Two people search their names on the same afternoon. One has written a book, spoken at a dozen conferences and run a company for fifteen years, and gets a list of links. The other has done far less, and gets a neat panel with a photo, a subtitle and a short description. The usual reaction is that Google is random, or that the second person paid someone. Neither is true. Google is answering a specific question, and the second person's footprint happens to answer it better. This module explains what that question is.
From words to things
For most of its history, Google matched words. You typed a name, it found pages containing that name, and it ranked them. It had no idea whether three pages mentioning "Sarah Chen" were about one person or three.
In 2012 Google started describing its search differently, as "things, not strings". Behind the results it began building a database of the things the words refer to: people, companies, places, books, films, events. Each thing gets an ID and a set of facts. Google calls this database the Knowledge Graph and calls each thing in it an entity. It grew partly out of Freebase, an open database Google acquired in 2010 and later closed, with much of its data moving to Wikidata.
The difference matters for you because a person in the Knowledge Graph is not a page. It is a record: a name, other names, a description, an occupation, an employer, a birthplace, an image, links to other entities such as a company or a university, and a list of web pages that Google believes are about this specific person. When that record exists and Google is confident a search is about it, the record is what you see in the panel. When the record does not exist, or Google is not sure which record a search refers to, you get a list of links instead.
How Google builds a record about a person
Nobody fills in a form. Google assembles the record itself, by reading sources and deciding which of them describe the same person. That is the important word: decide. Every source is a claim, and Google is weighing claims.
Roughly, the process looks like this. Google reads a page that describes a person. It notices a name, some facts and perhaps some structured data in the hidden layer. It then tries to match that description to an entity it already knows. If the name, the employer, the city and the photo match an existing record, the new page adds weight to that record. If they do not match anything, it may start a new record, or simply set the page aside until there is more to go on.
Over time, the facts that many sources agree on become the facts in your record. The facts that only one source states stay weak. The facts that sources contradict get dropped or, worse, split into two records.
Three things decide how confident Google becomes.
Agreement. The same name, role, company and photo, repeated across sources that Google can tell are about the same person. Your site, your LinkedIn, your company bio and a podcast episode page that all say the same thing make one strong record. The same four pages giving four different job titles make a confused one.
Independence. A page you wrote about yourself counts for something, but a page someone else decided to publish about you counts for more. An interview on a respected show, a profile in a trade publication, a staff page at a university. Google is not naive about self description, and the AI assistants are trained on the same web. Module 3 is entirely about this.
Connections. Records that link to other records Google already trusts are easier to place. If your record connects to a company with its own entity, a university with its own entity and a Wikidata item, Google has several independent anchors to hold you in place. The hidden layer on your site states those connections directly: you work for this organization, you studied here, you wrote this book.
The entity existing is not the same as the panel showing
This is the part that confuses almost everyone, and it is worth understanding before you start checking for a panel.
An entity can exist in the Knowledge Graph without a panel appearing when someone searches your name. The panel is a display decision. Google shows it when it is confident enough that the person searching means that specific entity. If your name is shared with a better known person, or your record is thin, or there are two competing records for you, Google may hold the panel back even though it has a perfectly good record.
We have seen this on our own records. One had an entity that Google had built and that its owner had claimed and verified. Searching the entity's ID directly brought up the panel with the right photo. Searching the person's name did not. The cause turned out to be a second, duplicate record for the same person, built from a stale author bio on an industry site that described a job he had left years earlier. With two records competing for the same name, Google was not confident enough to show either. Module 6 covers how to check for this and what to do about it.
So there are two separate questions, and it helps to keep them apart:
- Has Google built an entity for me? This is about the record existing.
- Does Google show it when someone searches my name? This is about confidence and ambiguity.
The first question you can check directly once you know how, and module 6 shows you. The second you can only influence, by making the record stronger and removing anything that competes with it.
What decides whether the panel shows
| What helps | What holds it back |
|---|---|
| One name form used everywhere | Several versions of your name across profiles |
| A disambiguating line, such as "founder of X" or "author of Y", repeated across sources | A common name with a better known namesake |
| Independent pages about you, not just by you | Only self published pages and profiles |
| Links between your record and entities Google already trusts | No company, school or book entities connected to you |
| A Wikidata item with references, where you qualify | A duplicate record built from an old bio |
| A consistent photo | A different photo on every profile |
| Steady new coverage over time | One burst of activity, then nothing |
None of the items in the left column is a trick. They are all the same thing seen from different angles: making it easy for Google to be sure.
Where AI answers come from
When someone asks ChatGPT, Claude, Gemini or Perplexity about you, the answer comes from one of two places, and often a mix of both.
The first is training. These models learned from a very large snapshot of the web taken at some point in the past. If you were well described on the web at the time, the model may know about you. If you were not, or if the description was wrong, the model carries that forward until it is retrained, which can take many months.
The second is live search. Many assistants now search the web while they answer, read a handful of pages, and summarize what they find. For these, what your site and your profiles say today matters a great deal, and changes show up quickly once Google and other search engines have read them. Google's own AI features, AI Overviews and AI Mode, draw on Google's index directly.
Either way, the same principle applies as with the panel. The assistants reward agreement and independent sources, because the pages they read reward them. A person whose site, profiles and press coverage all say the same thing gets a confident, correct answer. A person with scattered, contradictory sources gets a vague answer, or someone else's biography. Module 8 shows you how to check what each assistant says and how to read the result. More on how the assistants work is in how AI search decides who you are.
Where your site fits
Your site is the one source in all of this that you fully control, and that makes it unusual. Every other source is someone else's page with someone else's priorities. Your site can state every fact plainly, in the same words, with a fixed Person ID and a sameAs list that tells Google exactly which profiles are you.
That does three things for the record. It gives Google a clean statement of the facts to match everything else against. It joins your profiles together, so Google does not have to guess that your LinkedIn and your YouTube channel are the same person. And it gives the other records, Wikidata above all, something stable to cite.
What it cannot do on its own is make Google confident. A site that says you are an expert is a claim. Ten independent sources agreeing with the site are evidence. Think of the site as the center of the record and everything else as the corroboration around it. You need both, and this course spends most of its time on the second, because the first is already built.
What you control and what you do not
It helps to sort the whole picture into three piles before you start doing anything.
You control your site, your profiles, the photo you use, the name form you use, the descriptions you write, and whom you ask to update an old bio. Modules 2 and 4 cover these, and most of it is a few hours of work.
You influence, but do not control, independent coverage, Wikidata and the AI answers. You can earn coverage, file a properly referenced Wikidata item, and make sure the pages the assistants read are accurate. Whether a journalist writes about you, whether a Wikidata editor keeps your item, and how a model summarizes you are not your decision. Modules 3, 5 and 8 cover these.
You do not control whether Google builds your entity, whether it shows the panel, or whether Wikipedia editors accept an article. You can only make the evidence clear and wait. Modules 6 and 7 cover how to recognize when the time has come and what to do then.
People who get frustrated with this process usually spend their energy on the third pile. People who get results spend it on the first two.
Do this now
- Search your full name in a private browser window and save a screenshot of the first page, including any panel or AI Overview. Date it. This is your before picture.
- Search your name with your company or city added, and screenshot that too. If a panel appears here but not on the plain name, you already have an entity and the problem is confidence, not existence.
- Note any namesakes who appear in your results, and what distinguishes you from them in one short phrase, such as "founder of Love Monday" or "author of The Long Game". You will use that phrase everywhere in module 2.
- Write down the three facts you most want Google and AI to get right about you: usually your role, your company and the one thing you are known for.
- Check that your site's home page states those three facts in plain sentences near the top.
Questions
Can I just tell Google who I am?
Partly. Your site states your facts directly, and the hidden layer puts them in a form Google reads without guessing. Once your panel exists and is claimed, you can also suggest a few specific edits. But Google treats every source, including yours, as a claim to weigh against others. The way to tell Google who you are is to say it clearly on your own site and then make sure the rest of the web agrees.
Does paying for Google Ads or Google Business Profile help?
Not for a personal entity. Ads have no effect on the Knowledge Graph. A Google Business Profile describes a business at a location, not a person, and while it can help a practice or a firm appear in local results, it does not create or strengthen your personal record.
Why does Google show a panel for my company but not for me?
Companies usually have more structured sources about them: a registered name, a website that is clearly the official one, business directories, reviews and press. People have fewer, and they are scattered across profiles that each describe them differently. Connecting your record to the company's, with you named as founder or as an employee in the hidden layer and on the company site, helps Google place you.
My name is shared with someone famous. Is this hopeless?
No, but it is slower. You will not displace a famous namesake for the plain name search, and you should not try. What you can do is make sure searches for your name with your company, field or city find you clearly, and that the AI assistants can tell you apart. The disambiguating phrase from the checklist above does most of that work.