You have almost certainly had this conversation with yourself already. The work is good. The reviews are good. The website cost real money and the agency sends a report every month that looks fine. And somewhere in the last year you started noticing that new patients arrive having already decided — and there's a good chance your website had nothing to do with it.
The AI Authority Stack™ is how I break down what has to exist before an AI platform can name your practice to somebody who asks. It is four layers, built bottom to top: Foundation, Answer, Proof, Authority. The acronym is FAPA, and yes, it is pronounced FAY-puh.
Most practices invest from the top down. The trouble is that the bottom two layers are what make the top two usable.
What are the four layers?
Here they are, in build order. Bottom of the list is the bottom of the Stack.
Most practices I look at have spent a decade on 3 and 4 and roughly no time on 1 and 2. That is not carelessness. Proof and Authority are the two layers a human being can see. Foundation and Answer are plumbing, and nobody walks into a finished bathroom and admires the plumbing.
Why does a practice with great reviews still not get named in AI search?
Because credibility that isn't connected to the question doesn't get used.
Your reviews may be describing a practice that can't be placed — no clear statement of what you do, no consistent address, nothing on your site that answers the question in the form a patient asked it. The evidence exists. It just isn't attached to anything an AI platform can resolve back to a named practice in a named city.
That is why the Stack builds bottom-up even though the top is what you are proudest of. Foundation and Answer are what make Proof and Authority usable. Without them, your best asset is sitting in a room with no door.
AI can't recommend you for a category it can't tell you're in.
Foundation: can a machine state what your practice is, without guessing?
Layer one
Foundation is the set of plain facts about your practice that a machine can extract without inference: what you do, who you do it for, and where you are — plus a name, address, and phone number that say the same thing in every place they appear.
Read your homepage the way a machine reads it. No photography, no hierarchy, no vibe. Just the language. If the headline is Exceptional Dentistry. Extraordinary Smiles., that is a mood. It is lovely on a wall and it states nothing.
The consistency half is less glamorous and breaks more practices. Your NAP lives in more places than you remember — the Google Business Profile, the site footer, the Yelp listing nobody claimed, the insurance directory from when you were in-network, two association pages, a city guide, and whatever your last agency built without telling you. When those disagree, there is no single version of your practice to retrieve.
This is also the layer where the SEO conversation gets uncomfortable, because keyword optimization does not automatically produce any of it. A page can rank respectably and still fail to say plainly what the practice does. I've written about that distinction at length in GEO vs. SEO: Why Generative Engine Optimization Is the New Priority.
Answer: is there anything of yours that answers what a patient actually asked?
Layer two
Answer is content structured around the real questions patients bring to an AI platform, written in your own name, so there is something of yours available to retrieve and cite.
Patients don't type your specialty. They type situations. How much do veneers cost if my teeth are already crowded? Can you fix a smile without braces first? What's the difference between veneers and bonding for a chipped front tooth? Those are the questions, in roughly those words.
A services page built to rank for "veneers [city]" is a different artifact. It exists to appear in a list. Answering a question in a form that can be quoted back to a patient is a separate job, and most practice sites have never done it. This is the layer my own FAQ page is — it exists to answer, in plain language, what's the best way to improve AI visibility for dentists? and the nine questions next to it.
Proof: do you have evidence, and did you produce it?
Layer three
Proof is the evidence you generate yourself — reviews with substance in them, before-and-after work that shows real transformation, patient stories that contain specifics rather than adjectives.
Most serious cosmetic practices are strong here. If you are one of them, treat this section as a checkpoint rather than a to-do list. What matters at this layer is less the star average than the language — when ten patients describe the same experience in similar words, that repetition is corroboration. I made that case separately in Why Google Reviews Matter More Than Your Website.
If you are earlier than that — a handful of cosmetic cases photographed, reviews that read like general dentistry — Proof is real work and it belongs on the list. Just not before layers one and two.
Authority: why is this the one you can't build?
The outcome
Authority is the one you don't build. It is recognition conferred by other people — press, citations, references from sources that don't answer to you, being taught or quoted by someone who isn't on your payroll. You can commission a website. You cannot commission being the name people use.
That's the honest shape of this framework: three things you build, one thing you get. Foundation, Answer, and Proof are work you can schedule. Authority is what accrues when the first three are true for long enough that other people start repeating you.
Which is why it's the layer most likely to already be sitting unused. A practice can have real standing in its city — the referring dentists know, the study club knows, the lab knows — and still be invisible to a patient asking a chatbot, because nothing underneath connects that standing to a clearly stated, resolvable practice. The recognition exists. It just isn't attached to anything retrievable.
You don't fix that by chasing more Authority. You fix it by building the three underneath it, so the Authority you already earned has somewhere to land.
How do I tell which layers I already have?
Run this yourself, today. It takes about ten minutes and you don't need me for any of it.
- Foundation. Open your homepage and read only the words. Can you extract what you do, for whom, and where, without looking at a single photo? Then search your practice name and compare the phone number and address on the first five results.
- Answer. Write down the three questions patients ask you most often at the chair. Search your own site for each one. If nothing on your site answers them in that form, you don't have this layer.
- Proof. Read your last ten reviews. Do they describe specific treatment and specific experience, or are they four words and five stars?
- Authority. Name three places you're mentioned that you don't own and didn't pay for. Treat this one as a read on how much recognition is already sitting there waiting for the other three to catch up, rather than a task list.
Then ask ChatGPT to recommend a cosmetic dentist in your city — phrased the way a patient would, without your practice name — and read what comes back against those four. If a competitor got named, the question worth sitting with is which layer they had that you don't. In my experience it is rarely Proof.
If you want the version of this focused on why nothing surfaces at all, that's Why AI Search Doesn't Recommend Your Dental Practice.
Do the layers have to be built in order?
No. They depend on each other, which is a different thing.
Plenty of practices already own layers three and four and are missing one and two — that's the common shape, not a failure. The dependency runs one way: the top two layers can't do their job until the bottom two exist. Adding another before-and-after gallery to a site that can't state what you do is spending money on a layer you already have.
There is no rule about it. The layers hold each other up. Foundation makes Answer resolvable. Answer makes Proof and Authority usable. Fix the bottom and the top starts working without you touching it.
Where do the 3 Cs fit?
The 3 Cs are the conditions your practice has to be in. The AI Authority Stack™ is what you build to get there.
Clarity comes from Foundation plus Answer. Consistency comes from Foundation. Credibility comes from Proof plus Authority. Two of the three rest entirely on the bottom of the Stack — the half nobody looks at.
The Cs are a diagnostic: they describe the state of your information. The Stack is the work order. I laid the Cs out in full in The 3 Cs of AI Visibility, and this page is where that argument was heading.
The rest of the named frameworks — the 3 Cs, the Invisible Practice Problem, the Jen Test™, and the patient experience side — are defined together on the Frameworks page.