory listings, and the padding is visible.
04Part four: the profile alignment spec
| Claim you will hear elsewhere | What the evidence actually says |
|---|---|
| Schema markup gets you cited by AI | A controlled study found AI Overviews -4.6% after adding JSON-LD. Ship schema for rich results, never bill it as AI visibility. |
| 76% of AI Overview citations come from the top 10 | Superseded. The figure is 37.9%, from a sample roughly twice the size. |
| Write a listicle putting yourself first | When a brand's own listicle was cited, the brand was left out of the recommendation 69% of the time. |
| Longer articles get cited more | Correlation between word count and citation is 0.04. Over half of cited pages are under 1,000 words. |
| A knowledge panel takes 90 days | Our build window is 60 to 90 days. The panel tends to fire on the bare name at 6 to 12 months. |
This is a checklist, not a philosophy. Open every profile you own and make these six things identical. Mismatches are a confidence leak, and they are the most common silent regression six months after a build.
- The name string, character for character, including spacing and capitalisation
- The headshot, the same image file, not a different crop of the same session
- The first eight to ten words of the headline or title
- The city, written the same way, not Ahmedabad in one place and Ahmedabad, Gujarat in another
- The first line of the bio
- The primary link, pointing at your own domain
Do this across LinkedIn, YouTube, Instagram, X, your podcast feeds, any trade directories and any conference bios you can still edit. It is two hours of unglamorous work and it is the highest-return two hours in the entire toolkit.
05Part five: the claim statement
When a panel does appear, claiming it means answering a free-text field asking why you are claiming it. This is where most rejections happen, and the reason is a misread of the question.
What gets rejected
- I am the subject of this panel and I would like to manage it
- This is my personal brand and I want to control what it says
- I am a public figure and this represents me
- Long explanations of who you are with nothing Google can check
What gets approved
- A statement naming the exact records already in Google's index that identify you
- Links to the entity home, the author archive, and the coverage that names you
- The identity line, verbatim, matching what those records say
- Short. Three or four sentences of evidence, not of introduction
06Part six: the quarterly maintenance list
Source: Ahrefs, 1,885 pages that added JSON-LD against 4,000 controls, four methods
A panel is not a finished object. It drifts, it merges with namesakes, and it occasionally disappears. Fifteen minutes a quarter keeps it honest.
- Every weekThirty-minute review
Search your own name signed out. Note anything new on page one, good or bad.
- Every monthThe three-assistant baseline
Ask ChatGPT, Claude and Perplexity the same three questions about you, with the same wording, and save the answers with the date. The delta is the measurement.
- Every quarterThe nine-query audit
Name alone, name plus city, name plus role, name plus company, name plus each of your two strongest topics, and the three namesake variants. Note which results you control.
- Every quarterOne new third-party node
One genuine new record that names you and links home. One is enough. Consistency beats volume.
That is the whole toolkit. None of it is secret, none of it requires a tool you have to buy, and all of it takes longer than it looks. The parts are free. The hours are the price.
What to remember
- Person schema on the personal site, Organization on the company site, wired to each other.
- Write six bio assets once and paste from that file forever.
- sameAs is the load-bearing field, and padding it makes it weaker, not stronger.
- Profile alignment is six identical things across every surface you own.
- The claim statement answers why you deserve a panel, with evidence Google can already see.
- Fifteen minutes a quarter is what stops the whole thing drifting.
Questions people actually ask
Where exactly does the Person schema go?
In the head of your personal site's homepage, inside a script tag with type application/ld+json. One Person block for one person. Do not repeat it on every page, and do not put a second Person block for somebody else on the same site.
Should sameAs include my company's profiles?
No. sameAs on a Person block means these URLs are the same person. Your company's LinkedIn page is a different entity. Link the person and the company with worksFor and founder instead, which is the field pair built for exactly that relationship.
Will adding schema get me cited by ChatGPT?
Not on its own, and it is worth being clear about this because a lot of people are sold the opposite. A controlled study of 1,885 pages that added JSON-LD against 4,000 controls found AI Overview citations fell 4.6 percent. Ship schema because it earns rich results and makes your records machine-readable. Do not buy it as an AI visibility product.
What if my name is very common?
Then the geographic and role qualifiers in your schema and bios are doing more work than usual, and you should expect the panel to fire on name-plus-qualifier long before it fires on the bare name. Namesake collisions are real and they can be live, meaning Google serves a different person's panel for the same query on different days. Fixing that is a longer job than a first build.
Can I use the same headshot everywhere, or does that look lazy?
Use the same file everywhere. It is not a style choice, it is an identity signal. Image similarity is one of the ways a machine confirms that six profiles are one person, and using six different photos removes that signal for no benefit.