The ROI of automating fax triage in eClinicalWorks comes down to a simple equation: monthly fax volume times minutes saved per document times your loaded staff cost, minus platform and implementation cost. With commonly cited savings of two to three minutes per fax, a practice handling a few thousand documents a month is usually recovering the equivalent of a full staff role. The second-order returns — faster referral turnaround, fewer missed authorization deadlines, fewer denials from documents that never got filed — are frequently larger than the labor savings and almost always harder to get credit for.
The Core Calculation
Start with the number your CFO will actually accept, then layer the softer benefits on top.
Monthly inbound fax documents × minutes saved per document × loaded hourly staff cost ÷ 60 = monthly labor savings.
Three inputs, and only one of them is a vendor claim.
Monthly volume you can pull directly from your fax service or eCW reporting. Use documents, not pages — a 40-page referral packet is one triage decision, not forty.
Minutes saved per document is the contested input. Published figures for AI-assisted fax handling commonly land in the two-to-three-minute range per document, covering the manual sequence of opening, reading, identifying the patient, searching the chart, attaching, renaming, and notifying. Be conservative and model two minutes.
Loaded hourly staff cost means salary plus benefits, payroll taxes, and overhead — typically 1.25 to 1.4 times base wage. If a front-office staffer earns $22/hour, model $28–31.
A worked example. A practice processing 3,000 fax documents monthly, saving 2 minutes each, at $29 loaded hourly:
3,000 × 2 = 6,000 minutes = 100 hours per month. At $29/hour, that's $2,900 monthly, or roughly $34,800 annually — about two-thirds of a full-time position.
Run your own numbers before reading any further. If the annual figure doesn't comfortably exceed platform plus implementation cost, the rest of this article is interesting rather than actionable.
Why Documents, Not Pages, Is the Right Unit
Vendors price on different units — per page, per document, per user, per month — and comparing quotes without normalizing produces bad decisions.
Per-page pricing looks cheap until a practice with heavy referral packets does the arithmetic. A specialty practice receiving 800 documents a month averaging 8 pages each is buying 6,400 pages. At $0.05 per page that's $320; at $0.15 it's $960. Same volume, triple the cost, and neither number is visible from the rate card alone.
Per-document pricing aligns better with the value being delivered, since the work saved is per triage decision rather than per page. It also protects you from the packet problem, where your most valuable automation — splitting and routing a dense multi-document fax — is also your most expensive line item.
Whatever the model, build a normalized annual cost using your actual last-90-days volume and page counts. Ask every vendor to price against that same volume. The spread between quotes usually narrows considerably once you do.
The Second-Order Returns Operators Forget
Labor savings are the easiest number to defend and rarely the largest one.
Referral turnaround and patient leakage. A referral that sits in an unsorted queue for two days is a patient who may be scheduled somewhere else by day three. For most specialty practices, one retained new patient per week covers a meaningful share of the platform cost — and referral speed is a direct function of how fast documents reach whoever schedules.
Missed authorization deadlines. Prior authorization determinations and payer requests for additional information arrive by fax with clocks attached. A determination that sits unread past a resubmission window converts into a denial or a rescheduled procedure. The AMA's 2024 prior authorization survey found practices completing an average of 39 authorization requests per physician per week and spending roughly 13 hours on the process, with the large majority of surveyed physicians reporting that prior authorization delays care.
Denials from documents that never got filed. Missing documentation is one of the most preventable denial categories, and it's frequently a filing failure rather than a clinical one — the record existed, it just wasn't attached where the biller could find it.
Turnover in roles nobody wants. Fax duty is repetitive, low-judgment work, and the people doing it are the people you replace most often. Every avoided backfill saves recruiting, onboarding, and the productivity trough that follows. MGMA's 2025 prior authorization analysis found the majority of medical groups have hired or reassigned staff specifically to keep up with administrative volume — capacity most practices would rather redirect than keep buying.
Be Honest About the Cost Side
A business case that only models benefits is a business case nobody senior will believe.
Platform cost is the visible line. Get it annualized against real volume, not a demo tier.
Implementation time is real and internal. Discovery, document-type mapping, queue design, and validation consume 20 to 40 hours of your team's time across the project. That's not billed, but it's not free either.
The persistent human-review queue. No system files everything automatically. Budget staff time for the exception queue — realistically 10% to 25% of documents in early months, declining as the classifier tunes. Model the higher end and let reality beat your forecast.
The ramp period. Savings don't start on day one. A shadow run produces zero savings by design, and the phased ramp produces partial savings. A twelve-month ROI model with full savings from month one is wrong by roughly a quarter of the annual number.
Data cleanup. If your eCW patient index has duplicates, you'll deal with them during implementation rather than after. Better to plan for it than discover it.
The 2024 CAQH Index puts a manual administrative transaction at roughly $3.41 versus about $0.05 for a fully electronic one, a gap of more than 98%. Fax triage doesn't make everything electronic, but that spread is the shape of the opportunity — and it only applies to the volume you're genuinely still handling by hand.
Baselines to Measure Before You Sign
The difference between a defensible ROI claim and a hopeful one is whether you measured anything beforehand. Capture these over two weeks, pre-implementation:
- Documents requiring a human routing decision. Not total volume — the subset a person had to read in order to decide where it went.
- Average handling time per document. Have two or three staff time themselves on twenty documents each. This replaces the vendor's number with yours.
- Referral-to-schedule turnaround. Median days from referral fax arrival to scheduled appointment.
- Documents aged over 48 hours. The size of your current backlog problem.
- Denials attributable to missing documentation. Pull the last two quarters from your billing system.
- Data fields retyped per document. Count them for one day.
Six numbers, two weeks of light effort. Without them, you'll be arguing about whether the tool worked instead of demonstrating it.
Honey Health's Fax Triage agent reads each inbound fax, classifies it, matches it to the eCW chart, extracts the fields staff would retype, and routes it to the queue that owns that document type — which maps directly onto four of the six baselines above. The ones that move first are routing decisions and retyped fields; turnaround and denial metrics follow a quarter later.
What a Reasonable Payback Period Looks Like
Most practices should expect to see labor savings turn positive somewhere in months four through eight, accounting for a ramp that produces little in months one and two.
If a vendor's model shows payback in under 90 days, ask what ramp assumption they used. If yours shows payback beyond 18 months, either your volume doesn't justify the purchase or you're modeling a platform when you needed a point solution.
The strongest cases share a profile: high document volume, a document mix diverse enough that routing genuinely matters, multiple staff queues, and meaningful data retyping. The weakest cases are practices with moderate volume and a uniform mix, where native eCW document handling is already doing most of the job.
One last piece of advice for the board conversation: present the labor number as the case and the second-order returns as upside. Referral retention and denial reduction are real, but they're harder to attribute cleanly, and a case that leans on them tends to invite the argument you don't want to have.
How the Math Changes for Multi-Site Groups and MSOs
Running the same calculation across a group rather than a single practice shifts two things, both in your favor.
Volume aggregates but implementation doesn't. Discovery, document-type mapping, and classifier tuning are largely done once and reused across sites running the same eCW instance. A group processing 15,000 documents a month across six locations pays roughly the same implementation cost as a single practice doing 3,000 — which makes the per-document economics substantially better at the group level.
The second shift is standardization. Most multi-site groups discover during document mapping that each location handles the same document types slightly differently, with queue structures that evolved independently. Automating fax triage forces that conversation and usually produces a standardized taxonomy the group wanted anyway. That's real value, though it's the kind that shows up as fewer arguments rather than a line on a spreadsheet.
The counterweight: a group rollout has more stakeholders, and every additional site with an opinion about routing adds calendar time. Budget for coordination, not just configuration.
Frequently Asked Questions
How do you calculate the ROI of fax triage in eClinicalWorks?
Multiply monthly fax documents by minutes saved per document by your loaded hourly staff cost, divide by 60, and subtract annualized platform and implementation cost. Model two minutes saved per document to stay conservative, and use loaded cost — salary plus benefits and overhead — rather than base wage.
How much time does AI fax triage actually save per document?
Commonly cited figures land in the two-to-three-minute range per document, covering opening, reading, patient identification, chart search, attachment, renaming, and notification. Savings vary by document complexity, so timing your own staff on twenty documents produces a far more reliable input than any vendor benchmark.
What fax volume justifies buying a dedicated tool?
Volume matters less than the share of documents still requiring a human routing decision. A practice with high volume but uniform, auto-filing documents may not need anything; a practice with moderate volume where most faxes require reading before routing usually does.
How long until fax triage pays for itself?
Labor savings typically turn positive in months four through eight, since a shadow run and phased ramp produce little savings early. Payback claims under 90 days generally assume no ramp period, which doesn't match how these rollouts actually run.
What costs do practices usually forget to model?
Four: internal implementation time (20–40 hours of staff effort), the persistent human-review queue that never reaches zero, the ramp period where savings are partial, and any patient-index cleanup that surfaces during rollout.
Should we count denial reduction in the ROI case?
Include it as upside, not as the core of the argument. Denials from missing documentation are a real and preventable category, but attribution is messy enough that a business case leaning on it is easier to challenge than one built on labor hours with a measured baseline.
