SEO, AEO and GEO: How to Get Cited by AI Search, Not Just Ranked

A buyer used to type a phrase into Google, see ten links, and pick one. Now a growing share of them ask ChatGPT, Perplexity or Google's own AI summary a full question and read one paragraph of answer with three or four brands named in it.

The difference is not cosmetic. Ranking tenth still put you on page one. Being the tenth-most-credible brand in a model's answer puts you nowhere at all. There is no page two.

The three letters, plainly

They are not competitors. SEO still feeds the index the models train and retrieve from. AEO is the bridge. GEO is where the buying decision increasingly happens for considered purchases, which is exactly what professional services are.

What actually earns a citation

After watching this closely across our own site and our clients', five things move the needle. None of them are tricks.

1. Answer the question in the first two sentences. Models extract. A page that spends four paragraphs warming up gives them nothing to lift. State the answer, then support it. Every article on this site opens that way on purpose.

2. Be specific enough to be worth quoting. "Pricing varies by scope" is unquotable. "$750 a month plus a $3,000 implementation" is a fact a model can carry into an answer. Real numbers, real timelines, real process steps. Vagueness is the single most common reason a well-written page never gets cited.

3. Get mentioned somewhere other than your own site. This is the one most firms skip. Models weight consensus across sources. Your own page saying you are the best fractional BD firm counts for very little. A directory listing, a podcast appearance, a partner's case study, a trade publication quote, all count for a great deal. If your brand appears only on your own domain, you are asking a model to trust a single self-interested source.

4. Structure the page so a machine can parse it. Real headings that describe content, comparison tables, numbered steps, an FAQ block with schema markup. Models lift tables and lists far more readily than prose. This is why so many cited pages read like reference material rather than marketing.

5. Say who you are and stay consistent. Same brand name, same description, same service names everywhere. A firm that calls itself three different things across LinkedIn, its site and its directory listings dilutes its own signal.

The part almost nobody has done

Most modern marketing sites are JavaScript applications. A person's browser runs the code and sees a full page. Many AI crawlers do not run the code, and see an almost empty shell.

You can check this in thirty seconds. View source on your own homepage. If the text you see on screen is not in the raw HTML, that is roughly what a crawler gets.

The fix has a name now. Publish a plain-text version of every meaningful page, link to it from the HTML page with a rel="alternate" tag, list your key pages in an llms.txt file at your root, and serve the plain-text version to anything that asks for markdown. It is unglamorous work and it is the difference between being readable and being invisible.

This site does exactly that. Every significant page has an addressable plain-text twin at the same path with a .md suffix, they are indexed for AI crawlers separately from the Google sitemap, and /llms.txt maps the whole site for any model that looks for it.

One caution from our own experience: keep those plain-text twins out of your Google sitemap. A sitemap entry is a request to index that URL, and Google will not index a plain-text duplicate of a page it already has. Put them in a separate file that AI crawlers can find and Google is never asked to index.

How to tell whether it is working

Rankings will not tell you. You need to check the answers themselves.

• Write down the fifteen questions a real buyer would ask before hiring you. Not keywords. Questions.
• Run them through ChatGPT, Perplexity and Google's AI overview once a month.
• Record whether you are named, in what position, in what tone, and who else appears.
• Track your share of those answers over time, and which sources the models cite when they do not cite you.

That last column is the useful one. The sources a model cites instead of you are your actual competitive set for AI search, and they are frequently not the companies you thought you competed with.

Where this fits in the work

The mechanics above are half the job. The other half is having something worth citing, which means content built around what you actually do rather than generic roundups.

Our marketing services cover both sides: the positioning and messaging that makes a brand quotable, the content system that produces pages models want to lift from, and the technical work that makes those pages readable in the first place. Where the constraint is that nobody can see the buying decision clearly, strategy execution comes first, because no amount of visibility fixes an unclear offer.

If you want to see the pattern in practice, what is a business navigator and what to look for in an AI-powered CRM are both built to the rules above: direct answer up front, real specifics, comparison tables, and honest scope.

Start here

Three things, in this order:

1. Check whether a crawler can read your site at all. Everything else is wasted if it cannot.
2. Write down the fifteen questions and get a baseline for how often you are named today.
3. Fix the three pages that answer your highest-value questions. Specific numbers, tables, direct answers.

Most firms find the first step alone explains why they have been invisible.

Book a discovery meeting and we will run your fifteen questions, show you who gets cited instead of you, and tell you plainly whether the gap is technical, editorial or positioning.