Getting Found in the AI World: Rewrite Positioning to Be Quoted, Not Ranked

Forwarded to you? I am Heath. I build go-to-market systems and put AI to work in sales, the right way, then I write down exactly what I built, what broke, and what it moved. One story per week, receipts only. This one is about getting found where your buyer actually looks now, inside the machine that answers them.

The reflex: buyers search differently now, so add some AI keywords

Every team feeling the traffic drop reaches for the same move. Buyers are asking ChatGPT and Perplexity instead of Google, so you go update the page. You sprinkle in the new terms, you add an "AI" line to the headline, and you treat the answer engine like a search box with a different logo.

I did exactly that. At a company I was at, I rewrote positioning for the results page and the human skimming it, then bolted on the words I thought the machine wanted to hear. It did nothing. The engine kept returning an answer that did not include us, on the exact questions our buyers were asking at the moment they were ready to buy. We were not ranked low. We were absent. And you cannot keyword your way out of absent, because the engine does not reward the words you stuff onto a page. It repeats what it can source, in the buyer's own language.

The block was never the keywords. It was that our positioning was written for the category, not for the customer. There was no corpus of how buyers actually described the problem, so the machine had nothing true, and ours, to quote. The message just sat on the site in our words and died there, unquoted.

The AE feels this as "the buyer already has a shortlist before the first call." The CSM feels it as "the renewal team asked an engine to compare us and got someone else's framing." The marketer feels it as "we rank and still lose the answer." Same reflex every time: writing for the page, when the page is not where the buyer looks anymore.

The reframe: write to be quoted, not ranked

AEO and GEO are not SEO with a new coat of paint. The move is not to rank a page. It is to give the machine answering your buyer something true, and yours, to repeat. And the only way to earn that is to start where positioning always should have started: understanding the customer well enough to say the problem in their words, not the category's.

If your positioning cannot be quoted, you are not ranked low. You are absent, at the exact moment of intent.

Same move, other seats. A CSM watching a renewal drills it down: what does the buyer type into the engine when they reconsider us, and is our value in words they would recognize. A marketer staring at a traffic chart drills it down: which question precedes intent in our category, and does the answer to it source us or a competitor. The loud metric is never the thing you act on. The quotable sentence underneath it is.

How the best teams frame it

I am not the first person to argue that AEO is downstream of understanding the customer. The operators and analysts working the front of this shift mostly agree on where the leverage is, and it is not the keyword. It is the language and the source.

SOURCE

Andreessen Horowitz, "How Generative Engine Optimization (GEO) Rewrites the Rules of Search"

What it argues. GEO is about citations, not rankings. The goal is to be the source an AI answer trusts and repeats, not to hold a fixed position on a page nobody scrolls. Authority and clarity beat keyword coverage, because the engine is deciding what to quote, not what to list.

My take. Agree, and this is the part teams skip. They chase the citation like it is a ranking to win. The citation is the output. The input is whether the machine has something true, in the buyer's words, to source in the first place.

SOURCE

April Dunford, "A guide to advanced B2B positioning" (Lenny's Newsletter)

What it argues. Positioning has to account for what your specific buyers are actually looking for, in their terms. On AI, mention it as the enabler but lead with the value. Over-emphasizing AI because competitors are doing it obscures the thing you actually solve.

My take. Agree hard. This is why I reallocated the prompt budget instead of adding "AI" to every line. Name the AI only where it earns the line. Lead with the buyer's problem, or you sound like every other tool the engine already has fifty of.

SOURCE

Frase, "Answer Engine Optimization: The Complete Guide to Getting Cited by AI"

What it argues. AEO structures content so an AI system selects it as a cited source when it generates an answer. The win is the citation in the synthesized response, not the click, and clear, sourceable, direct answers to real questions are what get quoted.

My take. Extend. The order is the whole thing. Get the buyer's language first, then structure it. A perfectly structured page in the category's words still gives the machine nothing your buyer would recognize.

Even a curator has to concede when the field agrees: nobody who has actually worked this thinks the keyword is the hard part. The customer's language, and being a source the machine trusts, is the moat.

The method: Solve, Stack, Split

SOLVE THE CRUX

What is the real problem, framed as work and not a headcount?

The problem is not "we need AEO keywords." It is "the machine cannot quote us, because we never wrote in the buyer's words." So the first work is the customer. The same inputs that build a real ICP, customer interviews, reviews, call transcripts, and closed-won patterns, are exactly what tell you the language your buyer actually uses. For the AE that is the words that precede a deal, for the CSM the words that precede a renewal, for the marketer the words that precede intent.

STACK THE CONTEXT

What tech and signals turn a real problem into positioning a machine will quote?

Not a shopping trip. The buyer's own language lives in Gong call transcripts. Salesforce holds the demo and attribution data, read backward from the win instead of crediting the last click. Claude runs the answer-engine prompt research and drafts the rewrite in the customer's words rather than the category's. Notion holds the canonical story, locked and versioned so the site, the listing, and the pitch all pull from one source.

SPLIT · CUT THE DRAG

What low-judgment work goes to the system?

Pull the buyer's language out of the calls, read the attribution backward from the win, research the prompts the machine will source, and draft the quotable positioning. This is the mechanical part, reading and rewriting at a volume a person never gets through by hand, and it is the part AI does well.

SPLIT · KEEP THE JUDGMENT

What stays human?

The budget and the line. The operator decides the split: Defend the terms we own, Intercept the buyer's real question, Define the category on our terms. The operator names the AI only where it earns the line, and locks the canonical story so it stops drifting across three surfaces. The machine drafts. The operator decides what is true and yours.

The workflow: the board that runs it

Solve, Stack, Split is the shape. Here is the actual board, lane by lane: what the agents run, what stays human, and the tool at each step. Once it is wired, the story stays one source and gets refreshed instead of rewritten from scratch every quarter.

The receipt

At a company I was at, a B2B SaaS with a category full of near-identical tools, this is the exact build I ran. Start with the customer, read attribution backward, rewrite to be quoted, and reallocate the prompt budget on purpose.

We did not write more. We wrote in the buyer's words, on the questions that came right before a purchase, and let the machine source what was true and ours. The lift came from cutting the undifferentiated utility language and claiming the ground nobody had defined, not from adding volume.

Both of my receipts are positioning-side. Drop your own workflow in. A CSM turns the same build into a retention play: pull the language a renewing buyer uses when they reconsider, and make sure the answer they get sources your value in words they recognize. A growth marketer turns it into a category play: find the question that precedes intent, and own the Define ground in the answer instead of bidding on the keyword under it.

WHAT I LEARNED

1. AEO and GEO are not SEO with new keywords. The machine repeats what it can source, so you write to be quoted, not to rank.

2. The work runs backward. Understand the customer first, because you cannot be quoted on words your buyer does not say.

3. Spend the prompt budget on purpose. Defend the terms you own, Intercept the real question, Define the category, and name the AI only where it earns the line.

4. Positioning drifts the moment it lives in three places. Lock it to one versioned source or the message fights itself across surfaces.

The move this week

Do not rewrite the whole site. Pick the single question your buyer asks the engine right before they buy. Open your last ten call transcripts and pull the exact words they used to describe the problem, not your words for it. Write one paragraph that answers that question in their language, publish it where the machine can source it, and check whether the answer engine starts quoting you. That is the whole build in miniature, and it tells you if getting found is worth a quarter before you spend one.

Two builds that sit next to this one:

• The Win/Loss Backtest. "Before you trust a scoring model, backtest it against the deals you actually won and lost. The ones that feel important and the ones that are important are rarely the same list." Same instinct, reading backward from the win to find what is really true.
• The Governance File. "When you automate, your failures go silent. Write them down as enforced gates, not documentation." The same discipline that keeps positioning to one versioned source instead of three eras fighting.

This is one build from the Build Log. Every week I take one sales or revenue problem, run it through the loop, and show the receipts. If someone forwarded this, the subscribe button is right below. Keep building. Heath.