Your Radiograph Gets Read a Third Time

Researching accounts. Building lists. Writing sequences.

There's a better use of your team's time.

Apollo is the AI revenue engine that handles the busywork, so you can stay focused on selling.

Plus, everything you need is in one place:

Stop doing busywork and start building pipeline, faster.

With Apollo — the AI revenue engine powering 4M+ users.

Sign up for free

Two issues ago: every AI product sold to your side of the chair, and the four machines inside them. Last issue: the payer's copy of one, reading your radiograph against you. This issue: the third reader, and the one with the largest check.

When a DSO or a private-equity-backed group looks at buying your practice, the diligence team asks for your production reports, your payer mix, your lease, your adjusted EBITDA, and a chart audit, which in most of the 2026 guidance is still a few dozen charts pulled at random, radiographs checked against what was planned and completed. That part hasn't changed. What has changed is what a buyer can do with the whole archive. In 2021, Overjet's leadership and executives from Heartland Dental, North American Dental Group, Dental Care Alliance and Smile Brands wrote in Compendium that "every radiograph, both historic and recent, can be analyzed during due diligence." A year later a Pearl-powered diligence shop, SmartDiligence, launched with the stated aim that every DSO acquisition get its analysis before closing. Not a sample. All of it. Whether a given buyer asks for the archive is still deal by deal, and HIPAA lets you hand it over before close only to a covered entity or under a business associate agreement, which is worth reading, because it also governs what the vendor running the analysis gets to keep.

The same class of vision model we spent last issue on runs over the archive and cross-references it against the treatment plans in your practice management system. It produces the one number on the clinical side that a buyer's forward model actually needs: the gap between the disease your images show and the treatment you billed.

That gap has names, none of them flattering. The diligence glossaries call the diagnosed half "unscheduled treatment": care you recommended that never made it onto the book. The undiagnosed half, the part only a machine reading your radiographs can count, has no underwriting term at all. The vendors call it "care opportunities." The bankers call it upside, and they will tell you in writing that a buyer who believes it can create the upside with its own systems "may reserve more of that value for itself." Call it what it is: unrealized production. And the acquirer is pricing it as theirs.

Overjet, to take the vendor that says this out loud on its own pages, described the mechanism in 2023: its AI "analyzes 18 months of radiographs from previous visits and cross-references with treatment plans in the practice management system to identify potentially undiagnosed restorative and periodontal treatment needs." Overjet sells that as a chairside day-planning tool; the 2026 version, Chart Audit, runs 12 months of history and flags conditions "either identified but never scheduled or not identified at all." Point the same model at a target's archive and it is a diligence tool. This is the same Overjet the product guide flagged as Delta Dental-backed and selling to payers. One vendor, three customers, one model.

Its bone-level tool was FDA-cleared in 2021 on 161 films from 63 patients, where its measurements landed within about a third of a millimeter of a dentist-and-radiologist consensus on average (0.31 mm on bitewings, 0.35 on periapicals); Overjet's marketing now calls that "0.35mm precision." That is agreement with human readers, not a ruler, and it measures crestal bone on a bitewing, which a probe never touches. The label itself carries the caveat: cleared to aid in measuring bone levels, not to make or confirm a diagnosis. The point is not that it is finer than your hygienist. The point is that it is the same measurement in your office and in the buyer's 40 other offices, because the same algorithm is reading the same pixels.

The Compendium piece, "How DSOs Can Drive Growth With AI," listed what the archive gets read for: the periodontal health of the patient pool, the proportion of untreated Class II and III carious lesions, the proportion of existing crowns and restorations with recurrent decay and how old they are, the percentage of edentulous spaces that could convert to implants, and the share of posterior teeth carrying large, aging amalgams.

Read that list again from the buyer's side of the table. The chart audit reads your standard of care in both directions: over-treatment that becomes a billing liability, under-diagnosis that becomes a supervised-neglect liability. The model reads something else. Every item on that list is a revenue event that has not happened yet, and what the buyer is inventorying is the production embedded in your patient base that your practice has not harvested, to build a forward margin model on harvesting it under new ownership.

The periodontal line is the one that moves the multiple, so start there. Roughly 42% of dentate US adults 30 and older met the CDC/AAP surveillance definition of periodontitis in NHANES 2009–2014, the last cycles that put a probe in anyone's mouth, and 7.8% had severe disease. That is a full-mouth probing survey of the general population, and the definition counts mild and long-arrested attachment loss. The same survey supplies the honest comparator for a recall pool: among adults who had seen a dentist in the previous six months, roughly who sits in your hygiene chairs, the figure was 30%, with 4% severe. So the honest expectation sits nearer 30 than 42, and both are far above what most practices code.

How much higher now has a published answer. Dental Intelligence's 2026 State of Dentistry report, benchmarks from more than 10,000 practices, tracks the share of active patients over 30 who are in ongoing periodontal care. Bottom-tier practices: 1%. Average: 11%, down from 12% the year before. Top tier: 32%, up from 27%. The report's own gloss is that top practices diagnose and treat periodontal disease in nearly a third of their patients over 30 while the bottom tier barely cracks 1%, most likely by treating perio patients as prophy patients. Grade: one analytics vendor's client base, large but self-selected. Falsifier: your own trailing-12-month number, which I would like you to send me. If your honest answer lands in the low teens you are average, and average is the gap the buyer's model is about to price.

Last issue's tease was the group that bought the practice down the road and changed the recall interval first. This is why. For every patient whose chart supports it, a 6-month prophy becomes a 3- or 4-month perio-maintenance visit, D1110 becomes D4910, and the hygiene line moves before a single new patient walks in.

Buyers price that line on its own. McLerran & Associates' 2026 DSO multiples guide puts the premium at half a turn to a full turn of EBITDA for practices with hygiene above roughly 30% of collections (a turn is one multiple of EBITDA); the investment banks say the same thing without a number and tie it to recall adherence and the hygiene-to-doctor ratio rather than share alone. Below it the published view is a gradient, not a cliff: the same brokers call the mid-twenties to mid-thirties healthy, and the discount language starts around 20 to 25%, where the recurring-revenue story visibly thins. For scale, the bands bankers published this year are illustrative benchmarks, not disclosed deal data: FOCUS Investment Banking puts single-location general practices under $1M of EBITDA at 5–7x as DSO tuck-ins and regional add-ons (a practice bought to bolt onto an existing group) at $1–3M of EBITDA at 7–9x, other appraisers put the small end a turn or two lower, and most offices under $1M are still priced on a percentage of collections rather than on EBITDA at all. On a practice with $700,000 of adjusted EBITDA, one turn is $700,000 of purchase price. At $300,000 it is $300,000, which is still more than the equipment credit you will spend a week fighting over. That is the going rate for the difference between a hygiene department that finds and treats disease and one that polishes and reappoints.

Buyers price hygiene this way because recurring hygiene revenue is the part of a practice that survives the owner walking out the door. A restorative practice built on the selling doctor's hands takes an owner-dependence haircut: brokers' published rules of thumb start at 10–20% off once the owner produces 90% of the work and run to a turn or two once one doctor produces most of the collections, and in 2025 the more common outcome was a buyer who walked, or came back with a five-year employment term and an earnout. A perio-active hygiene department does not depend on the selling doctor's hands. It does depend on hygienists, which is why the staffing line in the term sheet matters as much as the code mix. And it depends on a treatment protocol that closes the distance between what a full-mouth probing finds and what the ledger shows, not one that aims at a percentage. The right treatment rate for your practice is whatever your perio charting says it is, and both the buyer's clinical team and the payer's model will check.

So the vision model and the valuation model are reading the same thing from two directions. The AI reads your radiographs and reports undiagnosed bone loss. The banker reads your hygiene percentage and reports an under-earning department. They are the same fact. One is measured in millimeters, one in multiples.

The buyer's model finds your perio gap, and the buyer does not pay you for it. The offer is a multiple of trailing-twelve-month adjusted EBITDA, what you billed, not what your archive shows, and a thin hygiene line sits on the risk side of that multiple, not the growth side. If you capture any of the upside, it comes through an earnout measured over the two years after closing, not through the headline price. Exposure runs both ways, for the record: supervised neglect is the classic periodontal malpractice claim, and SRP coded to a percentage target is the classic payer-recoupment claim. The chart is the defense against both.

Then the buyer closes and installs the same class of AI across the operatories: Dental Care Alliance rolled Overjet into 400-plus practices, and Heartland put Videa into 1,500-plus in under ten weeks. Clinical leads use the machine's read as the reference they coach providers against, chart by chart, until the perio line moves. That last step is the one the product guide told you never to allow: a hygienist measured against what the overlay found. The buyer allows it because the buyer is not the one holding the license. In vendor-run pilots the lift shows up inside a quarter, and the two years after closing, the window earnouts are written to measure, is when it becomes the new owner's EBITDA. It is the same production that was sitting in your archive the whole time. The buyer bought your trailing EBITDA and will sell the platform's; the broker shorthand is buy the practice at seven, recapitalize the group at fourteen, and the perio production your archive predicted is part of what gets multiplied.

That is arbitrage, not a conspiracy, and it works because the information was asymmetric. The buyer read your radiographs with a machine. You read them with a human, between patients, for years, and billed the prophy.

The remedy is not complicated, but it has a clock on it. Everything a buyer will do to your hygiene department after closing, the protocol, the AI, the coaching, you can do before you go to market, and because buyers price a trailing twelve months, you need at least a year of the result on the books before the LOI for it to count. Start eighteen months out and every turn you recover is yours instead of theirs. The transition advisors who do this all day put the trip from the mid-twenties to the low thirties at 12 to 24 months of real work; Transitions Elite, whose August piece on multiples this section leans on, calls 24% to 32% "about eighteen months and some genuinely hard hiring." The protocol and the re-diagnosis pass move in months. What takes the time is the staffing to run it.

The pathway has an order, and payers enforce it. Comprehensive periodontal evaluation (D0180) first: six-site probing, bleeding, recession, mobility, furcations, staged and graded under the 2017 classification. Then, for patients with pockets and attachment loss, scaling and root planing by quadrant (D4341 for four or more teeth, D4342 for one to three), with the charting and the bitewing attached to the claim, because the payer's reader from last issue will ask for them. For generalized inflammation without attachment loss, D4346, not a prophy. And only after active therapy, periodontal maintenance (D4910). A perio share that jumps from 10 to 40 in one quarter is what the payer's model and a plaintiff's lawyer both look for. Eighteen months is the honest speed.

Your bitewing now has three readers, and each has trained a model on it with a different objective. Your own AI reads it to find pathology, because you pay for sensitivity. The payer's AI reads it to find grounds for denial, because the plan pays for consistency with prior denials. The acquirer's AI reads it to find unrealized production, because the sponsor, the fund behind the group, pays for forward margin. Same pixels, three training targets, and only one of them is on your side of the chair or the patient's.

Stop treating the archive as a storage cost. Two other parties already read it as the most information-dense asset in the building. Read it first.

If a transition is within three years, this is the highest-leverage project in the practice, ahead of the equipment and the associate hire. If you are never selling, it is still the standard of care. Either way the work is the same.

Three weeks ago, after the product guide, I promised a pricing issue: what practices actually pay for AI, per operatory, versus what they were quoted. I've been collecting list prices. What I don't have is invoices, and the only source of those is you. One tap below, and if you're willing, the text box after you vote is where the useful detail lives: which products, the quote, the actual invoice, the renewal term and notice period, and whether you're billed per scan or per chart. Results run the Sunday after there are enough replies to bucket without identifying anyone, and I'll print how many came in; until then the pricing issue runs on list prices and contract terms, labeled as such. No names, no practices, nothing traced back to a respondent.

Still owed, in order: the annual-maximum story promised in July, and the first Bench Test verdict. Both come after the pricing results, and if either slips again I'll say so here rather than go quiet. The three-pillar shape from Issue #2 is paused for the three-reader series; the Diaries return with new numbers.

Disclosure: I'm building a small group of practices myself (that's the Diaries), so I sit on the buy side of this table, and I have a commercial interest in periodontal care. Nothing in this issue names it, links to it, or is sold by it.

No industry sponsors, no affiliate links. The product guide said no sponsors at all; the rule is now no dental-industry sponsors, and the ad at the top is a sales-software company that has never seen a bitewing. It buys adjacency, never a verdict. If this one landed, send it to the colleague who is three years from selling.

— Thad

Sources: Compendium, March 2021, "How DSOs Can Drive Growth With AI" (Overjet with Heartland Dental, North American Dental Group, Dental Care Alliance and Smile Brands executives) · DSOPro, September 2022, SmartDiligence.ai launch · Overjet: the 18-month retrospective (2023), Chart Audit (2026), the 0.35 mm claim and the 2021 FDA clearance summary, K210187 · Pearl Practice Intelligence · CDC, NHANES 2009–2014 (Eke et al., JADA 2018, Table 2 by time since last dental visit) · AAP staging and grading guide · Dental Intelligence, The State of Dentistry 2026 (Perio Patient % Age 30+, benchmarks from 10,000+ practices) · Umbrex DSO diligence glossary ("unscheduled treatment") · FOCUS Investment Banking, "Dental Practice EBITDA Multiples 2026" (table labeled illustrative) · McLerran & Associates, 2026 DSO multiples guide · Transitions Elite, August 2026 · CT Acquisitions · TUSK Practice Sales (offers on trailing-twelve-month adjusted EBITDA) · Large Practice Sales (buy at 7x, recapitalize at 14x) · Dental Care Alliance and Overjet rollout · Heartland and Videa case study · Auxo Capital Advisors. Evidence grade for the valuation bands: broker-reported, no closed-deal dataset disclosed. Falsifier: a 2026 term sheet where hygiene share moved the multiple by less than a quarter turn. Corrections append here, never silently rewrite.

The Practice Ledger is a neutral data publication. Nobody buys a number: sponsors buy adjacency, never editorial. Sources and methodology for every benchmark we publish are available on request.