Picture this. A founder in Cleveland sells custom orthotics through a Shopify store. She wants Perplexity, ChatGPT, and Google’s AI Overviews to mention her brand when someone asks about high-arch insoles for runners. She has product pages, a blog, and a mailing list. What she does not have is an About Us page that gives an AI model anything to cite. Her About reads like a yearbook entry. “Founded in 2019 with a passion for foot health, our team is dedicated to crafting premium orthotics for active lifestyles.” That sentence contains no facts a model can use. No founder name, no city, no production count, no third-party mention. A model that reads it learns nothing it can repeat back to a user.

The About Us page is the single most underrated AEO asset on a small business site. It is the page where your entity gets named, defined, and connected to every other source on the open web. When an AI model builds its internal record of who you are, the About page is one of three or four anchor documents it returns to. The other anchors are your homepage, your Wikipedia or Wikidata entry if you have one, and the highest-authority third-party article that mentions you. If your About is vague, the model has nothing to attach the rest of its citations to. The brand stays a ghost.

This piece walks through the structure of an About Us page that earns AI citations, the exact schema markup it needs, the way to write it so models extract clean facts, and the test you should run before you publish. If your business sells anything, your About page is doing entity work for you whether you designed it that way or not. Most small business About pages do that work badly.

Why an About Us page is an AEO asset, not a brand essay

When someone types “best orthotics company for runners with high arches” into Perplexity, the model does three things in sequence. It interprets the query. It retrieves a candidate set of brands and sources. It assembles an answer with citations. The retrieval step is where most small businesses lose. The model is looking for entities it can verify. An entity is verifiable when its core attributes (name, category, founders, location, credentials) appear in multiple sources that agree with each other. The About Us page is the canonical statement of those attributes. Every other source on the web either confirms or contradicts what the About says.

A vague About page creates a contradiction surface. If your About says you are a “premium wellness brand” and your Crunchbase listing says you are “footwear ecommerce,” the model has to pick a category, and it usually picks the third-party source because external citations carry more trust weight. You lose control of your own entity record. A precise About with the exact same category language as your Crunchbase, LinkedIn, and Google Business Profile gives the model a clean, redundant signal. Redundancy is what lets a model cite you with confidence.

I tested this in March 2026 with a client whose About said the company was a “lifestyle brand for the modern professional.” Perplexity, asked to define the company, called them an “apparel and accessories retailer.” After we rewrote the About to specifically say “men’s footwear company specializing in dress shoes for size 14 and larger,” and updated their Crunchbase and LinkedIn to match, Perplexity’s description shifted within six weeks to match the new wording. The model is not creative. It is reading what is in front of it.

The Entity Anchor Stack: seven elements every About page needs

I call the structural template the Entity Anchor Stack because each element acts as an anchor that ties your About page to a different layer of the AI model’s retrieval pipeline. Skip an anchor and the model has to guess. Include all seven and the model has nothing to guess at.

The seven anchors, in the order they should appear on the page:

  1. Entity declaration. First sentence names the company, names the category, names the audience. “Northgate Orthotics is a custom orthotic insole company serving runners and walkers with high arches in the United States and Canada.” No metaphors. No brand voice flourishes. Just the entity record.

  2. Founding facts. Second paragraph contains year founded, founder names, original location, and the precise event that started the company. “Northgate was founded in 2019 by Dr. Maria Velasquez, DPM, in Cleveland, Ohio, after seventeen years of clinical practice at the Cleveland Clinic Foot and Ankle Center.” Dates and named institutions create verifiable hooks.

  3. What you do, in one paragraph. Describe the offering with specificity. Include the product, the service, the price band if appropriate, and the customer outcome. Avoid abstractions. “We make custom 3D-printed orthotic insoles, priced between $179 and $299 per pair, scanned remotely through a smartphone foot-mapping app, and shipped within ten business days.”

  4. Who you serve. Name the audience with attributes a model can match against query language. “Our customers are typically distance runners, hikers, and retail workers between ages 28 and 65 with confirmed pes cavus or moderate-to-high arch profiles.”

  5. Credentials and proof. List the named credentials, certifications, partnerships, and third-party mentions. Link out to each one. “Dr. Velasquez is board-certified by the American Board of Foot and Ankle Surgery (ABFAS, 2008). The company is a registered medical device manufacturer with the FDA (Class I, Establishment 3017284532) and has been featured in Runner’s World (March 2024), Outside Magazine (June 2024), and Podiatry Today (October 2025).”

  6. Where to find you. Physical address, hours if applicable, primary contact channel, and links to your verified profiles on the platforms AI models cross-reference (LinkedIn company page, Google Business Profile, Crunchbase, Wikidata if you have it).

  7. Cited press and links to reviews. The bottom of the page should contain a “Press” or “As mentioned in” block with linked logos or text citations. This is the citation surface AI models scan when deciding whether to repeat one of your claims.

That is the complete stack. Most About pages contain elements 1, 2, and a watered-down version of 3. The rest get cut because they do not feel “brand-friendly.” The brand cost of including them is real but small. The AEO cost of leaving them out is the entire reason you cannot get cited.

Schema markup that makes the page legible to machines

The About Us page should be marked up with JSON-LD structured data. Specifically, it should declare an Organization (or LocalBusiness, if you serve a defined geography) and a Person entity for each founder or principal. The Organization block should include the sameAs property pointing to every external profile that confirms your existence. Wikidata, Wikipedia, Crunchbase, LinkedIn, X, Facebook, Instagram, YouTube, and any niche industry directories you appear in. Each sameAs URL is a vote that you exist as you claim.

A minimum viable Organization block contains: name, alternateName (if you use one), url, logo, foundingDate, founder (linked to a Person block), address (PostalAddress), contactPoint (ContactPoint with telephone and contactType), and sameAs (array of URLs). The Person block contains: name, jobTitle, worksFor (linked back to the Organization), and sameAs (the founder’s LinkedIn, X, professional bio pages).

A test I run with every client: I take the rendered HTML of their About page, paste it into Google’s Rich Results Test, and check whether the structured data parses without errors. If it does not parse, the page is invisible to half the AEO pipeline. The Schema.org spec is finicky about required fields. Get the markup wrong and the model treats it as if you did not declare anything.

How to write the prose so models can extract it

Writing for AI extraction is a different exercise than writing for humans. Humans tolerate paragraphs that wander toward a point. Models do not extract from wandering paragraphs. They extract from sentences that contain a subject, a verb, and an object that map cleanly to a fact. Three rules.

First, declarative sentences over rhetorical ones. “We were born from a frustration with cookie-cutter insoles” is rhetoric. “Northgate Orthotics was founded in 2019 to address the underserved market of runners with high arches” is declarative. The first sentence contains zero extractable facts. The second contains four (entity name, founding year, founder motivation, target market).

Second, name everything. People, places, products, partnerships, dates, certifications. Models cite specific entities, not abstractions. A page that says “we partner with leading hospitals” gets ignored. A page that says “Northgate provides custom orthotics to the Cleveland Clinic Sports Medicine program (since 2022) and the University Hospitals Drusinsky Sports Medicine Institute” gives the model two named partnerships it can verify and cite.

Third, repeat the core entity description in slightly varied wording across the page. The first paragraph says “Northgate Orthotics is a custom orthotic insole company.” The fifth paragraph might say “As a manufacturer of 3D-printed orthotics for runners and walkers, Northgate has…” The model averages these statements. Consistent repetition reinforces the canonical description. Inconsistent statements (calling yourself an “insole company” once and a “wellness brand” later) introduce the contradictions that flatten your citation rate.

A live test you can run today

Go to Perplexity. Type “Tell me about [Your Company Name]” and read the answer. Then type “What does [Your Company Name] do?” and read that answer. Then type “Where is [Your Company Name] based?” and read that one. The three answers should match each other and match what your About page says. If they contradict each other, your entity record is fragmented. If they all return “I cannot find specific information about that company,” you have no entity record at all. If they return information that is wrong, your About page is being out-cited by a stale third-party source.

I ran this test on a client called Hilltop Coffee Roasters in April 2026. Perplexity told me they were “a coffee subscription service.” Their About said they were “a small-batch roaster with retail and wholesale operations.” The discrepancy traced to a 2023 Yelp listing that described them as a subscription service before they had pivoted to retail. We updated Yelp, rewrote the About to lead with “small-batch coffee roaster,” and added schema. Six weeks later, Perplexity’s answer matched.

The test takes two minutes and tells you exactly where the gap is. Most small businesses have never run it.

Common mistakes that kill citation potential

The five most common About Us mistakes I see, in order of severity:

The metaphor opener. “Born from a love of…” or “Crafted with passion…” kills the entity declaration on line one. The model has to scroll to find a fact. By the time it does, it has already given up and gone back to the search results. Lead with the entity statement.

The pronoun trap. Pages written entirely in “we” and “our” instead of the company name. “We are committed to excellence” cannot be extracted as a fact about your specific business. Use the company name early and often. At least once per paragraph.

The credentials dump as image. Founders’ credentials presented as a photo carousel or a designed graphic with no alt text. Models cannot read the image. Put the credentials in HTML text first, then optionally add the visual.

Missing or broken outbound links. Press mentions listed as text without links. The link is the verification path. A press mention without a link is a claim. A press mention with a working link to the article is a citable fact.

Schema set to the wrong type. Marking yourself up as Organization when you are a LocalBusiness, or as Person when you are a company, scrambles the model’s category assignment. Pick the most specific applicable type from Schema.org and stick with it across all pages.

What to do this week

Take your current About page. Open a blank document next to it. Write the seven Entity Anchor Stack elements from scratch in order, using only verifiable facts. When you finish, compare what you wrote to what is on your site. The gap is your work order. Get the prose updated, get the schema added (if you cannot write JSON-LD yourself, hand the seven facts to a developer and they will produce it in twenty minutes), and ship the changes. Then run the three-question Perplexity test in fourteen days and again in sixty days. The first test tells you the baseline. The second tells you whether the model has reindexed.

This is the cheapest, highest-impact AEO move you can make. The page already exists. Most of the work is replacing brand voice with extractable facts. The model is waiting. Give it something to cite.