AEO Checker

Is AEO a Scam? How to Tell a Real Audit From a Pitch Deck

Cal Evans 2026-09-10 9 min read

The message that starts most of these conversations

Somebody sends you a message. Maybe email, maybe a LinkedIn cold pitch, maybe a guy who cornered you at a chamber of commerce mixer. The message says your competitor down the street is showing up in ChatGPT and you're not, and for a monthly retainer they can fix that, guaranteed.

You ask how they know your competitor shows up. They don't really answer. You ask what they'd actually do to your site. Vague words happen: visibility, optimization, AI-readiness. You ask how you'd know if it worked. More vague words.

Here's the uncomfortable part. None of that automatically makes them a fraud. It's entirely possible the pitch is honest and the person just isn't great at explaining it. It's also entirely possible you're about to pay a retainer for something nobody, including the vendor, can actually deliver. The pitch sounds identical either way. That's the problem this post is trying to solve. Not "is AEO real" (it is, this whole product exists because it's real) but "how do I tell a real audit from a guess wearing a nice suit."

Why this pitch is everywhere right now

Answer engine optimization, AEO, sometimes called GEO for generative engine optimization, is a genuinely new field. Not new like "SEO but rebranded." New like the tools measuring it barely existed two years ago.

You can watch the money prove it. Profound, a company that tracks how brands show up in AI answers, raised a $20 million Series A led by Kleiner Perkins in 2025. By February 2026 it had raised a $96 million Series C at a $1 billion valuation, the first unicorn in this exact category. That's not a knock on Profound. They're a real, well-funded, named platform, and none of this is about them specifically.

But watch what happens whenever a category goes from zero to a billion dollars that fast. There's no accreditation body for "AEO expert." No licensing exam. No professional association that can revoke your card if you overpromise. Anybody can print business cards that say GEO Strategist and start pitching by Thursday. Money moving that fast into a field with zero gatekeeping is exactly the condition that pulls in people selling confidence instead of results, because for a little while, nobody's in a position to check their work.

What Google itself actually says about this

Google updated its own AI-optimization guide as recently as July 10, 2026, and it's worth reading the actual sentence instead of a summary of it. Here it is: "Be wary of third-party tools that promise ranking success or claim to use 'internal' Google metrics. No third-party tool has access to our internal ranking or AI systems."

Notice what that sentence does not say. It does not say AEO or GEO vendors are illegitimate. It does not say the category is a scam. What it says is narrower and more useful: if anyone claims access to Google's internal ranking or AI systems, that claim is false on its face, because nobody outside Google has that access. And in the same breath, Google says to be wary of anyone promising you ranking success. Google's actual guidance for evaluating a third-party AEO or GEO service is the same guidance it already gives for evaluating any SEO vendor. Same bar, not a special new suspicion reserved for this one category. That distinction matters, because it means the standard here isn't "avoid anyone who says AEO." It's "apply the same skepticism you'd already apply to anyone promising you Google rankings."

The checklist that actually holds up

Between Google's own warning and a company and a critic who both cover this industry closely, a short list of concrete red flags shows up over and over. None of these require you to understand a lick of the technology behind them.

  • Guaranteed placement. DesignRush's 2026 guide on this puts it flatly: "No Generative Engine Optimization company can guarantee that a brand will appear in AI-generated responses. Generative engines select sources dynamically, and outputs vary by user, prompt, and model updates." Anyone promising a guaranteed result is promising something outside their own control.
  • Vague language standing in for a real answer. Phrases like "guaranteed citations," "instant AI rankings," or "exclusive GEO tactics" sound impressive and mean nothing. Ask what they actually mean by any of them and watch what happens to the sentence.
  • No stated methodology. A legitimate vendor can tell you what gets audited, which pages get prioritized, and how they'll measure whether anything improved. If the answer to "how will I know this worked" is another vague phrase, that's your answer.
  • No mention of the fundamentals underneath. AEO doesn't replace clean HTML, working structured data, and pages that actually answer questions. It sits on top of that. A pitch that never mentions any of it is a pitch built entirely on the new buzzword.
  • Fabricated expertise. Industry critic Nikki Pilkington put this one bluntly: "Nobody is a Generative Engine Optimisation expert yet. Nobody." Her point is about arithmetic, not opinion. AI search has been usable in any real way for roughly two years. Anyone claiming years of deep GEO expertise, or an "exclusive partnership" with an AI platform, or charging five figures a month for a "cutting-edge GEO strategy," is claiming a track record the calendar doesn't support.

None of these five require you to know what structured data is. They just require you to notice when an answer refuses to get specific.

A question worth asking, not a test you pass or fail

Here's one more thing worth trying, with an honest caveat attached before you try it. Ask the vendor to show you their own business actually showing up in an AI answer for the kind of query they're pitching you on, and ask exactly what query they typed to get that result.

This is a real signal. If they can't produce it at all, that tells you something. But don't treat it as a clean pass or fail. AI answers shift by prompt, by session, by which model answered, sometimes by nothing more than when you asked. A vendor who genuinely does good work might still not show up for a broad, competitive term on the day you happen to check. So use this as one more thing to probe, alongside the checklist above, not as the single question that settles it.

The FTC angle, kept honest

This next part isn't about AEO at all, and I want to be upfront about that rather than stretch it into something it isn't. On May 21, 2026, the FTC announced proposed consent agreements with CMG Media Corporation (doing business as Cox Media Group), MindSift LLC, and 1010 Digital Works LLC, over a product called "Active Listening" that was marketed to small businesses as AI-powered ad targeting based on listening to conversations near a smart device. It didn't actually listen to anything. It was a marked-up resale of email lists bought from data brokers. CMG paid $880,000. The other two firms paid $25,000 each. Those consent agreements were proposed in May and went through a public comment period before the orders were finalized on August 26, 2026, and all three companies are now under 20 years of FTC reporting and compliance obligations.

Again, none of that is about AEO or GEO. It's about an unrelated "AI-powered" service sold to the exact same audience this post is written for. What it proves is narrower than "AEO is a scam category." What it proves is that "AI-powered service sold to small businesses that quietly doesn't do the thing it claims" is a pattern regulators are actively finding and actively punishing, right now, in 2026. Not a hypothetical risk somebody's warning you about for fun.

What a checkable methodology actually looks like

So what does the opposite of a vague pitch look like in practice? I'll use our own product, because it's the one I can describe honestly without guessing at somebody else's internals.

The score you get from an AEO Checker report doesn't come from an AI model guessing at how impressive your site seems. It comes from six modules, Structured Data, Content Structure, Entity & Trust, AI Crawler Access, Technical Health, and llms.txt, each scored by fixed point rules written in plain PHP. The AI Crawler Access module checks robots.txt against five named bots, GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, and Bingbot, by name, not by vibe. Every individual check inside every module comes back PASS, WARN, or FAIL, with a specific finding attached, not a vague "improve your AI visibility" score with nothing under it.

Now the part I'm not going to leave out, because leaving it out would make this post fail its own test. We do use the Claude API. Just not for the score. Once the score is already locked in by those deterministic rules, Claude writes the fix instructions in plain English, generates the copy-paste assets like your llms.txt file and schema markup, and writes the narrative sections of the report so a non-technical reader can act on them. The number itself never touches an AI model. The words explaining what to do about that number do. That's the honest shape of it, not "no AI here," but "the part you'd want to verify is the part that doesn't depend on a model's mood that day."

The actual test

None of this checklist requires you to become an expert in query fan-out or embeddings or whatever term shows up in next month's pitch deck. The test was never "does this person sound confident." Confidence is free and anyone can generate it. The test is "can you verify a single thing they just told you." A guaranteed placement, you can't verify, because nobody can guarantee it. A vague methodology, you can't verify, because there's nothing underneath it to check. A named crawler getting checked against your actual robots.txt, you can verify yourself, in about a minute.

If you want to see what that looks like on your own site before anybody pitches you anything, the free Grade Check on aeocheck.net scans one page and hands you a real score off the same scoring engine behind our paid reports. It's free, it takes about a minute, and the number comes off the same fixed rules, not a model's opinion of your site. Run it, and you'll have something concrete to hold the next pitch up against.