DX / Differentials / PE PLATFORMS & ROLLUPS

D-01 — PATIENT ACQUISITION COST · PE-BACKED PLATFORMS

Cost per new patient is climbing — PE platforms

Blended CAC is rising and nobody can say why

DIFFERENTIAL — LIKELY CAUSES, MOST LIKELY FIRST

1. 01

   There is no benchmark for the number to be "up" against — the platform is comparing a moving denominator to itself

   Blended CAC moves whenever the site mix, the specialty mix or the acquisition mix changes, with no channel performing worse. A single top-line CAC across a changing site set is arithmetic, not a signal. The readable version is cost per new patient by site, on one definition, built from the ledger.

   Across a 48-source harvest — advisory firms, transaction databases, valuation practices and public filings — no source publishes CAC, LTV or payback benchmarks for PE-backed healthcare platforms. Sponsors compute them internally per platform and never disclose. This is the single largest gap in the vertical, and it is a gap in the market. Anything found publishing these numbers is, on present evidence, manufacturing them.

   ADMEN PE/rollup source harvest, 48 sources, 2026-07-25

   NOT THIS IF — CAC is already computed per site and per cohort on a fixed site set, and the rise persists inside that set. Then it is real and the cause is below.

2. 02

   A scale effect running in reverse — add-ons import the high-CAC cohort

   Acquisition cost per patient is structurally higher at smaller sites. A platform that grows by buying sub-scale practices raises its blended CAC mechanically, by arithmetic, while every individual site holds its own cost flat.

   Cain Watters, compiled from actual client books in pediatric dental: cost per new patient averages $26.46 — but $44.47 at practices under $1.65M in collections against $25.42 at practices above it. That is 2.97% of collections versus 1.23%. The sub-scale cohort costs roughly 1.75× more per new patient. This is the only real acquisition-cost benchmark located anywhere in healthcare, and it comes from a dental-CPA channel rather than an association.

   Cain Watters (dental CPA practice-financial data, from client books)

   NOT THIS IF — Acquired sites' cost per new patient converges on the platform average within their first four quarters. Then you are integrating faster than you are diluting.

3. 03

   The numerator is flat and the denominator is falling — conversion, not cost

   Cost per inquiry is unchanged; cost per booked, shown and treated patient rises, because fewer inquiries survive each stage. Every stage that leaks inflates CAC without any media price moving.

   Patel et al., Journal of General Internal Medicine 2018: of 103,737 referral scheduling attempts inside a single integrated health system, only 34.8% completed and 38.9% had no documented status at all. Henry Schein One's Catalyst Index shows the same mechanism at the practice level — 7-day versus 23-day time to appointment, 45% versus 75% case acceptance. In cosmetic derm, Zenoti's CY2025 platform data shows rebooking within 24 hours at 40% average against 69% among top earners.

   Patel et al., J Gen Intern Med 2018 (verified at source); Henry Schein One 2026 Catalyst Index; Zenoti CY2025 medspa platform data

   NOT THIS IF — Cost per inquiry is rising in step with cost per treated patient. Then the auction really did get more expensive and this is a media-market problem.

4. 04

   Market density — sites are now bidding against each other

   A platform that has consolidated a geography ends up buying the same demand twice, at auction, from itself. CAC rises with owned density and no external competitor has to do anything.

   PESP's Private Equity Hospital Tracker, built on CMS Medicare-enrolled hospital and self-reported ownership files, shows how far this goes: 447 PE-owned hospitals as of June 2026, 9.5% of all private hospitals and 26.8% of proprietary for-profit hospitals. Texas holds 92 — 20.3% of the state's private hospitals. New Mexico holds 16 — 42.1% of the state's private hospitals. Density is real and it is state-specific.

   PESP, Private Equity Hospital Tracker (2026 update, CMS December 2025 data), read at source

   NOT THIS IF — Cross-site patient-origin ZIP overlap is negligible and no two owned sites appear in the same paid auction. Then density is not your cost driver.

5. 05

   Mix — CAC is fine, the acquired unit is simply worth less

   The cost to acquire a patient did not move; the revenue that patient generates did. Payback stretches because LTV fell, and the operating partner reads it as a marketing failure.

   CMS Medicare Physician & Other Practitioners PUF, CY2024, across 12,616 dermatology-typed providers: the top-decile dermatologist bills 4.1× the median on 2.7× the patients. The gap is service mix, not panel size — Mohs first stage pays $581.52 against $6.22 for actinic keratosis destruction beyond the first lesion. Allowed amounts run $26,773 at the 10th percentile, $188,072 at the median and $777,069 at the 90th.

   CMS Medicare Physician & Other Practitioners Public Use File, CY2024 (federal claims data)

   NOT THIS IF — Revenue per encounter, per case or per chair is flat across the site set. Then the acquired units are worth what you thought and the problem is upstream.

HOW TO TELL THEM APART

How to tell these apart in your own numbers

Each of these is a measurement you can run yourself, without us.

WHAT RESOLVES EACH

What resolves this, and how you will know it resolved

There is no benchmark for the number to be "up" against — the platform is comparing a moving denominator to itself	Rx 04 · marketing attribution →	Site-level CAC on a fixed cohort, with cohort payback, reconciled to the income statement. Because no external CAC benchmark exists in this vertical, the only defensible comparison is your own frozen cohort over time — build that and you have something diligence can test.
A scale effect running in reverse — add-ons import the high-CAC cohort	Rx 04 · marketing attribution →	First make the mix visible so the board stops reading arithmetic as decay. Then Rx 03 · intake-and-ops is what actually moves a sub-scale site toward the larger-site cost structure — it gets there on conversion, not on spend.
The numerator is flat and the denominator is falling — conversion, not cost	Rx 03 · patient conversion →	What does not work: more spend. Against a 34.8% completion rate, doubling spend doubles the leak at full price. See also /marketing-problems/private-equity/new-patient-volume/.
Market density — sites are now bidding against each other	Rx 01 · patient acquisition →	Deduplicate the auction, geo-fence by catchment, and route shared demand to capacity rather than to whichever site bid highest. Real work, and unglamorous.
Mix — CAC is fine, the acquired unit is simply worth less	not a marketing engagement, mostly →	Acquisition can shift which patients arrive; it cannot change what a provider is credentialed, trained or contracted to perform. A 4.1× billing gap driven by Mohs versus lesion destruction is a clinical, credentialing and payer-contracting decision. Say so to the sponsor before they spend against it.

HOW THIS DIFFERS BY SCALE

How this differs by scale

Single site	One site: CAC is a real, computable number and needs none of this. The specialty-level differentials are the right read — /marketing-problems/dental-groups-dsos/patient-acquisition-cost/ for dental, /marketing-problems/eye-care-groups/patient-acquisition-cost/ where cost per exam is the unit.
Group	At a handful of sites the blend starts lying, because a single acquisition can move it several points. Freeze the cohort now; it costs nothing before there is history to restate.
Platform	At platform scale, blended CAC is close to meaningless as a management number and is still the one most boards see. The useful artifacts are site-level CAC, cohort payback by channel, and an explicit statement of what is mix. Expect diligence to ask for exactly those three.

OTHER PRESENTATIONS — PE PLATFORMS & ROLLUPS

• Same-store flat while the unit count rises
• Demand arrives; booked, shown, treated work does not
• Same brand, same playbook, a distribution instead of a result
• The growth story has to survive diligence, not just the pitch