What is the ROI of a revenue cycle project?
The return on investment (ROI) of a revenue cycle project is the ratio between the financial result the project generates and what it costs, over a set horizon. The formula is: ROI = (benefits for the period − costs for the period) ÷ costs for the period.
The project may be billing outsourcing, claim review and denial management software, or staff restructuring. To choose the model, see the guide Hospital billing outsourcing; this guide covers how much each option gives back.
A business case has three parts: a 12-month baseline, improvement assumptions by lever and costs. ROI is the summary; what convinces the board is the formula behind each line.
A note on terms. In Brazil, a denial (claim denial) is a full or partial refusal by the health plan (operadora) to pay a claim. Hospitals appeal through a recurso de glosa. The private sector is regulated by ANS (Agência Nacional de Saúde Suplementar, the National Agency for Supplementary Health), and Anahp (Associação Nacional de Hospitais Privados, the national association of private hospitals) publishes the main hospital benchmarks.
The five levers and the costs, as formulas
Each lever turns an operating indicator into Brazilian reais. All use the hospital's own data, by health plan when possible.
| Lever | Annual formula | Nature | Input |
|---|---|---|---|
| 1. Final denials prevented (prevention) | G × p × (1 − r) | Recurring | G = amount denied at first review over 12 months; p = expected reduction in initial denials; r = current overturn rate |
| 2. Denials overturned (appeals) | G × (1 − p) × Δr | Recurring | Δr = expected increase in the overturn rate |
| 3. Days to payment | ΔDTP × DR × i | Recurring (financing cost avoided) | ΔDTP = fewer days; DR = daily gross health plan revenue; i = annual cost of capital |
| 3a. Cash released by days to payment | ΔDTP × DR | One-time | Not in ROI; goes to cash flow |
| 4. Discharged not final billed (DNFB) | ΔDNFB × DR × i | Recurring | ΔDNFB = fewer days of discharged accounts not yet billed |
| 4a. Cash released by DNFB | ΔDNFB × DR | One-time | Same as 3a |
| 5. Staff productivity | FTEs converted × total annual cost per FTE | Recurring, if converted | Only hires avoided, redeployment or overtime eliminated |
| Implementation cost | Integration, training, parallel run | One-time | Proposals and internal budget |
| Recurring cost | Monthly fee, variable fee, licenses, contract management | Recurring | Variable fee: recalculate with the benefits |
In the table, DR = annual gross health plan revenue ÷ 365. FTE is a full-time equivalent.
Two definitions prevent many disputes. The overturn rate is the amount recovered on appeal ÷ the amount denied at first review. The current final loss is G × (1 − r), which should match accepted denials in the accounting records. The formulas for each indicator are in the guide Revenue cycle KPIs, and the difference between denial measures in Initial and final denials.
Why does separating prevention from appeals avoid double counting?
Separating the two levers with the formula above keeps the same real from being counted twice. A denial prevented at the source cannot be overturned later, because it no longer exists. That is why lever 2 applies the overturn gain only to the remaining denials, G × (1 − p).
Nor should the whole reduction in initial denials count as revenue: part of it would be recovered on appeal anyway. The revenue gain is only the share that would become a loss, (1 − r); the rest is a gain in time and work.
A map of where revenue leaks, before and after billing, is in the guide Hospital revenue leakage.
Faster payment is a financial gain, and it happens once
Cutting days to payment releases cash only once. If the hospital is paid 10 days sooner, receivables drop by 10 days of revenue, which enter cash once; with days to payment stable, there is no new release.
What repeats every year is the financing cost avoided: interest no longer paid or income now earned. Adding released cash to annual benefits inflates ROI.
The rate must reflect the hospital's situation:
| Hospital's situation | Reference rate | Value (retrieval date) |
|---|---|---|
| Finances working capital with banks | Average interest rate on total working capital loans, legal entities, non-earmarked credit (BCB, series 20725) | 22.89% per year in August 2026 [5] |
| Has cash invested | Effective monthly Selic rate, annualized (BCB, series 4189), or the actual return on investments | 13.78% per year in September 2026 [6] |
| Sells receivables | Effective cost of the factoring contract | From the hospital's own contract |
Both Central Bank of Brazil (BCB) rates were retrieved on 10/17/2026 and change monthly. Update them before presenting the business case. The cost of each day is detailed in Cost of payer delays, and financing options in Hospital working capital.
DNFB adds to days to payment only if the clock starts at the invoice
Discharged not final billed (DNFB) accounts are those of discharged patients that have not been billed yet. The Healthcare Financial Management Association (HFMA) measures it in days: gross DNFB amount ÷ average daily gross revenue [7].
Lever 4 adds to lever 3 only if days to payment start at the invoice, as in the definition used by Anahp [2]. If days to payment are measured from discharge or service, DNFB is already inside them, and adding both is double counting.
Accounts on hold also risk loss: TUSS table 38 (the ANS denial message code set) includes denial messages for billing after the contractual deadline, such as 3214 and 3161 [8]. If the hospital has a history of these denials, include the amount in lever 1. See Discharged not final billed (DNFB) and Hospital days to payment.
Productivity only becomes money when it is converted
Freed-up hours enter ROI only when they become cash: a hire avoided, a vacancy not filled, overtime eliminated or redeployment to revenue-generating work. Otherwise, the gain is capacity and appears as a qualitative benefit.
The weight is significant. At Anahp hospitals, staff costs were 39.03% of total expenses in 2024, for the hospital as a whole [1]. The cost per FTE should include salary, payroll charges, benefits and turnover. The full comparison is in In-house vs. outsourced billing.
What can the Anahp benchmark do, and not do, in the business case?
The Anahp benchmark puts the hospital's baseline in context; it does not set the target or the size of the gain. The figures cover members that reported data and use different denominators.
| Anahp indicator | Value | Denominator | Caution in the business case |
|---|---|---|---|
| Initial denials (management) | 15.93% in 2025 [2] | Amounts receivable from health plans [1] | Do not compare with denials over billed amounts; since 2026 the denominator is amounts received in the month [4] |
| Accepted denials (accounting) | 1.72% in 2025 [2] | Gross health plan revenue | Do not divide one denial rate by the other to estimate an overturn rate |
| Days to payment | 77.35 in 2025; 69.91 in 2024 [2][3] | Days from invoice to payment | Formula not published; the benchmark swings from year to year |
| Financial expenses | 3.77% of total expenses in 2025 [2] | Total expenses (includes interest) | Shows the weight of financing cost; it is not the hospital's cost of capital |
Treating the gap to Anahp as a “possible gain” is a mistake. The target comes from the hospital's baseline, by health plan and by denial cause.
Illustrative example: a hypothetical hospital
This example is entirely illustrative. The hospital, figures and assumptions were invented to show the calculation. They do not represent the industry, any client or an expected result, and they are not derived from Anahp figures.
The hypothetical hospital has gross health plan revenue of BRL 240 million a year, or BRL 657.5 thousand a day (DR). In the baseline, denials at first review totaled BRL 28.8 million over 12 months (G). The overturn rate was 80% (r), and the final loss was BRL 5.76 million. The project costs BRL 0.8 million to implement and BRL 1.2 million a year. In year one, the levers deliver half of the full effect (implementation ramp).
| Assumption | Base case | Conservative case |
|---|---|---|
| Reduction in initial denials (p) | 25% | 10% |
| Increase in overturn rate (Δr) | +5 p.p. | +2 p.p. |
| Reduction in days to payment | 10 days | 5 days |
| Reduction in DNFB | 3 days | 1 day |
| Cost of capital (i) | 22.89% (working capital, Aug 2026) | 13.78% (Selic, Sep 2026) |
| Productivity converted | 4 FTEs × BRL 90 thousand | None |
| Implementation cost | BRL 0.8 million | BRL 1.0 million (25% overrun) |
| Result (BRL millions) | Base | Conservative |
|---|---|---|
| Final denials prevented, per year | 1.44 | 0.58 |
| Denials overturned, per year | 1.08 | 0.52 |
| Financing cost avoided (days to payment), per year | 1.51 | 0.45 |
| Financing cost avoided (DNFB), per year | 0.45 | 0.09 |
| Productivity converted, per year | 0.36 | 0 |
| Full recurring benefit, per year | 4.84 | 1.64 |
| Benefits over 3 years (50% in year 1) | 12.09 | 4.10 |
| Costs over 3 years | 4.40 | 4.60 |
| Net result over 3 years | 7.69 | −0.50 |
| 3-year ROI | 175% | −11% |
| Payback | 8 months | Does not pay back within 36 months |
| One-time cash released (outside ROI) | 8.55 | 3.95 |
See the numbers in a table
| Item (3 years) | BRL millions |
|---|---|
| Final denials prevented | +3.60 |
| Denials overturned | +2.70 |
| Financing cost avoided (days to payment) | +3.76 |
| Financing cost avoided (DNFB) | +1.13 |
| Productivity converted | +0.90 |
| Total benefit | 12.09 |
| Implementation | −0.80 |
| Recurring costs | −3.60 |
| Net result (175% ROI) | 7.69 |
The same project goes from 175% to −11% with more cautious assumptions alone. In the base case, the largest lever is financial and depends on the rate chosen. And the BRL 8.55 million of released cash matters to treasury, but stays outside ROI.
The most common business case pitfalls
| Pitfall | Why it distorts | How to fix it |
|---|---|---|
| Adding released cash to annual benefits | Balance sheet effect that happens once | Show separately; in ROI, only financing cost avoided |
| Counting prevention and appeals on the same denial | Double counting | Apply Δr only to G × (1 − p) |
| Treating every initial denial prevented as revenue | Part of it would be overturned on appeal | Multiply by (1 − r) |
| Adding DNFB to days to payment measured from discharge | DNFB is already inside | Write down the days-to-payment formula |
| Counting freed-up hours as cash | Capacity is not money | Only converted FTEs |
| Treating recovery of the old denial backlog as recurring | It is a one-time gain | Separate line, in the year it occurs |
| Using the Anahp benchmark as a target | Different denominators and sample | Target from the hospital's own baseline |
| Ignoring taxes on revenue and variable fees | Denials prevented are gross revenue | Deduct before calculating the result |
| Crediting the project with external effects | Rate increases, payer mix and seasonality | Measure by health plan; log events |
| Instant ramp | The full effect depends on implementation and may not arrive in year one | Explicit ramp and conservative case |
What we see in operations
Accrua provides revenue cycle services and therefore takes part in business cases like the one in this guide, so what follows is the method we use, not a neutral assessment. The diagnostic starts from the history of claims the hospital has already sent to health plans and measures, claim by claim, the baseline for each lever: final denials, reversed denials, collection time and unbilled claims.
That measured baseline is what goes into the business case, even when it differs from the team's first impression. We measure collection time from the invoice and, separately, the time between discharge and submission, so the same gain is not counted twice; for the same reason, we keep denials prevented before submission apart from denials reversed on appeal.
Engagement is modular: the platform only, platform plus operation, or everything with Receita Garantida. The cost of each format goes into the business case as its own line. We do not estimate results before the diagnostic; the rollout ramp is designed with the hospital, based on what its history shows.
Where to start
- Measure the 12-month baseline by health plan: amount denied at first review, overturn rate, accepted denials, days to payment, DNFB and staff cost.
- Write down the formula for each indicator, including when the days-to-payment clock starts, before projecting any gain.
- Set the cost of capital with the finance team and note the date of the rate used.
- Build both scenarios, base and conservative, with a year-one ramp and all costs.
- Present separately the recurring result, the one-time cash released and the qualitative gains.