Margin Architecture
Synthetic worked example — not client data
Worked Example · Multi-Site HealthcareThe board was about to close the wrong clinic.
An eight-clinic outpatient roll-up, $31.2M of revenue, two years into a buy-and-build. A board meeting in six weeks carries three decisions — close the smallest clinic, clone the flagship for the next two acquisitions, approve the integration budget. All three rest on one exhibit: EBITDA by clinic, with $4.6M of corporate cost allocated by revenue share.
Illustrative scope: eight clinics and a corporate centre, nine accounting entities, $31.2M of revenue. Illustrative fees comprise a $12k Financial Signal Diagnostic and a $30k Financial Signal Fix. Illustrative Diagnostic fees reflect the decision scope, entity and evidence-stream count, evidence readiness, and reconstruction effort. Actual fixed fees are scoped after discovery and evidence-access review. Modeled opportunity amounts are illustrative scenario assumptions informed by the diagnostic findings. They are not realized or guaranteed results.
EBITDA by clinic.
Corporate cost is not caused by revenue. It is caused by claims complexity, call volume, and support load. Switch the view to allocate it by what each clinic actually consumes — and to return $720k of physician compensation that was being paid through the corporate entity.
Consolidated EBITDA is $3,584k in both views. The diagnostic moves no profit whatsoever. It moves only the truth about where the profit comes from — which is the entire difference between closing Clinic G and keeping it.
Every board conclusion inverts.
Ordered by how far the reported figure sat from the truth. The closure candidate is profitable. The flagship is mid-pack once its billing load is priced. And the clinic that actually needs attention was not on the agenda.
Clinic A consumes roughly 40% of the billing team’s effort while paying 22% of the pool. Its payer mix is 34% commercial, with heavy government and complex claims. Clinic G — simple payer mix, self-sufficient front desk — consumes about 3% and was charged as though it consumed its revenue share.
What happened to the corporate pool.
Two clinics’ medical directors were paid through the corporate entity — understating those clinics’ direct costs and inflating the allocation base for everyone else at the same time.
$720k returned, $3,880k reallocated. Physician compensation went back to Clinics B and E where the work is done. What remained was distributed by support consumption — claims complexity, call volume, billing effort — rather than by revenue share.
The template was the wrong clinic too.
Revenue on the horizontal, true EBITDA on the vertical. Brookfield — quiet, mid-size, 62% commercial payer mix, low support consumption — is the profile worth acquiring more of. It was never the one being discussed.
What the work meant.
The diagnostic took four weeks and cost $12,000 — longer and dearer than a single-entity engagement because nine sets of books had to be conformed before anything could be compared. The Financial Signal Fix — a driver-based allocation model with written definitions, a restated 24-month clinic EBITDA series, a board exhibit that ties to consolidated, an acquisition-screening scorecard, and a monthly close cadence for the allocation — cost $30,000. Total illustrative investment: $42,000.
The corrective actions model to roughly $305,000 in the first year: $210k from a payer-mix and billing-workflow programme at Clinic A — denials, coding, collections — and $95k from reallocating billing and call-centre capacity now that its true consumers are visible. That is a 7.3× ratio of modeled opportunity to fees.
The larger consequence is not in that number. The board was six weeks from closing a clinic that makes money, cloning a flagship that is mid-pack, and leaving the genuinely weak clinic unexamined. Corrected clinic economics changed both the closure decision and the profile the roll-up should be acquiring against — which is a different kind of value from the one a first-year figure can carry.
What the board can act onDo not close Clinic G
At +3.0% true EBITDA it is profitable, not loss-making. The −6.0% on the board exhibit was an allocation artifact, not an operating result.
Put Clinic E on the watch list instead
At 3.0% true EBITDA it is the weakest site in the group, and it was not on the agenda at all.
Run a payer-mix and billing-workflow programme at Clinic A
The flagship is strong but not what the board was told. Its claims complexity consumes roughly 40% of billing effort. Modeled at $210k.
Re-aim add-on acquisitions at the Brookfield profile
High commercial mix, low support consumption. The correlation between payer mix and true EBITDA is the acquisition thesis.
Reallocate billing and call-centre capacity
Cost drivers are now visible per clinic, so support capacity can follow demand rather than revenue. Modeled at $95k.
Renegotiate the Clinic G lease
A related-party rent inherited in the acquisition sits roughly $120k above market. Treated as an adjustment outside the model — see below.
The allocation method was not a bookkeeping detail. It was about to destroy a profitable clinic and misdirect two acquisitions.
Modeled avoided value destruction.
Had the closure of Clinic G proceeded on the strength of the original exhibit, the modeled exposure comprises two parts:
- ~$190k — normalized annual EBITDA that would have been forgone by closing a clinic that is in fact profitable.
- ~$350k — illustrative one-time closure-cost exposure avoided. This is an illustrative scenario assumption outside the operating-data model; unlike every other figure on this page, it does not reconcile to the clinic financials.
Neither amount is added to the $305k headline or to the 7.3× ratio. Both depend on a decision that was never executed. They are shown separately, and labelled as a counterfactual, because the finding is the most consequential thing the diagnostic produced — and because presenting it inside the headline would be exactly the kind of unearned number this firm exists to find.
Clinic G pays roughly $120k above market in rent to a former-owner landlord, under a lease inherited in the acquisition. Normalized, the clinic improves from +3.0% to approximately 9%.
The interactive view and the table above show +3.0% — the figure that reconciles to the operating data. The rent normalization is deliberately held outside the model so every number shown ties cleanly, and it is excluded from headline economics.
Median healthcare services EV/EBITDA sits around 11.5×, with specialty practices trading at roughly 7–11× and platforms above $3M of EBITDA commanding materially higher multiples — premium valuations driven explicitly by multi-site density, stable payer mix, and strong financial reporting. Private-equity underwriting commonly assumes shared back-office synergies of 200–300 basis points, which is precisely the corporate cost pool this example shows being misallocated. Exit multiples are priced partly on reporting quality; an allocation method that nearly closed a profitable clinic is a valuation risk, not a bookkeeping detail.
Sources: FOCUS Investment Banking healthcare EBITDA multiples dashboards; ClearlyAcquired and Sofer Advisors practice-valuation guides. Accessed 2026.
Release controls
- Synthetic worked example — not client data. Calder Health Partners, its clinics, and its sponsor are invented.
- Invented company, real method. The analytical method and the arithmetic are real; the company is not.
- Modeled opportunity, not realized or guaranteed results.
- Impact estimates include illustrative scenario assumptions informed by the diagnostic findings.
- Fees shown reflect this illustrative scope only.
- The Financial Signal Diagnostic and the Financial Signal Fix are separate phases.
Start with the report or number leadership no longer fully trusts.
If a margin, profitability, pricing, cost, or revenue-unit signal is important enough to act on but not trusted enough to act on confidently, that is the conversation.
Allocation distortion appears wherever a shared cost is spread by a convenient base rather than by what causes it. Revenue share is the most common convenient base in existence. It is also almost never the cause of anything.