Quick answer: Multiply your daily inbound fax volume by the minutes staff currently spend opening, classifying, matching, and filing each document, subtract the share the software handles without human review, and value the remainder at loaded staff cost. For a mid-to-large pulmonary practice with heavy fax volume, fax triage software for pulmonology clinics typically reclaims several staff hours per day — but the number that matters is the one built from your own measurements, not a vendor's calculator.
Build the number from your own week of data
Every vendor has an ROI calculator, and every one is tuned to produce a favorable answer. Build your own. It takes a week and it's the only version a CFO or a partner group will accept.
You need four inputs:
- Daily inbound fax volume. Count for a full week and average. Mid-sized pulmonary groups typically land between 80 and 250 documents a day.
- Current handling time per document. Time twenty documents end to end — opening, identifying, patient lookup, filing, routing. Most practices land between 90 seconds and four minutes, depending on how much EHR searching each one requires.
- Post-automation blended handling time. This is the average across auto-filed documents (near zero staff time) and reviewed documents (30 to 60 seconds to confirm). Get this from a shadow-mode pilot, not from a vendor estimate.
- Loaded hourly cost of the staff doing the work — wages plus benefits plus payroll tax, typically 1.25 to 1.4 times base wage.
The formula:
Annual labor savings = daily volume × (current minutes − post-automation minutes) ÷ 60 × loaded hourly rate × 250 working days
A worked example to substitute into
Take a hypothetical ten-provider pulmonary and sleep practice. These numbers are illustrative — replace them with yours.
- Daily inbound volume: 180 documents
- Current handling: 2.2 minutes per document
- Post-automation blended handling: 0.6 minutes per document
- Loaded hourly rate: $28
Time saved per day: 180 × 1.6 minutes = 288 minutes, or 4.8 hours daily.
Annual hours: 4.8 × 250 = 1,200 hours, roughly 60% of a full-time position.
Annual labor value: 1,200 × $28 = $33,600.
That's the floor. Present it as the floor, not as the case — it's the number you're most confident in, and if the investment is defensible on labor alone, everything else is upside. In practice labor alone usually gets a practice to somewhere near break-even against subscription cost, which is why the other buckets matter.
What the published benchmarks are worth
Two numbers circulate in this category, and both are useful with caveats.
The first is the per-packet estimate: staff spend roughly 15 to 30 minutes per referral packet manually indexing pages, verifying the patient match, and uploading the PDF. That figure describes multi-page referral and results packets, not single-page correspondence, so applying it across your whole document mix will overstate savings badly. Use it for the packet share of your volume and nothing else.
The second comes from implementations reporting roughly an hour per day saved on routing incoming faxes to patient records. That's a real, conservative, single-workflow number — routing only, not the full open-classify-match-file cycle — which makes it a reasonable sanity check on the low end of your own estimate.
The 2025 CAQH Index offers the industry-level frame: roughly $21 billion in remaining savings sits inside manual and partially manual administrative transactions, and fully automated workflows save an estimated 70 minutes per patient visit. Useful for context in a board deck. Not a substitute for measuring your own queue.
The costs most ROI models leave out
A business case that shows only upside doesn't survive a finance committee. Put these in explicitly.
Implementation time. Four to eight weeks, with real hours from a practice administrator and whoever owns the EHR relationship. That's internal cost even when the vendor charges nothing for setup.
The parallel-run period. During shadow mode you're paying for the software while still doing the work manually. Budget a month of overlap.
Integration effort. Depending on your EHR, this ranges from a straightforward API connection to a custom interface build, sometimes with a one-time fee. Ask before signing.
Exception-queue staffing. Someone still works the exceptions. Model residual manual volume at 15 to 25% rather than assuming full automation — overstating the automation rate is the single most common way these models go wrong.
The accuracy ramp. Weeks one through four run at a higher review rate than steady state, by design. Don't model month-one savings as month-twelve savings.
The returns that don't fit on the labor line
Labor is the easiest bucket to measure and usually the smallest. Three others move the number more, and a pulmonary practice can defend all three.
Referral conversion. A referral worked the same day converts to a scheduled appointment more often than one that sits for three. If your practice receives 300 referrals a month and same-day processing lifts conversion by two percentage points, that's six additional new patients monthly. At any reasonable first-year value per new pulmonary patient, this bucket alone can exceed the labor savings. Measure your current time-from-referral-receipt to scheduled appointment first, so you have a before number.
DME and PAP deadline capture. Medicare's PAP coverage standard requires documented use of at least four hours per night on at least 70% of nights in a consecutive 30-day period within the first 90 days (CMS LCD L33718). A compliance download that sits unread past the window turns into a denied claim and a patient without device coverage. Count how many of those happened last year — most practices have never counted, and the number is usually higher than expected.
Turnover avoidance. Document sorting is among the least engaging work in a practice, and MGMA has tracked front-office turnover as a stabilized but unresolved problem heading into 2026. Replacing a frontline support staff member runs into the tens of thousands once recruiting, onboarding, and the productivity gap are counted. Avoiding one departure a year is a real line item, even if it's the hardest of the four to defend on a spreadsheet.
A worksheet you can fill in
Structure the model in five rows and it fits on one page for a partner meeting.
- Baseline labor. Daily volume × current minutes ÷ 60 × loaded rate × 250. Your starting cost.
- Post-automation labor. Same formula with the blended post-automation minutes. Get this from shadow mode.
- Gross labor savings. Row 1 minus row 2.
- Second-order value. Referral conversion lift × value per new patient × 12, plus avoided DME denials, plus a conservative turnover-avoidance figure. Keep every assumption visible and sourced to your own data.
- Costs. Annual subscription + amortized implementation + parallel-run month + exception-queue staffing.
Net first-year benefit is rows 3 plus 4 minus row 5. Divide one-time costs by the monthly net to get payback period, which is how operators actually think about it. Most mid-to-large pulmonary practices land between 6 and 12 months when referral conversion is included, and 12 to 24 months on labor alone.
Honey Health's Fax Triage agent files directly into the practice's existing EHR rather than a separate portal, which matters to the model specifically: a tool that routes documents into its own inbox leaves the lookup-and-file labor in place — the exact cost row 2 is supposed to reduce.
Recovered hours only become savings if you spend them
The part practices underestimate isn't the technology. It's what happens on the other side of it.
When five hours a day come off the fax queue, that time doesn't convert to value on its own. It diffuses unless somebody directs it. The practices that get the most out of this decide in advance where the hours go: prior auth follow-up, converting inbound referrals to appointments faster, chasing DME compliance downloads that never arrived, patient outreach that had been quietly deprioritized for years.
Be honest with your finance committee about which version you're proposing. "We will eliminate a position" and "we will redeploy this person onto referral conversion" are different business cases with different numbers and different internal politics. Picking one before the rollout starts is what separates a project that delivers from one that produces a nicer inbox and no measurable change.
Frequently Asked Questions
What automation rate should we model?
Model 75 to 85% of documents filing without human touch once the rollout is mature, not 100%. The residual is handwriting, multi-patient batch faxes, and poor scans. If a vendor quotes above 95% across your whole mix, ask which categories that covers — it's usually true of the structured documents and not of the mix as a whole.
How do we value a recovered staff hour?
Use the loaded hourly cost of the person actually doing the work, not a practice-wide average. If the fax queue is worked by front-office staff at $18 to $22 base, the loaded figure is roughly $23 to $30. Valuing recovered hours at a clinical or management rate inflates the case and gets it challenged.
Is the ROI different for a multi-site pulmonary group?
It usually improves. Subscription pricing tiers tend to flatten per-document as volume rises, and multi-site groups often carry duplicated document-handling labor at each location that consolidates once triage is centralized. Implementation effort rises, but not proportionally.
Should we include referral conversion in the case?
Yes, but conservatively and against your own baseline. Measure current time-from-referral-receipt to scheduled appointment before you start, then track it after. A one to three percentage point conversion improvement is defensible; anything larger needs your own before-and-after data.
What if we can't measure our current handling time?
Measure it before buying anything. One week of tallying volume by category and timing twenty documents gives you both the business case and the baseline you'll need to prove the result. Practices that skip this step almost always end up unable to demonstrate the return they actually got.
