MPI have been doing search in Louisville since the work involved arguing with people about whether meta keywords mattered. They did not. I was right about that one and wrong about plenty of others, which is roughly the trade you make for staying in a field long enough to watch it turn over three times.
The Part Where I Tell You I Saw It Coming
In early 2025 the shape of the thing became obvious. Not to everybody, but obvious. Google started answering questions instead of listing links, people started opening ChatGPT instead of a search bar, and the entire discipline I had spent two decades on quietly changed what it was optimizing for. Ranking was becoming a means to an end, and the end was being named in somebody's answer.
So I started Agency AEO and pointed it at that. Not as a pivot or a rebrand, just a plain acknowledgment that the target had moved and the old aim was going to miss. For a while that felt early in the way early feels before it works, which is to say it felt like being wrong.
Then the phone rang from Texas. Then South Carolina. Then Washington, D.C. Then somebody in England filled out the contact form. All of them had typed some version of "AEO agency" into Google and found us sitting at number two nationally, on a site that at the time was exactly one page long. That is not a brag so much as an indictment of how few people were trying.
Watching The Catch-Up
What is happening now in Louisville, and everywhere else, is the catch-up. Agencies that spent 2025 telling clients AI search was a fad spent 2026 adding an acronym to their service menu. You can watch it happen in real time on their websites: a new page appears, the word GEO shows up in a nav that had four items last quarter, and the case studies underneath it are all from 2023.
I am not going to pretend that annoys me. Competition is fine and most of those firms are good at what they were already good at. What is worth saying plainly is that adding a page is not the same as doing the work, and the work is mostly not on your website at all. If you want the argument for that, we laid it out in where AI actually gets its answers, and it is uncomfortable reading for anybody selling content packages.
Which brings me to the part of this article that is either a good idea or a bad one.
We Graded Ourselves
We sell an AI visibility audit. Five pillars, a grade on each, no partial credit for trying. It felt dishonest to keep running it on other people without running it on ourselves, so here it is, graded in September 2026 against the live site, with nothing pulled.
Site audit comes back 100 out of 100, zero errors, zero warnings across 20 pages. PageSpeed SEO score 100. Sitemap live with 11 URLs and declared in robots.txt. An llms.txt file at the root, which the evidence says the AI companies ignore, but it costs nothing and we would rather be wrong cheaply than right expensively.
One Organization, one WebSite, a ProfessionalService with full address and hours, 12 services and 10 questions answered in structured data, validating at zero errors and zero warnings. This is the pillar we are best at, which you would hope, since it is the one we lecture people about.
Nine researched articles, cross-linked, each carrying article, FAQ and breadcrumb markup. Good work. Also: we have no pricing page, no about page and no service pages. The site is a homepage and a blog wearing a trench coat. A prospect who wants to know what we charge currently has to email and ask, which is exactly the friction we tell clients to remove.
Here is the bad one. We are not listed on Clutch, which in our own category study was the single most-cited source across 224 AI answers. No Bing Places listing at all. Apple Maps has us unclaimed and at the wrong street. No company page on LinkedIn. No social accounts. Our Google Business Profile is verified and properly built out, and it is carrying this entire pillar by itself.
We track this monthly across ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews, which already puts us ahead of most of our category, because most of our category is not measuring at all. It is a B and not an A for one reason, and it is the reason most tracking is worthless: the prompt list is the measurement. Watching whether AI names you for broad industry questions tells you nothing if your buyers ask something narrower. Ours is built from the questions that actually produce our inbound, and it gets rebuilt as the category changes.
What The D Actually Means
Strip the theater off that report card and it says something specific: our paperwork is immaculate and our corroboration is thin. We built a beautiful, machine-legible, technically perfect website, and websites are the half of this job that agencies are comfortable doing because it is the half they control.
We are excellent at the part we control and mediocre at the part we have to earn. So is almost everybody.The honest summary
The uncomfortable arithmetic is that a business with a mediocre site and forty recent reviews on three directories will usually get recommended over a business with a perfect site and none. Machines weight evidence over claims, and a directory listing with real ratings on it is evidence. Your homepage is a claim. We wrote the long version of that in how AI picks local businesses, and then we went and got a D in it.
What We Are Doing About It
Publishing a bad grade and then not fixing it is just content marketing. So, in order:
- Clutch first. The most-cited source in our category and we are not on it. The free listing goes up, then the reviews, then we decide whether their paid tier is worth it. Their own methodology says paying does not raise your rank, reviews do, so the order matters.
- The missing pages. Pricing, about, and real service pages. If we are going to tell people to publish their prices, we can publish ours.
- Bing and Apple. One listing does not exist, the other has our address wrong. Both are ten-minute fixes we have been too busy to do, which is the most ordinary reason anything stays broken.
- Keep the prompt list honest. Buyer questions shift as a category matures. Ours gets rebuilt against the searches that actually produce inbound, not a template somebody shipped with a dashboard.
We will re-run this same audit in ninety days and publish it again, including whatever we fail next time. If the D has not moved, that will be on the page too.
Go Grade Yourself
The five pillars above are not proprietary. Here they are as a scorecard you can fill in right now, for your own business, in about four minutes. Be as unkind to yourself as we were.
Does your site load fast, work on a phone, have a working sitemap, and pass a basic SEO audit with no errors?
Is your business name, address, phone and service list identical everywhere it appears, and is any of it in structured data a machine can read?
Does a page on your site answer each of the top questions a buyer asks before hiring you, in the first paragraph, including price?
Are you listed, completely and with recent reviews, on the directories and communities that get cited in your category?
Do you actually know, this month, whether ChatGPT and Google AI Overviews name you when somebody asks for a business like yours?
Four or five honest yeses and you are ahead of nearly everyone. Two or fewer and you are where most of the market is, which is also where the opportunity is. The 20-minute AI visibility audit turns these five questions into an actual number, and it is free and takes exactly as long as it says.
And if you want the version of this where somebody else does the grading and then fixes what is broken, that is the whole business. But start with the scorecard. You may find you need less help than you feared, or more than you hoped, and either one is better than guessing.
Questions People Ask
Why would an agency publish its own bad grade?
Because a report card where everybody gets an A is worthless and readers know it. The useful version shows a real gap and what is being done about it. It is also the fastest way to demonstrate that the audit we sell is a real instrument rather than a sales prop.
What are the five pillars of an AI visibility audit?
Technical foundation, entity clarity, content, third-party corroboration, and AI citation measurement. Together they cover whether a machine can find you, identify you confidently, quote you, verify you through other sources, and whether you know any of it is happening.
What does a D in third-party corroboration mean?
It means the business is not present, or not present with recent reviews, on the directories, review platforms and communities that AI systems actually cite in its category. It is the most common weak pillar and usually the highest-leverage one to fix.
Is Agency AEO really ranked number two nationally?
We reached number two on Google for the search AEO agency in 2026, which produced inbound from several states and from England while the site was still a single page. Rankings move, so treat it as a fact about a moment rather than a permanent condition.
How often should a business re-run this audit?
Quarterly. Pages that are not refreshed roughly every quarter are meaningfully more likely to lose their AI citations, and third-party listings drift constantly as reviews age out.
Do you work with businesses outside Louisville?
Yes. The work is digital and our clients are nationwide. We are based in Louisville, we know this market well, and we serve the region directly.