I Ran an AI Visibility Audit on My Own Business and Scored 22 out of 75

TL;DR: I ran an AI visibility audit on my own business and scored 22 out of 75 — findable, but AI was describing a company I closed five years ago. This post is why that happened, why it matters more than the traffic numbers suggest, and exactly how to raise your flag so AI finds you, understands you, and recommends you. Fair warning: AI finds you faster and fact-checks you faster. I'll show you the claims of my own that didn't survive.

First, why I cared enough to audit myself

Kim and I have been believers in long-form, searchable content for our entire careers. And the way I've always explained it is the Energizer Bunny.

Remember those commercials? Every other battery brand's toy runs for a few seconds and quits. The Energizer Bunny is still going. That is exactly long-form versus short-form content. The average social media post has a lifespan of maybe twenty-four hours — a sprint for relevance, and then it's gone. (TikTok stretches that a little, and there are exceptions.) But a blog post, a YouTube video, a Pinterest pin? We have videos that are almost a decade old still pulling views-per-hour today. That's the Energizer Bunny. It keeps running long after everyone else's content has died.

We didn't just believe it — we bet our businesses on it

This isn't a theory I read somewhere. It's how we made our money.

I grew a moving company by outranking everybody — surfacing content that was relevant when no one else was bothering, and leapfrogging nationwide players a fraction of our size. Kim posted her first YouTube videos and got discovered by a corporation that hired her, which led to her building an entire social media department from the ground up — managing over 150 accounts at once for a single client — which led to her launching her own business. Our companies would not exist without long-form content. Literally.

Can you win with short-form? Sure — if you're willing to out-hustle the bunny every single day, forever. And AI has made pumping out that content easier than ever. But ask yourself the real question: why wouldn't you also build content that lives for years, compounds over time, and can be re-optimized whenever you want?

Which brings us to AI.

What AEO actually is (and why it's not a new game)

When this whole conversation started, a term showed up that you may not know yet: AEO — Answer Engine Optimization. Think of it as SEO for AI. Where SEO structured your content so humans could find you through a search engine, AEO structures your content so AI can find you, crawl you, and — the important part — recommend you to other people.

I'll do a whole separate breakdown on AEO versus SEO, because the differences matter. But the core essence is simple: make your content discoverable and friendly for AI to both find you fast and recommend you confidently.

And here's the reframe I want you to sit with, because it's the whole point: this is not a reinvention. It's the next evolution of the exact game we've already been winning for fifteen years. Search engines rewarded us for long-form, searchable, helpful content for over a decade. AI is about to reward the same behavior — harder.

Why this matters more than the traffic numbers suggest

Here's the shift most people haven't clocked yet.

Back in the day, a human being was never going to read four hundred of your blog posts or watch two hundred of your YouTube videos. They consumed your content in silos — one post, one video, then gone. But guess what AI can do in a fraction of a second? Read all of it. Every blog, every video transcript, every page. And then turn to a stranger and say, "This business is exactly what you're looking for."

That's the unlock. The long-form library you built for humans becomes the corpus AI uses to vouch for you.

And the timing is the opportunity. Right now, AI referrals are only about 1% of website traffic (Conductor, 2026). Sounds tiny — until you see the slope. AI referral traffic has grown roughly 16x in two years (SE Ranking) and is climbing faster than any other channel — by one measure 165x faster than organic search (Superlines). Already, about 75% of Americans use AI search weekly (Yext). And the people it does send you are better: AI-referred visitors are roughly 2x more engaged than standard visitors (Semrush), and brands that show up in ChatGPT recommendations are 2.5x more likely to get a site visit within a week (Similarweb).

That 1% is not "too small to matter." That 1% is the SEO-in-2009 moment — the window where getting in early, before everyone else, is the entire advantage. The GEO market is projected to grow from under $1B to over $33B by 2034, and more than half of US marketers say they're implementing it within the next few months (Superlines). The crowd is coming. You want your flags up before they arrive.

The pendulum is already swinging back toward humans

Here's the part that should make you optimistic, not anxious.

A couple of years ago, everyone was going to fire their teams and replace them with AI. Even Mark Zuckerberg was out front on that. Now? He's predicting an AI-driven job boom. The pendulum swung, and here's why: AI is a thought tool — you'll hear me say that a lot — but people are increasingly finding that it doesn't have the right answer for them. Worse, AI is starting to reference AI-generated content and feed it back to the next person. I joke that it's getting high on its own supply.

And the data backs the joke. AI search engines fail to retrieve correct citation information more than 60% of the time (SearchSignal). Only 10% of people trust the first AI result, and 48% verify answers across multiple platforms (Yext). So what happens? People come back out into the world looking for a real human they know, like, and trust — someone who actually does what they do — to cut through the noise.

The gaps are where you win. Legal, health, finance — those are easy layups; AI actively hands those off to humans. But the bigger opportunity is niche expertise. I have a friend in a niche industry, with a niche offering, for a niche person — a gap AI simply can't cover well — who told me he gets "more leads than he can handle" and doesn't "even need marketing." Now, as a marketer, that's deeply offensive, and I told him so (we laughed). But the real lesson underneath it: he filled a gap AI can't, and AI is now sending him people. As time goes on, more of those gaps will either become obvious or get sought out. In the meantime — raise your flags.

So how do you raise your flag?

You signal, clearly and consistently, three things: who you are, what you do, and who you do it for. And the best news of the entire post: a highly trained AI thought partner — like mine — can do about 95% of the lift.

I know, because we did all of the following exclusively together, over a single weekend sprint. Here's what that looked like — do it yourself, or hand it to your developer:

Tell the AI crawlers they're welcome. A robots.txt file that explicitly names GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and the rest. Most sites either block them by default or say nothing.

Add an llms.txt. An emerging convention — a plain-text map of your site written for language models. Build it from the same source as your sitemap so it can never drift out of sync.

Check the layer above your website. This is the one almost nobody knows exists: since mid-2025, Cloudflare blocks AI crawlers by default on new sites, at the network level — upstream of your robots.txt. Your site can look wide open and still be invisible, because the block happens before the crawler ever reads your permissions. Check it. (This is just a fact of how the setting ships, not a complaint about Cloudflare.)

Deepen your structured data. Not "add schema" — deepen it, especially the credibility layer: your credentials, a real Organization entity, your product as its own defined thing. And give your named method its own definition the machines can read — name it, define it, credit its creator. That turns a phrase you say on sales calls into a real, machine-readable thing.

Publish your method on a canonical page. Most businesses have a named framework they use in conversations and never publish anywhere indexable. To an AI engine, that method does not exist. Put it on one authoritative page. (Bonus: nobody competes for your own term — this is the rare case where a single page can own a phrase outright.)

Answer the real questions, as answers. Write the actual questions your buyers type into ChatGPT, and answer them in the plain, helpful voice an assistant would use when recommending someone — then place each one on the page where it genuinely belongs, not dumped into one FAQ block.

Make your bio identical everywhere. Same facts, same phrasing, every platform. This feels wrong — most people write a fresh, "more polished" version per platform. That's exactly what keeps your entity data fuzzy: many slightly different descriptions read to a machine as many loosely-related things. Consistency is the mechanism.

Choose specificity over aggregates. "More than 40 bestselling authors" is unverifiable and gets discounted. "Jack Canfield, Marie Forleo, Daymond John" are entities an engine can actually resolve. Names beat counts — for machines and for skeptical humans. (Hold onto that one. It's about to bite me.)

And then I pointed all of it at myself

Here's where it gets uncomfortable, and useful.

I didn't guess at my score. I tested it — fifteen queries across five categories — and landed at 22 out of 75.

• Clarity: 9/25. Findable, but described as the wrong person.
• Crawlability: 8/25. Infrastructure excellent — and two days old, so nothing was indexed yet.
• Credibility: 5/25. Real credentials, almost none of them machine-readable.

Three findings, and every one of them is probably true for you too.

My method didn't exist. I searched the exact phrase for our framework — the thing the entire business is built on — and got zero results referencing me, Kim, or the company. Ten results, all generic articles from big publications. The foundation of my positioning did not exist as far as the index was concerned.

AI was describing a business I closed five years ago. My name pulled up bios from a company I ran half a decade back — old interviews, old directories, my own outdated copy quoted back at me. My website was not the primary source of truth about me; other people's summaries of my 2021 self were. So AI didn't say "I don't know." It confidently described the founder of a company I no longer run. If you've rebranded or repositioned in the last five years, assume AI is doing this to you right now — and remember, a confident wrong answer is harder to fix than a blank one, because the old data has years of corroboration and your new positioning has however long your new site has existed.

And I did it to myself. The single most-cited stale source wasn't some third party. It was my own Apple Podcasts description, still naming me as founder of the old company — and Apple feeds dozens of downstream directories. Nobody did this to me. I did it by not updating things.

The lesson that cost me the most

Now the part I debated including, because it's about me getting caught — but it's the most valuable thing in this whole post.

When I audited my own claims, a couple didn't survive. My bio said I'd worked with "40+ New York Times bestselling authors." Another said we hit "Top 5 on iTunes worldwide." Here's the honest truth: those weren't lies. The "40+" was a number I ballparked in my head, thinking about the guests I'd had on my show. The "Top 5" was actually Top 10 in New & Noteworthy — a promotional shelf for recent launches, not the main chart. They were good-faith estimates. Spitballed. And in a pre-AI world, that was fine — because precision was never being enforced. Nobody was fact-checking your about page in real time.

That world is gone. And here's the hard part: AI treats "technically imprecise" as "technically wrong." It will call your BS a hundred times faster than it finds you — and it'll say so, to the person who just asked it about you. You can no longer tuck an audacious claim onto a blog page nobody visits, because AI reads every page. So the exact same speed that lets AI vouch for you is the speed that lets it discredit you. Coherence and precision across your entire web presence just went from "nice to have" to load-bearing.

Kim and I have said this to each other for years: be careful what you put on the internet, across your socials or anywhere else — because someday it surfaces. "Someday" is now, and it surfaces fast. The fix isn't to say less. It's to get precise — and to have a thought partner audit your claims for accuracy before AI does it for you, in public. (I'll do a full episode on that auditing process, because it deserves one.)

What to do this week

You don't need to be a household name — you need to be legible and accurate. Start here:

1. Search your own name and your own method in ChatGPT, Gemini, and Claude. Note whether you appear, how you're described, and whether it's the current you. That's your baseline.
2. Make your method exist on one canonical, defined page — and make your entity (who you serve, what you do, where) unmistakable.
3. Check the layer above your website — your Cloudflare AI crawler settings — because your permissions don't matter if the gate upstream is closed.
4. Audit your claims for precision. Every number, every superlative. Tighten it before AI tests it for you.

You're not behind on this. Almost nobody has figured it out yet — which is exactly why doing it now, deliberately and accurately, is the advantage.