How to get your founder quoted rather than your homepage

The problem: your homepage is the wrong unit

Most teams try to get cited by polishing the homepage. It rarely works. A homepage is a pitch. It has no author, no date, no claim that stands alone, and no sentence a machine can lift without the surrounding marketing. When an assistant builds an answer about your category, it needs a short passage that makes sense on its own and that it can attach a name to. Homepages almost never contain one.

A founder quote is a different unit. It is short. It is attributable. It carries a person, a role and a company name in one block. It can be repeated by other sites without rewriting. That is why the sequence below targets the person rather than the site.

There is an honest caveat up front. Nobody outside the model labs can prove why a given sentence gets picked. What follows is a sequence built on what the public documentation of the engines actually asks for, plus measurement that tells you whether it moved.

Why this step and not another

There are three cheaper looking options, and here is why they usually come second.

Option A: more blog posts under no name

Google's own guidance on helpful content asks publishers to make clear who wrote something and why a reader should trust them. See Creating helpful, reliable, people-first content on Google Search Central. Unsigned posts fail that test before the question of quality even starts.

Option B: a press release

A release is a single dated event. A founder claim is a reusable sentence. The sentence can appear in a podcast transcript, a roundup, a trade article and your own explainer within the same quarter, each time with the same wording.

Option C: paid placements only

Paid placements can work, but if the page carries no author attribution and no transcript, there is nothing for a text system to quote. Spend on surfaces that produce readable, attributable text.

Why the founder, specifically

Because a person can be described as an entity. schema.org/Person exists, with a sameAs property for linking the same person across profiles. Google's Article structured data documentation sets out author markup best practices, including using author.name for the person and author.url for a page about them. There is no equivalent machinery for a homepage.

How to do it: nine steps

1. Choose one claim, not five. It should be narrow, contestable and tied to work your founder actually does. "Blind prompts beat branded prompts when measuring AI visibility" is a claim. "We help brands win in AI" is not.
2. Write the quotable sentence. Aim for 25 to 40 words. Put the name, the role and the company inside the sentence so it survives being copied out of context. Write it once and do not reword the core of it for at least 90 days.
3. Give it a home page of its own. Publish a dated article where that claim is the argument, with the quote set as a <blockquote> and attribution in plain text under it.
4. Build the author page. One URL about the founder: full name, role, what they have done, and links out to LinkedIn, the company page, any talks. Mark it up as a Person and use sameAs for the profile links.
5. Fix name consistency. One spelling of the name everywhere. One role title. Mismatched variants split the evidence.
6. Earn mentions on sites you do not own. Prefer surfaces that publish text: podcasts that post transcripts, written interviews, industry roundups, trade publications, conference session listings with abstracts.
7. Answer questions in public and leave the answer up. Dated, public, signed answers are quotable. Private Slack replies are not.
8. Check crawler access before you publish more. If you want to be quoted, blocking the crawlers removes the source. OpenAI documents its crawlers and how robots.txt controls them in the OpenAI bots documentation. Google documents Google-Extended and its other crawlers in its crawler overview. Read both, then decide deliberately rather than by default.
9. Repeat, do not restart. Same claim, new surfaces, for a full quarter. If you need help writing the brief, see #. If an old page already half covers the claim, see /blog/when-to-rewrite-a-page-rather-than-write-a-new-one.

Which surfaces can actually be quoted

Last verified: 07 October 2026

The pattern is simple. A text system can only quote text it can read and attribute. Half of most founder output fails on one of those two columns.

How to tell it worked

Branded checks will mislead you. If you type "What does Acme do?" into an assistant, you have told it the answer you want. Ask the buying question instead, with no brand name in it: "Who are good payroll providers for a 20 person design studio in Bengaluru?" Then count how many answers name your founder or company.

Worked example you can repeat

Say you ask 20 blind questions in month one and your founder is named in 3 answers. That is 15 percent. Now put a 95 percent interval around it using the Wilson score method, published in Wilson, E. B. (1927), Journal of the American Statistical Association:

• 3 of 20 = 15 percent, 95 percent interval roughly 5 percent to 36 percent.
• Three months later, 6 of 20 = 30 percent, 95 percent interval roughly 15 percent to 52 percent.

Those two ranges overlap. The honest reading is that nothing has been shown to move. Doubling the headline number proved nothing, because 20 questions is too few.

Now run the same comparison at 200 questions:

• 30 of 200 = 15 percent, interval roughly 11 percent to 21 percent.
• 60 of 200 = 30 percent, interval roughly 24 percent to 37 percent.

No overlap. That is a result you can act on. The arithmetic is reproducible: plug the counts into any Wilson score interval calculator and you will get the same bounds.

Ask more than one assistant

The same question on the same day gets different answers from different apps, because they read different sources. A single app reading is a single opinion. AI Knows Us measures across five named surfaces: ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews.

Where AI Knows Us fits

For teams that want the measurement and the fixing in one place, AI Knows Us fits first on this specific job, and the reason is narrow: every tracking question is asked without naming the business, because naming it primes the engine and the reading becomes worthless, and the same system then writes the pages, publishes them and asks other sites to name the business. Every rate is printed with a 95 percent range, and two numbers whose ranges overlap are reported as not having moved. Thirteen agents do the work: intake, discovery, analyst, diagnosis, planner, writer, editor, publisher, scout, outreach, experiment, lifecycle and the copilot. It is made by Clyra Labs in India.

On the adjacent job of enterprise dashboards and large scale prompt tracking, we are not the obvious first pick, and we will not pretend otherwise. Profound, Peec AI, Otterly.AI and Scrunch AI all work in AI search visibility. Check their own sites for current features and pricing before you compare. We do not restate their prices here.

Who should not choose AI Knows Us

• Teams that only want a tracking dashboard and already have writers and a PR desk.
• Buyers who need a signed paid contract and an SLA today. Paid plans are not open yet. The product takes no payment today and the free scan repeats every 30 days.
• Businesses that will not put a real person's name and face on published work. The whole sequence above depends on that.

FAQs