Yes, AI visibility is worth paying for when a practice has a credible website and new patients who compare practices before booking, and it is not worth paying for when either of those is missing. The decision turns on one figure: what one new patient is worth to you over a year, not on how many times a model mentions you.
Plenty of practices run this backwards, because mentions are visible and revenue is not. This article works the other way round: one new patient, and the share of its value an AI answer can touch.
One new patient
What one new patient is worth to you over a year.
What AI can touch
The share of that value an AI answer can reach.
Fund it, or not
Only when the numbers support it.
Two inputs come from your own records: what one new patient is worth over twelve months, and how many new patients a month arrive in a market where people research first.
Start With One New Patient, Not With Mentions
The unit of value in a dental practice is a new patient who stays and comes back. Mentions are inputs, and inputs nobody can price are inputs nobody should fund.
Price one new patient using the money you collected in their first year, not the price list. List prices rarely survive adjustments, and an AI answer rarely decides a decade of care.
The Payback Sheet
Fill this in with your own numbers. Every cell is something you can look up or ask for, and the last line answers the title question.
Swipe the table sideways to see every column.
| Line | Your Figure | Where It Comes From |
|---|---|---|
| 1. New patients a month | Your figure | Report, 90 days, divided by three |
| 2. Share who research before booking | Your figure | Front desk, out of ten |
| 3. Share of those who asked an assistant | Your figure | Same conversation |
| 4. Share of the answers you could win | Your figure | Measured, not assumed |
| 5. New patients from AI each month | Line 1 multiplied by lines 2, 3 and 4 | Multiply down |
| 6. Value of one new patient, year one | Your figure | Collections in year one, over new patients |
| 7. Value AI brings each month | Line 5 multiplied by line 6 | Multiply |
| 8. What you spend each month on the work | Your figure | Your quote or your hours |
| 9. Months to payback | Line 7 divided by line 8 | The answer to the title |
How To Work The Sheet
- Count the new patients. Take the practice management report for the last 90 days and divide by three. Use completed visits, not booked appointments, because an appointment is not yet a patient.
- Price one new patient. Sum what new patients actually paid in their first twelve months, then divide by the number of new patients in that cohort. Collections, not billed fees.
- Ask the two questions. At booking, ask every new patient whether they looked you up before calling, and whether they asked an AI assistant about dentists in the area. Record the answers out of ten, then write them as decimals: eight in ten is 0.8.
- Measure line 4, then multiply. Run your question set, count the answers where your practice is named and recommended, then divide by the answers you could win. The method is in the article on how to measure AI visibility. Line 7 over line 8 is your payback period.
These figures are invented to show the arithmetic. A practice with 40 new patients a month and eight in ten researching first has 32 candidates. Winning one in five of the answers it could win gives 24 new patients a year.
The Three Questions To Answer First
Before any budget discussion, answer three questions from records. A weak answer to the first makes the other two irrelevant.
How Many New Patients a Month?
Pull it from the practice management report, not from memory, and use the last 90 days. A practice taking twenty new patients a month and one taking a hundred are different investment cases.
What Is One New Patient Worth Over a Year?
This is the number most practices cannot produce on request, because it sits in three places at once: the practice management system, the bank account and the head of the practice.
How Much Of The Pipeline Is Local Intent?
Local intent is the share of new patients who chose you because you are near them. AI answers touch local intent first, because an assistant answering a question about dentists in this area is serving someone who intends to come in soon. Ask the front desk to sort the last thirty new patients.
Where The Value Actually Comes From
Value comes from category coverage, not from winning one big answer. One answer is one chance. Six answers across six questions is six chances, typed by six different patients.
This is why chasing position is a dead end. There is no position one in a paragraph. The citation research agrees: content with specific numbers is more likely to be quoted (the Princeton GEO study by Aggarwal and colleagues, KDD 2024).
A score with no local comparison attached is close to decoration. The article on AI visibility benchmarks explains how to build one from your own city. The useful question is which categories of question you own, and how many are open questions where nobody is named.
The Cases Where It Is Worth It
Three conditions, and all three have to hold at once.
A Market Where Practices Compete For The Same Patients
If six practices in your area have websites and treatment pages, the answer to a local question has a list to choose from, and being on that list is worth something. If you are the only practice within a reasonable drive, it matters much less.
Treatments That Get Researched Before They Get Booked
Implants, orthodontics, whitening and full mouth work are decisions that take weeks, and a patient doing research is normal for that kind of spending. A practice built on walk-ins has less for the same work to act on.
A Website That Can Carry The Work
Visibility work makes a good website easier to find and a poor website slightly easier to find. The work multiplies a site that already has treatment pages with specific facts, a price range somewhere and reviews to point to. If none of that exists, the first job is the website. drsmiles.ai runs the full audit across forty questions and four engines and returns the raw answers this sheet needs.
The Cases Where It Is Not Worth It
Four situations where the honest answer is no, and spending more does not fix any of them. One of them is having nobody to own the work, since visibility work without a monthly run and a named person decays into a bill.
No Website, or a Site an Engine Cannot Read
A missing site is not a visibility problem but a business problem. The same goes for a site that blocks search crawlers. Google states in its AI features optimization guide that it can process JavaScript, so a scripted site is a quality problem, not an access one.
No Reviews and No Plan To Fix That
Local answers lean on what other people say, and the directories carrying those reviews are read widely. Without a steady flow of genuine reviews the work has no evidence to point at. The FTC rule on consumer reviews and testimonials shows that incentivised or written reviews create a legal problem, not a ranking one.
A Market That Sends Patients By Referral or Panel
If patients arrive because a relative recommended you, or because your name is on a panel they were locked into, the questions they ask are about availability, hours and insurance rather than which practice is best. Visibility work targets the second kind of question.
How To Measure The Payback Honestly
Baseline Before You Spend
Run the question set before any work starts and keep the raw answers. Without a baseline you cannot show movement, and movement is the only evidence that matters. The protocol is in the article on measuring AI visibility.
Re-measure With The Same Wording
Same questions, same city, same day of the week, same time window, same number of runs. Change the wording and you are measuring a different thing. Re-run monthly on a fixed date.
Attribution You Can Actually Get
Expect attribution to be thin. An AI answer rarely hands over a referral the way a search result does. What you can get is a front desk question: ask every new patient how they found you, and give them an option that names an AI assistant.
Do Not Count Traffic As Value
Page views are the wrong unit in both directions. They understate the effect, because plenty of people read an answer and never visit, and they overstate it, because a visit from someone checking whether you are open is not a new patient.
If you would rather buy the work than run it yourself, the article on hiring an AEO agency sets out the threshold and the contract terms worth insisting on. The pricing and terms FAQ covers what a quote includes.
- I can state what one new patient is worth to me in collected revenue over twelve months.
- I know how many new patients arrive each month and how many chose us for local reasons.
- My website has pages an engine can read, with facts a patient can check for themselves.
- We collect genuine reviews steadily, with nothing in the process that would breach the review rules.
- A named person runs the same question set on the same date every month.