Writing an About page a machine can read

Who you are, what you do, who you serve — in the first two hundred words, in plain sentences, with no adjectives doing the work of facts.

Your About page is where a machine goes when the homepage is all headlines and social proof. It is also, on most sites, the least edited page — written at launch, quietly outdated, and full of the founder's origin story instead of the facts.

The fix is not to make it boring. It is to put the facts first and the story after.

The first two hundred words carry everything

A model extracting a description of your company reads from the top and stops early. Everything below the fold is a bonus. So the first paragraph has to answer three questions:

1. Who are you? The brand name, spelled the way it appears in your schema and your title tag.
2. What do you do? The category, in words a stranger would use. Not your internal term for it.
3. Who is it for? The specific customer. "For businesses" is not an answer.

A version that works:

Northstar Analytics is a product analytics platform for B2B SaaS teams. Product managers and growth engineers use it to see which features drive activation and retention, without waiting on a data team. Founded in 2021, based in Denver, used by around 400 companies.

Three sentences, all four facts, no adjectives doing the work of nouns. A model can turn that into a recommendation. It can also turn it into a correct recommendation, which matters more — you do not want to be suggested to people you cannot help.

The version that does not work is the one every third About page opens with:

We believe great products are built by teams who truly understand their users. Our journey began with a simple question…

There is nothing extractable in that. No category, no audience, no name after the first line. It reads fine to a human already on your site and it gives a machine nothing at all.

Facts worth including

Once the opening paragraph has done its job, these earn their place:

• Founded — a year. Corroborates your schema and your Crunchbase record.
• Location — city and country. Anchors you geographically, which matters more than people expect for "best X in Y" questions.
• Size — a range is fine. "Around 25 people."
• Scale — customers, users, volume. Any real number.
• Named people — founders and leadership, with titles. Real names with real roles are a strong signal that you are a real operating company.
• Credentials — certifications, licences, memberships, compliance. Verifiable claims, not badges.
• What you do not do — genuinely useful, genuinely rare. "We do not work with agencies" or "not a replacement for a data warehouse" prevents bad matches.

Facts to leave out

Superlatives you cannot support. "Leading", "world-class", "revolutionary", "best-in-class" — a model has read ten thousand pages that say this and has correctly learned to ignore it. Worse, it displaces the specific claim that would have been believed.

Also: vague year ranges, "trusted by thousands" without a number, and awards from bodies that exist to sell awards.

Structure that helps

Use real headings. <h2>What we do</h2>, <h2>Who we work with</h2>, <h2>Team</h2>, <h2>Contact</h2>. Extraction follows headings, and a wall of undifferentiated prose is harder to pull facts from than the same text in labelled sections.

Use plain paragraphs and lists rather than text baked into images or split across animated panels. If your About page is a scroll-triggered experience, check what it looks like with JavaScript disabled — that is what the crawler sees.

Put it at /about or /about-us. Those are the two paths every crawler tries, ours included. A brilliant About page at /company/our-story gets found less often than a mediocre one at /about.

Make it agree with everything else

The About page is one of four places your company gets described, and the four have to say the same thing:

• the description in your Organization schema
• your meta description
• your LinkedIn and Crunchbase blurbs
• this page

Not identical prose — the same category, the same audience, the same name. Where they disagree, a machine either picks one arbitrarily or hedges, and a hedge is how you end up out of the shortlist. More on why that matters.

A five-minute test

Copy your About page text into a fresh assistant session and ask: "What does this company do, who is it for, and where is it based?" If the answer is vague, hedged, or wrong, you have your edit list. That is the same extraction the real thing performs, run in a place where the result costs you nothing.