What Is AEO (and the Lies Marketers Are Telling You)

Sometime this week, you asked ChatGPT for a recommendation. Best accounting software, best sunscreen, best patio for a Friday beer. You read the answer, you never clicked a single link, and you got on with your day.

Your customers are doing the exact same thing about whatever you sell. Right now. While you read this. And there is a very real chance the answer they just got did not mention you.

Want to see how strange this new world is? Ask ChatGPT to name the best brand in your category. Write down the answer. Ask the exact same question again. Different list. A third time. Different again.

That is the shelf your customers are shopping from now. And it is why most of what gets sold as "AEO" is built on a promise nobody can keep.

First, the plain definition

AEO stands for answer engine optimization. It is the work of making your brand show up when AI tools like ChatGPT, Gemini, Perplexity, and Google's AI Overviews answer a question instead of handing over ten blue links. SEO was about ranking on a page of results. AEO is about being named inside the one answer people actually read.

And people are reading those answers in enormous numbers. ChatGPT passed 900 million weekly active users in early 2026. About 37 percent of consumers now say they start their research with an AI tool instead of a search engine. Meanwhile, when an AI Overview shows up on a Google search, a randomized field experiment measured a 38 percent drop in clicks to websites. The traffic you were counting on from ranking well is getting absorbed into answers. If your brand is not in the answer, you are not in the conversation.

So far, so good. Now for the part most pitches skip.

Lie one: "We will make you the AI answer"

If a vendor tells you they will "make you the AI answer," ask them to show you the same prompt run twice.

Rand Fishkin's team at SparkToro had 600 volunteers run identical prompts through ChatGPT, Claude, and Google's AI nearly 3,000 times. The result: less than a 1 in 100 chance that any two responses returned the same list of brands. These systems are probability machines. Variety is not a bug they will patch. It is how they work.

It shows up in tracking data too. AirOps studied over 45,000 citations and found that only 30 percent of brands stayed visible in back to back responses. Brands drop out of the answer and resurface constantly. The era of holding a fixed number one position is over.

One thing does hold, though. In that same SparkToro research, the biggest names in each category kept appearing across most runs. In headphones, the established players showed up in 55 to 77 percent of responses no matter how the question was phrased. The individual answer is a coin flip. The consideration set is not.

That is the whole game, and it is the honest version of AEO: no one can guarantee you are the answer. You can only become enough of an expert, with enough independent proof behind you, that you are a potential answer more often than your competitors are. You are not buying a position. You are earning a probability.

Lie two: "The AI already knows your brand, so you are covered"

This one is sneakier, because usually nobody sells it to you. You sell it to yourself. You ask ChatGPT about your brand, it comes back with a decent description of who you are and what you do, and you relax. The machine knows us. We are fine.

Knowing you and recommending you are two different things. AI engines carry two separate pictures of your brand: the entity picture (who you are, what you sell, where you operate) and the recommendation picture (whether you belong in the answer when someone is ready to spend money). You can ace the first one and be completely invisible in the second.

The proof is brutal. A 2026 audit of 17 established DTC brands found every single one had zero AI visibility on purchase queries, despite real brand equity. One premium baby care brand in the study had strong domain authority, 150,000 monthly organic visits, hundreds of Google AI Overview appearances, and 66 to 70 percent positive AI sentiment across five platforms. Ask the machines about the brand and they speak well of it. Ask them for the best or safest product in its category, the questions actual buyers ask, and the brand never comes up. Researchers running multi turn conversation tests found the same pattern: brands stay visible and well described early in a conversation, then get dropped at exactly the moment the system is asked what to buy.

This is why you cannot grade your AEO by asking "what is my brand" and admiring the answer. The queries that matter are the ones with commercial intent, and they run on different fuel: independent proof, reviews, comparisons, and third party coverage that gives the machine confidence to put your name next to the word "best." Being known gets you described. Being corroborated gets you recommended.

Lie three: AEO is one single thing

The next lie is that AEO is a deliverable. A schema plugin, a rewritten FAQ page, an invoice, done.

The data says the machine barely cares what you publish about yourself. Muck Rack analyzed more than 25 million links cited by ChatGPT, Claude, and Gemini and found that 84 percent of AI citations come from earned media. Journalism alone is 27 percent. Paid placements? 0.3 percent. Ahrefs studied 75,000 brands and found web mentions of your brand correlate about three times more strongly with AI visibility than backlinks do. And a controlled study found that pushing the same content through third party news outlets lifted citation rates from 8 percent to 34 percent.

Read that again. The strongest levers are things other people say about you. Which means real AEO work spans several disciplines at once:

• PR and press releases. Getting your story into publications AI engines treat as credible.
• Mentions and citations. Directories, review platforms, industry roundups, podcasts, anywhere your name appears next to your category in a source the machines trust.
• Authorities in your sector. When the recognized experts in your space talk about you, the machine listens, because it corroborates. Your website saying you are great is a claim. A sector authority saying it is evidence.
• Long form and short form video. Video content accounts for roughly 16 percent of LLM citations, and a long form piece with a transcript gives the machine something to quote while the short form clips build the audience that makes you worth quoting.
• Your own site and content. Still matters. It is the entity record the machine checks your story against. It is just not the whole story, and it was never going to be.

Which of those levers matters most for your brand? I can tell you the honest answer: it depends, and anyone who says otherwise before looking at your data is guessing. When we run our What AI Knows About You audit, we ask the machines the questions that actually move money, the queries with commercial intent: best in your category, who should people hire, is your product worth it, you versus the competitor everyone names. Then we look at where those answers are coming from. Sometimes the gap is press. Sometimes it is video. Sometimes AI is confusing you with another company entirely (I have watched that happen to a manufacturer, and it is not fun to explain). The data picks the plan. Not the package.

Lie four: AEO is one and done

Now the part of the pitch deck nobody puts in bold: even when the plan works, you are not done.

SISTRIX tracked AI citations for 17 straight weeks and found 54 to 59 percent weekly churn in the sources Google's AI Mode cites, with no stabilization anywhere in sight. The models retrain. The engines rebalance for freshness. Your competitors publish, pitch, and post. And they are not slowing down: 94 percent of digital leaders planned to increase their AI search investment in 2026.

The ground moves. Your competitors move. The answer machine reshuffles the deck every single time someone asks. Treating AEO as a project with an end date is like mowing your lawn once and calling landscaping solved.

I spent 25 years in software, and part of that work was usability testing. The discipline here is identical. One round is not testing. It is a single data point pretending to be a pattern. The real loop looks like this:

1. Baseline. Ask the engines the commercial intent questions about your brand and category, the ones real buyers actually type. Record what they say and who they cite.
2. Act on the biggest gap. Press, mentions, video, site fixes, whatever the data flagged.
3. Retest. Same questions, weeks later, multiple runs each (remember, single answers are noise).
4. Adjust and go again. Beat your own last result. Then do it after the next model update too.

We run content the same way at NAPD: test organically first, small and cheap, find what works, then run it again and try to beat the winner before a dollar of paid budget shows up. AEO is that loop pointed at answer engines instead of ad accounts.

Keep the scoreboard honest

One warning before you spend anything on this. "We appeared in a ChatGPT answer" is the new "we got 100,000 views." It feels great and it pays for nothing.

Lead with the actual goal. Usually that is selling more. Sometimes it is bookings, signups, foot traffic, or awareness ahead of a launch. Name the goal first, then measure AEO against it: are AI referred visitors landing, converting, buying? Visibility in answers is a leading indicator, the same way brand lift is. It feeds the sale. It is not the sale.

Red flags when someone pitches you AEO

• They guarantee you will "be the answer." You now know why that is impossible.
• Their proof is that AI describes your brand correctly. Description is table stakes. Show me the commercial intent queries.
• One flat fee, one deliverable, no retest. That is a project, not optimization.
• Everything they propose lives on your own website. The citations live everywhere else.
• No baseline. If they never asked the machines about you before quoting a price, the plan came off a shelf.
• They report appearances and impressions with no line back to sales or your stated goal.

What good looks like

An honest operator starts by showing you what AI already says about you, tells you the truth about the gap even if the truth is "this is a small fix, you do not need us for much," builds the plan off your data instead of a template, tests in rounds, and reports data proven results against the goal you actually care about. That should not be a premium tier. That should be the standard.

Questions to ask anyone selling you AEO (including me)

1. Can you guarantee my brand appears in AI answers? (Correct answer: no, and anyone who says yes just failed the interview.)
2. What did the engines say about my brand before you built this plan?
3. Which levers are you pulling: press, mentions, sector authorities, video, my site? And why those, based on what data?
4. When do we retest, and how many runs count as a test?
5. What number are we measuring this against, and is it the one I actually care about?

I built and sold a software company before I turned 30, and I spent decades watching what happens when teams ship once and stop looking. The brands that win the AI answer are not the ones who bought a package. They are the ones still testing when everyone else declared victory.

If you want to know what AI says about your brand right now, ask it. Then ask again and watch the answer change. When you are ready to do something about it, talk to us. We will show you the answers before we show you a price.