A pattern that appears repeatedly across the audit data. Consider a group like this one.
Daniel Reyes is the CFO of a 22-location group headquartered in Charlotte. Same brand on every door. Same fee schedule. Same supported clinical scope at every site. And in his quarterly pack, one number that refuses to behave: his best location produces nearly twice the revenue per patient of his weakest one.
He has checked payer mix. Comparable. He has checked provider experience. Comparable. He has run case presentation coaching at the bottom five locations for six consecutive quarters. The spread has not moved.
Every quarterly review ends the same way. Someone says the lower sites need to "present more treatment." Everyone nods. Nothing changes, because the diagnosis is wrong.
If your portfolio has a revenue-per-patient spread that survives every operational fix you throw at it, this article is about the layer you have not looked at yet.
Why Do Identical Locations Produce Different Revenue per Patient?
When a revenue-per-patient spread shows up in a portfolio review, the explanations follow a familiar order. Payer mix. Fee schedule discipline. Provider mix. Case presentation skill at chairside.
Those things matter. Each one moves the number at the margin. But here is the uncomfortable finding: groups control for all of them and the spread persists. Because none of those variables answers the question that actually sets revenue per patient: what kind of patient walked in the door in the first place?
Revenue per patient is not a productivity number. It is a case mix number. A location whose schedule is heavy with implant and cosmetic cases will out-produce a location whose schedule is heavy with hygiene visits, at identical fees, with identical clinicians, at identical capacity. This is the core problem of Portfolio Revenue Optimisation: the spread you are trying to close is not created inside the operatory. It is created in the moment a patient decides which practice to contact, and for what.
And that moment has moved. The Dental Index national practice audit found that AI-referred patients book high-value procedures at 2-3x the rate of other referral sources. A patient who asked ChatGPT or Perplexity to recommend an implant provider arrives having already decided to treat. A patient who found the same location through a generic directory listing arrives wondering whether their insurance covers a cleaning.
Same location. Same doctors. Entirely different revenue per patient. The difference was decided before your front desk ever picked up the phone.
What Is Actually Setting Each Location's Case Mix?
Think of patient discovery in your markets as a ladder of intent. Every rung delivers a different patient, and a different case.
- AI search referrals: the highest-intent rung. These patients researched the procedure before they researched the provider. They ask an AI assistant a specific question ("best implant dentist near me who does full-arch") and book high-value treatment at 2-3x the rate of any other source. Only the locations the AI can see ever meet them.
- Specific-service Maps searches: patients searching a named service and choosing from the map. The audit data shows 82% of dental searches end in a Google Maps interaction. A location that ranks for its high-value services in Maps intercepts decided patients. A location that appears only for its brand name does not.
- Generic, low-intent discovery: insurance directories, "dentist near me," a plan's provider list. For preventive care, 28% of patients arrive through an insurance directory. These patients fill chairs. They rarely fill them with $4,500 cases.
Here is the mechanism that creates your spread: a location that is invisible on the top rungs does not sit empty. It back-fills from the bottom of the ladder. The schedule still looks full. The operations dashboard still looks clean. But every chair-hour is occupied by a patient whose case value was capped at discovery.
Now scale that across a portfolio. The audit data shows 70% of US dental practices are invisible to AI-referred patients, and fewer than 8% score above 65 out of 100 on AI readiness. Inside a 20-location group, that distribution almost guarantees some of your sites are visible where the high-value patients are looking and some are not. Your revenue-per-patient spread is that visibility distribution, expressed in dollars.
| How the Patient Found the Location | High-Value Booking Pattern | Case Mix the Channel Delivers | Revenue-per-Patient Pattern |
|---|---|---|---|
| AI search referral (ChatGPT, Perplexity, AI Overviews) | Books high-value procedures at 2-3x the rate of other sources | Weighted toward implants ($4,500 avg case) and cosmetic ($3,800 avg case) | Highest in the portfolio |
| Specific-service Google Maps search | Strong when the location ranks for the named service; 82% of dental searches end in a Maps interaction | Mixed; tilts high-value where the service is visible | Middle of the portfolio |
| Generic and directory discovery ("dentist near me," insurance lists) | Low; patient is still deciding whether to treat | Weighted toward preventive ($350 avg case) and urgent visits ($650 avg case) | Lowest in the portfolio |
Look at the value gap between the rungs. An implant case averages $4,500. A preventive visit averages $350. That is not a 10% difference a coaching program can close. It is a different order of magnitude, and it is assigned by the discovery channel, not by the clinician.
One more layer makes this urgent rather than merely interesting. The high-value categories are the fastest-growing ones: implant demand is growing 8.5% per year and cosmetic demand 6.8% per year, against 2.8% for preventive. The gap between your visible and invisible locations is not static. It compounds every year the demand mix keeps shifting toward the procedures your invisible sites never get asked for.
"Your revenue-per-patient spread is not a ranking of your clinical teams. It is a map of which locations your highest-value patients can actually find."
What Do the Groups Getting This Right Do Differently?
The groups that close the spread do not have better clinicians than yours. They read a different report.
Instead of ranking locations by revenue per patient and asking why the bottom five underperform, they segment each location's production by discovery channel and ask a sharper question: which rung of the intent ladder is feeding each site? The moment you cut the data that way, the pattern usually stops being mysterious. The bottom locations are not producing less from the same patients. They are being handed different patients.
Then they treat visibility as a portfolio asset, allocated market by market, the way they would allocate capital. Demand is not uniform across your footprint. One market is heavy with implant-age patients researching full-arch options; another skews young families searching for orthodontics ($5,500 average case, growing 5.1% per year). A location positioned generically in an implant-heavy market is leaving its most valuable local demand to whoever is visible for it. Understanding how implant patients search for and choose a provider is worth a finance executive's time precisely because that search behaviour decides which location books the $4,500 cases.
And they stop assuming brand strength covers this. The audit tracks 432,000 AI dental searches per month nationally, and almost none of them contain a brand name. Patients ask about problems and procedures, not about your group. With DSOs already holding 32% of the market, the question in every city you operate in is simply which organisation's location the AI names when a high-value patient asks. Every month, one practice in each market gets that referral. It is either yours or it is not.
What Separates the Groups That Close the Gap From the Ones That Don't?
It is not effort, and it is not spend. It is where they believe the number lives.
Groups that never close the spread treat revenue per patient as an operations metric: something the practice manager owns, coached from the inside, one chairside conversation at a time. Groups that close it treat revenue per patient as a discovery metric: something set in the market, upstream of the practice, and therefore fixable at the level where finance actually operates. That reframe changes who owns the problem. Case mix stops being a critique of your clinical teams and becomes a positioning question: what is each location known for, and can the patients who want that find it?
Once the problem moves upstream, the moves become clear. Six of them, in the order they compound.
Read revenue per patient by discovery channel, not just by location
A patient who asked an AI which implant provider to trust arrives pre-sold: the decision to treat was made before the first call, which is why this channel books high-value cases at 2-3x the rate of others. A patient who arrived from an insurance directory arrives comparing coverage, not committing to treatment. Until your reporting separates these patients, your bottom locations will keep being judged for a case mix they never had the chance to change.
Score every location's visibility for its highest-value service, not its brand name
Patients do not search for your group. They search for their problem. A location can dominate for its own name and be absent when someone in its market asks an AI about implants, which is the only search that carries a $4,500 case. When you test visibility the way a high-intent patient actually searches, the sites at the bottom of your revenue table tend to be the ones the question never reaches.
Match each location's positioning to its own market's demand profile
Implant demand is growing 8.5% per year, but it does not grow evenly across your footprint. In a market full of patients researching implants, a generically positioned location reads as interchangeable, and interchangeable practices collect the low-intent overflow. When a location's visible identity matches what its market is actually asking for, the patients arriving are the ones who were already looking for exactly that.
Give every location one named clinical identity
A patient who arrives with a specific prior reason to trust a practice behaves differently: they question less, defer treatment less, and accept the case they came in expecting. A patient who arrived because a location was simply nearby brings nearby-level commitment. One clear identity per site ("the full-arch practice," "the cosmetic practice for this side of the city") is what converts local demand into a deliberate patient rather than a convenient one.
Make what AI says about each location specific enough to pre-sell
When an AI can only describe a location in generalities, patients treat it as a generality and keep researching. When the answer is specific about what the location is known for, the patient's evaluation happens inside the AI conversation, and what reaches your front desk is a booking, not an enquiry. The audit found practices with fully completed profiles receive 7x more AI-referred clicks: specificity is the difference between being mentioned and being chosen.
Put visibility variance next to revenue variance in the monthly pack
What your leadership reviews monthly is what your organisation fixes. When each location's AI visibility for high-value services sits beside its revenue per patient, the correlation stops being a theory and becomes an agenda item, and the conversation about your bottom locations changes from "coach harder" to "reposition here." Patients feel the result of that shift long before they can name it: the right practice keeps appearing exactly where they are looking.
How Long Before the Revenue-per-Patient Spread Narrows?
Faster than culture change, slower than a fee increase.
Visibility signals build over weeks and months, not days. And case mix follows visibility with a lag, because high-value patients take longer to decide: an implant patient may research for months before booking, which is exactly why the channel that reaches them early matters so much.
The pattern in the audit data is sequential. Visibility for high-value services moves first. The enquiry mix at the repositioned locations shifts next, as the patients arriving start looking like the patients the location is now visible to. Revenue per patient moves last, as those cases complete. These are patterns observed across the data, not a guaranteed timeline for any single group. But the direction is consistent: the locations that become findable for high-value demand stop back-filling from the bottom of the intent ladder.
What does not work on any timeline is waiting for chairside coaching to fix a schedule that was filled by low-intent discovery. The clinical conversation cannot upgrade a case the patient never came in for.
Which brings us back to Daniel Reyes and his 22 locations.
He stopped commissioning case presentation training for the bottom five sites. Instead, he ran one exercise: for every location, he tested what an AI assistant said when asked for the best implant and cosmetic provider in that location's market.
His top locations came back named, described, specific.
His bottom five were invisible for everything except their own brand name.
The spread he had spent six quarters coaching was a visibility map.
So the group repositioned its weakest sites around the high-value demand already present in their markets. No new clinicians. No new fee schedule. Several quarters later, the enquiry mix at those locations had changed, and the revenue-per-patient spread had started to narrow.
Same chairs. Different patients.
If your quarterly pack shows a spread that operations cannot explain, run Daniel's exercise before your next review. The answer usually takes ten minutes to see.
Where Is Your Portfolio's Case Mix Actually Decided?
Here is the conclusion the data keeps pointing to: your portfolio's revenue per patient is decided in the answers AI assistants give and the map results patients tap, before any patient reaches any of your front desks. Positioning each location for its highest-value local demand only produces revenue if that positioning is visible where high-intent patients look. In 2026, they look in AI search and Google Maps first.
That is also why this is a finance conversation and not only a marketing one. Your marketing teams can only amplify the position each location holds; when a location is invisible for its most valuable services, every channel works harder for less return. Fix the visibility layer and the same spend, the same teams, and the same chairs start compounding instead of leaking.
You have built a portfolio worth finding. The open question is whether the patients who would book your highest-value cases can find each location in it. That is a measurable fact, location by location, and it is worth knowing before your next quarterly review.