.health Confidential

Margin defense & downside — the aggressive number, stress-tested

The margin is earned, not assumed

REV's most aggressive figures — a ~78% gross margin and a ~65% run-rate EBITDA margin — are not switched on at launch. They are earned over a multi-year ramp as the self-learning system captures more of every encounter. This page shows the real ramp from the proforma, proves the returns stay excellent even if we miss by a lot, and explains why REV beats athenahealth on primary care alone.

Primary care only

Every number on this page is the independent primary-care (PCP) case — 1–5 clinician practices running routine established-patient E/M (the 99213/99214-dominated visit mix). Specialists and non-ambulatory settings look different — usually lower margin and more coding complexity. REV wins primary care first; specialty and enterprise expansion is upside, not assumed.

The margin is earned, not assumed — and even if we miss, the math still wins. The model reaches $1,071,000 in collections per provider per year only at scale; Year 1 runs below ~$750K/provider (~70% of the steady-state level). Gross margin builds toward ~78% as the AI codes more accurately and works more denials with every encounter. We have a couple of years to ramp — capture and margin compound; they are not assumed on day one.
Gross margin at scale
78%
PCP, steady-state target
Run-rate EBITDA
~$33.1M
$2.76M/mo × 12, ~65% margin
Revenue run-rate
$50.9M
at exit (2032)
Collections / provider
$1,071,000
/yr at scale — $89,250/mo
Gross MOIC
15.3×
6× downside floor · 10× base case → 25.4× · $4M @ 20%, no dilution

The ramp — how the margin is actually built

Straight from the proforma's month-by-month series. Gross margin climbs from ~29% to ~78% and EBITDA margin from break-even to ~65% over roughly five years — not a flat assumption. The shaded area is monthly revenue (right axis), which compounds as practices onboard and each provider ramps from Year 1 (~$750K collections) toward the $1,071,000 at-scale level.

Monthly revenue (right axis, $M/mo) Gross margin % EBITDA margin %

We have a couple of years to ramp. Capture and margin compound as the model learns from every encounter — coding accuracy, denial recovery and AR all improve with volume, so the steady-state ~78% gross margin is a destination the system grows into, not a switch flipped on day one. Early months show negative gross margin only because a fixed cost-to-serve base is absorbed against tiny initial revenue; it inverts quickly as providers onboard.

Collections / provider — Year 1

~$749,700 /yr · ~70% of scale

Collections / provider — at scale

$1,071,000 /yr · $89,250/mo

Gross margin — Year 2 → Year 5

38.5% → 77.3% → 78% at exit

Capture rate at scale

98.5% ~1.5% leakage target

Ramp basis: the model applies a Year-1 ramp factor of 0.70 to per-provider collections ($1,071,000 × 0.70 ≈ $749,700), with practices onboarding continuously thereafter. Margin figures are computed directly from the monthly proforma series (gross profit ÷ revenue; EBITDA ÷ revenue).

Why ~78% is defensible for primary care

The fair comparison is athenahealth — the scaled, public incumbent — but on the same metric. Athena ran a ~54% non-GAAP gross margin and, on the line that's actually apples-to-apples with ours, a ~29% adjusted (non-GAAP) EBITDA margin in 2018. Its headline GAAP operating margin was only mid-teens, but that's after stock comp, executive pay, heavy S&M and R&D, D&A and public-company overhead — so it understates the real economics. Even measured against athena's ~29% adjusted EBITDA, REV runs higher because it's built paperless, cashless and AI-native for one narrow, repeatable visit mix — it should beat athena on primary-care economics alone. (REV's ~65% is likewise a run-rate, at-scale EBITDA, before financing and taxes.)

Dimensionathenahealth (incumbent)REV (PCP-native)
Gross margin~54% non-GAAP (FY2018)~78% (PCP, at scale)
Adjusted EBITDA margin~29% non-GAAP (FY2018)~65% run-rate (at scale)
GAAP operating marginmid-teens (after SBC, exec comp, S&M/R&D, D&A)n/a (pre-GAAP)
ScopeBroad, multi-specialty platformIndependent primary care only
RCM modelLarge human RCM / billing workforceAutomated; clean claims + auto denials/appeals
Documentation / codingTemplating + post-visit reworkAmbient AI scribe; coded correctly at the visit
Accuracy over timeRules-based; staticLearns from every encounter (compounding)
RCM cost-to-serveLabor-heavy % of collectionstoward ~1.3% of collections

The four structural advantages

Paperless

An ambient AI scribe writes the note — no transcription, templating or scribe-labor cost. The same model that documents the visit also drives the code, so documentation and coding are one step, not two.

Cashless / automated RCM

Clean claims on first pass plus automated denial and appeal generation and continuous AR — pushing RCM cost-to-serve toward ~1.3% of collections, versus the human billing shops every incumbent carries.

Shift-left on encounter / coding

The visit is coded correctly at the point of care, to documentation-supported specificity — not under-coded for safety and reworked later. PCP's narrow E/M surface (99213/99214) makes this low-risk and highly automatable.

AI/ML that learns every encounter

Accuracy compounds with volume — coding, denial recovery and AR all improve as the system sees more encounters. No rules-based incumbent gets this flywheel; it is exactly why the margin ramps rather than starting maxed.

Downside sensitivity — even if we fail by a lot, we still win

Every cell is computed off the base model: MOIC = ARR × (% of plan) × (exit multiple) × 20% ÷ $4M seed, with base ARR = $50.9M and the seed owning 20% (held to exit — no Series A/B dilution). Rows haircut the exit multiple (10× → 4×); columns haircut ARR vs. plan (100% → 50%). The base case is highlighted; the amber cell is a deliberately pessimistic floor. Even 6× on half the plan still returns ~7.6× the seed.

Exit multiple ↓ / ARR vs plan → 100% of plan
$50.9M ARR
75% of plan
$38.2M ARR
50% of plan
$25.4M ARR
10× 25.4×$101.8M · 10× base case 19.1×$76.3M 12.7×$50.9M
20.4×$81.4M 15.3×$61.1M 10.2×$40.7M
15.3×$61.1M · 6× floor 11.5×$45.8M 7.6×$30.5M
10.2×$40.7M 7.6×$30.5M 5.1×$20.4M

Read it this way: the 6× downside floor is 15.3×, with the 10× base case at 25.4×. To fall below ~5× on the seed you would need to simultaneously halve the plan and have the exit compress to 4× ARR — and even that pessimistic corner returns 5.1× ($20.4M). Every other cell returns 7.6× to 25.4×. The seed is structurally protected by how far the company has to miss before the return stops being excellent.

Margin haircut — still strongly EBITDA-positive

Hold revenue at the $50.9M run-rate and instead crush the gross margin. Below-the-line corporate opex (core eng/DevOps + G&A + a lean ~15% external sales motion; ~85% of the market is captive) runs only ~$6.6M/yr, so the company stays strongly EBITDA-positive even at a 60% gross margin — nearly 20 points below plan.

Gross-margin scenarioGross profit (on $50.9M)− Corp opexImplied EBITDAEBITDA margin
78% — plan$39.7M$6.6M$33.1M65%
70% — haircut$35.6M$6.6M$29.0M57%
60% — severe haircut$30.5M$6.6M$23.9M47%
50% — collapse$25.4M$6.6M$18.8M37%

Even a 50% gross margin — a near-collapse of the entire thesis — still leaves ~$18.8M of run-rate EBITDA (37% margin). The downside is not "do we make money," it is only "how large is the win."

Illustrative computations. The downside cells are arithmetic sensitivities off the base proforma (ARR $50.9M, 20% seed ownership, $4M seed); they hold all else constant and are not separate forecasts. They show the shape of the returns under stress, not a prediction that any single haircut will occur.

Why a miss doesn't break us: the burn flexes with the market

The cost base isn't committed ahead of the revenue. Outside a small fixed core — core engineering, DevOps and product — we add downstream capacity only after the physicians and practices that pay for it are signed. And because ~85% of the market is captive (partners feed the practices) and the external sales motion is lean (~15%), there is no big sales org to carry ahead of revenue. Burn follows revenue, not the other way around, so the model adjusts to the market in near-real-time.

Slower than plan?

Hiring slows with it. We staff against signed practices, so a soft quarter stretches cash instead of draining it — we're not carrying a team built for revenue that hasn't arrived.

Faster than plan?

We scale into demand we've already won — capacity is added against booked revenue, not a forecast, so growth funds its own ramp.

Tiny fixed floor

Core engineering/DevOps/product plus a lean ~15% external sales motion is the only standing cost — ~85% of the market is captive. Everything downstream — onboarding, support, RCM ops — is variable and demand-triggered.

That's the deeper reason a miss doesn't sink us: it isn't only that the exit math still works at a fraction of plan — it's that the company flexes its spend to whatever the market gives it, so it doesn't burn ahead of the revenue it has actually earned.

On the captive channel: growth depends on Empower / partner networks adopting REV — the single biggest dependency to manage. The lean, demand-triggered cost base is exactly what lets us self-fund through a slower channel instead of burning ahead of it.

On cash: the base plan's thinnest point is ~$911K, around month 38 — right around EBITDA breakeven at month 39. That's a thin but positive cushion — but it stays positive on the $4M seed alone, with no Series A/B. The flex above is what protects it: in a slower scenario we hire and spend later, so the trough doesn't deepen the way a fixed-cost build would. The company that runs out of cash in a downside is the one that staffed ahead of revenue — by design, this one doesn't. (A genuinely faster cert/onboarding slip is the real risk to watch, not demand.)

And we keep what we earn — the tax structure

The run-rate EBITDA above is before tax — and the gap between EBITDA and cash kept is unusually small here. REV is a Texas company with years of planned losses before profit. Between NOL carryforwards, R&D incentives, and Texas having no income tax, the model carries near-zero cash tax for years, then a low-teens effective rate — well under the 21% federal statutory ceiling, and nowhere near the combined rate a California or New York competitor pays.

Texas state income tax
0%
no corporate income tax — franchise/margin tax only
Cash tax during build + NOL years
~$0
losses carry forward and shield first profits
Mature effective federal
~13–16%
21% statutory, less R&D credits
TX franchise tax
$0 → ~0.33%
$0 under $2.47M revenue, then EZ rate
PhaseFederalTexasEffective cash tax
Build + early growth (losses)$0 (losses)$0 (<$2.47M rev)~0%
First profitable years (NOLs apply)$0 (NOL shield)~0.33% margin~0–1%
Mature (NOLs exhausted)~13–16% effective~0.33–0.75% margin~14–16%

Three structural deferments

NOL carryforward

Every dollar of loss in the build and early-growth years offsets future taxable income dollar-for-dollar. Several planned loss years stack into a pool that shields the first profitable years entirely — cash tax stays ~$0 well past breakeven.

R&D incentives

Software development is qualified R&D. R&E expensing and the R&D credit let software companies deduct or credit a large share of spend — a primary reason profitable tech firms post effective rates well below 21%.

Texas domicile

No state corporate or personal income tax — only a franchise/margin tax ($0 under $2.47M revenue, then ~0.33% EZ). A structural edge worth several points of margin versus a high-tax-state competitor.

Mechanism in the model: early losses accumulate into an NOL pool, applied against later positive pre-tax income before any tax; the remainder is taxed at a ~15% blended effective federal rate (21% statutory net of R&D credits) plus the Texas margin tax (~0.33% of revenue, $0 below $2.47M). Net: minimal cash tax for years, modest thereafter.

Illustrative, not tax advice. Effective-rate and franchise-tax assumptions for the proforma; final rates to be confirmed with the company's CPA. Federal effective-rate context: GAO/ITEP show large profitable firms averaging ~16%, with many tech companies far lower after R&D and depreciation incentives.

Scope — this is the primary-care case

Primary care only

Every number here is the independent primary-care case.

The model is built for independent primary care — 1–5 clinician practices running routine established-patient E/M. That is what makes ~78% gross margin defensible: a narrow, repeatable, automatable visit mix (99213 + 99214 dominate) where ambient documentation and AI coding do nearly all the work.

Other specialties and non-ambulatory settings will look different — often lower margin, with more coding complexity (procedures, modifiers, prior auth, facility billing). We are not assuming those economics. REV wins primary care first; specialty and enterprise expansion is upside layered on top of a defended PCP base, not baked into these returns.

Sources & footnotes

  1. athenahealth margins. From athenahealth's last full year as a standalone public company (FY2018, prior to take-private): ~54% non-GAAP gross margin; ~29% adjusted (non-GAAP) EBITDA margin (Q2 28.9% / Q3 29.2%); GAAP operating margin only mid-teens — the latter is after stock-based comp, executive compensation, S&M, R&D, D&A and public-company overhead, so it understates underlying economics. The adjusted-EBITDA line is the apples-to-apples comparison to REV's run-rate EBITDA. Directional benchmark for a scaled, multi-specialty EMR/RCM platform with a large human RCM workforce.
  2. athenahealth transaction values. Taken private by Veritas Capital and Evergreen Coast Capital for ~$5.7B in 2019; subsequently acquired by Bain Capital and Hellman & Friedman for ~$17B in 2021 — evidence of the durable value the market assigns to EMR/RCM platforms even at incumbent margins.
  3. REV base figures. From the proforma export: revenue run-rate $50.9M, gross margin ~78%, run-rate EBITDA ~$33.1M ($2.76M/mo × 12), end-state 866 physicians across 346 practices, exit at a 10× multiple → EV $509M, investor proceeds $101.8M, gross MOIC 25.4× on a $4M seed @ 20%, held to exit with no Series A or B (~85% captive market + a lean ~15% external sales motion). Per-provider collections at scale $1,071,000/yr ($89,250/mo) at a 98.5% capture target; Year 1 ~70% of scale (~$750K). Pricing: $595/provider/mo + 4.9% of collections.
  4. Downside matrix. Illustrative arithmetic off the base model (MOIC = ARR × plan% × multiple × 20% ÷ $4M); margin-haircut row holds revenue at $50.9M and below-gross corporate opex at ~$6.6M/yr. Sensitivities, not separate forecasts.