Outpatient Rehab KPIs: Formulas, Benchmarks, Sources

Most “top PT clinic KPIs” lists give you a name and a range: visits per case, units per visit, arrival rate, done. They rarely give you a numerator, a denominator, a period, or a source you could defend in a partner meeting. This post gives you all four, built from primary sources only: U.S. Physical Therapy’s SEC filings, the APTA Private Practice KPI Benchmarking Program, HFMA’s MAP Keys, two CMS manuals, and two peer-reviewed studies.

U.S. Physical Therapy (NYSE: USPH), a publicly traded operator of outpatient therapy clinics, reports these numbers in its SEC filings every quarter. Its clinics averaged 33.5 visits per clinic per day for the quarter ended June 30, 2026, up from 32.7 a year earlier, and physical therapy revenue per patient visit was $107.59 for that quarter, up from $105.33, according to its Form 10-Q filed August 7, 2026. Full-year 2025 average daily visits per clinic were 32.2, and net rate per patient visit was $105.76, per its Form 10-K filed February 27, 2026. These are the only public visits-per-day and revenue-per-visit figures we could find for outpatient rehab.

The other benchmark source practices actually cite, the APTA Private Practice KPI Benchmarking Program, publishes its definitions but not its results: results go only to practices that submit their own data. We could locate no published source for the visits-per-case, units-per-visit, or evaluation-conversion ranges (8 to 10 visits, 3.0 to 3.8 units, 85 to 92 percent) that circulate in vendor content. This post does not present those as benchmarks; it uses only figures traceable to a filing, a defined program, or a peer-reviewed study. Current as of September 2026.

What are the KPIs for a physical therapy clinic?

The KPIs below split into three families: scheduling (who showed up), production and coding (what got billed), and revenue cycle (what got collected, and how long it took). We note where a KPI matches a named industry definition (APTA Private Practice or HFMA’s MAP Keys), because the definition determines the number you get.

A few edge cases decide whether these numbers are comparable month to month. A reschedule made before close of the prior day is an advance cancellation plus a new booking, counted once, in the month the original slot fell. Double-booked or group slots should count visits, not slots, in a productivity numerator, while HFMA’s schedule-occupied metric (PA-1) counts slots occupied divided by slots available. An eval-only visit still counts as a visit and a new patient, which is why visits per new patient falls when eval-only volume rises. Home care and telehealth visits get excluded from USPH’s per-clinic daily count; do the same, or label the figure if you include them. Units per visit counts billed units regardless of what the multiple procedure payment reduction later pays; it measures what was billed, not paid.

Days from referral to first visit catches the leak before a patient reaches your schedule at all. In a small single-site study, Leatherwood et al. surveyed 100 patients six weeks after a PT referral at one orthopedic outpatient system (August to December 2022): only 60 attended PT, averaging 11.75 sessions (range 1 to 36). Not a benchmark, but a reminder that referral-to-arrival attrition belongs in the review too.

Here is every public, dated benchmark figure we could source, labeled by type: a public SEC filing, a member-only program, and a peer-reviewed study are three different kinds of evidence, not interchangeable.

How do you calculate arrival rate in physical therapy?

Arrival rate looks like a simple ratio until you check whose definition you’re using, because the definition changes the number for an identical schedule.

The APTA Private Practice glossary defines it as “((1-(Cancellations + No Shows)/(total appointments scheduled)) X 100,” and defines a cancellation narrowly: “Patient cancelled an appointment within the same day it was scheduled. Cancellations prior to close on the previous day are not counted as cancellations.” So an advance cancellation, made before close of business the prior day, does not reduce arrival rate at all: the slot had a chance to refill.

A stricter definition, arrived visits divided by all scheduled appointments regardless of status, counts every cancellation against you. For the same schedule the two produce two different numbers, and neither is wrong; they answer different questions. Comparing one clinic’s APTA figure to another’s strict figure, or to a study using a third definition, is a definitions mismatch, not a benchmark comparison.

A third definition shows up in the peer-reviewed literature. Bhavsar et al., studying 444,995 patients and 6,023,363 encounters at a large national outpatient PT provider from 2016 to 2017, defined a “no-show” as “a visit that was scheduled but had a final status of ‘canceled’ or ‘pending’ instead of ‘completed,’” folding cancellations of any kind into the count, the opposite of the APTA approach. No-show rates ranged from 15 to 31 percent across clinics, states, and providers, and 73 percent of the population missed at least one scheduled appointment over the study window. Real, peer-reviewed rates for one large provider’s 2016 to 2017 population under a broad definition, not a benchmark, and never to sit beside an APTA arrival rate as though they measured the same thing.

What is a good number of visits per day for a physical therapy clinic?

The honest answer is that the only public figure describes a multi-clinician clinic, not a single therapist, and it moves by quarter and by payer mix. USPH’s average daily visits per clinic (excluding home care) has moved as follows, per its Form 10-Q filed August 7, 2026:

USPH’s Form 10-K for fiscal year 2025 defines average daily visits per clinic as patient visits (excluding home care) divided by business days operated, further divided by the average number of clinics in operation. As of June 30, 2026, the company owned and/or managed 781 locations; this is a per-clinic figure across a multi-clinician clinic, not a per-therapist caseload.

Revenue per visit also depends on payer mix. For fiscal year 2025, Medicare and Medicaid patients were 40.1 percent of visits but only 35.8 percent of net patient revenue, per the 10-K. A clinic with a rising Medicare share will see revenue per visit drift down with no change in coding; one adding cash-pay or workers’ compensation volume will see it drift up. Read revenue per visit alongside payer mix, not on its own.

How many units per visit should a PT clinic bill?

There is no published, citable benchmark for units per visit; the 3.0 to 3.8 range that appears in vendor content has no source we could locate, so we’re not printing it as a target. Here is how to build the number instead.

Units per visit is billed CPT units (timed and untimed together) divided by billed visits, pulled from claims for a given month. It measures what was billed, not what was paid; MPPR and NCCI edits change reimbursement, not the unit count, so keep units per visit and net revenue as separate lines. The 8-minute rule sets the floor for how a unit gets justified, and the interaction matters most when a clinic pushes units per visit up by treating patients concurrently: timed units still require documented one-on-one minutes, so concurrent treatment raises billing exposure rather than revenue.

What are good AR days for a physical therapy practice?

HFMA’s MAP Keys, “developed for the industry, by the industry,” are the industry-standard revenue cycle definitions across 29 keys in five groups: Patient Access, Pre-Billing, Claims, Account Resolution, and Financial Management. Like the APTA program, the page publishes definitions without benchmark values, so there’s no single defensible “good AR days” number to quote. There is, however, a floor you can calculate.

FM-1, Net Days in Accounts Receivable, is net AR divided by average daily net patient service revenue (the “most recent three-month daily average”). AR-1, Aged A/R, buckets AR at 0-30, 31-60, 61-90, 91-120, and over 120 days from service or discharge date. AR-5, Remittance Denial Rate, is claims denied divided by claims remitted, pulled from 835 files; AR-6, Denial Write-Offs, is dollars written off as denials divided by average monthly net patient service revenue. PB-4, Total Charge Lag Days, is the sum of days between posting date and date of service across billed CPT codes, divided by the count of CPT codes billed.

The floor comes from Medicare’s payment rules. CMS’s Medicare Claims Processing Manual, Chapter 1, §80.2.1.2, sets a “payment floor” waiting period of “13 days for electronic claims and 26 days for paper claims” before the contractor may finalize payment, meaning Medicare pays clean electronic claims no earlier than the 14th day after receipt; AR days below roughly two weeks are structurally impossible for that payer. At the other end, the same manual, §70, sets the outer wall: under 42 CFR 424.44, claims generally must be filed “no later than 12 months, or 1 calendar year, after the date the services were furnished.” Everything you control on AR days sits between those two limits, which is also why net collection rate must lag: measuring collections received this month against charges billed this month understates performance, because a meaningful share of this month’s charges (Medicare’s electronic claims especially) legally cannot be paid until after the payment floor, and commercial payers often take longer. Measure net collection rate on a trailing 90 to 120 day window against charges from the matching prior period, not the current month.

How do KPIs interact?

None of these numbers should be read alone.

Units per visit versus one-on-one billing exposure. Pushing units per visit up by treating several patients at once raises billing exposure under the 8-minute rule, because timed units require documented one-on-one minutes.

Visits per case versus plan-of-care rules. CMS’s Medicare Benefit Policy Manual, Chapter 15, §220.3, requires a progress report “at least once every 10 treatment days,” and the same manual’s §220.1.3 requires recertification “not to exceed 90 calendar days from the initial therapy treatment.” For CY 2026, the KX modifier threshold is “$2,480 for PT and SLP services combined,” per CMS’s therapy services page; at roughly $100 per visit in allowed charges, a case approaches that threshold around visit 25. A visits-per-case average with no episode-level view hides all three triggers.

AR days versus denial prevention. Denied claims sit in AR until reworked, so rising AR days with a stable remittance denial rate (HFMA’s AR-5) signals a payment-timing problem; rising AR days with rising AR-5 or AR-6 (denial write-offs) signals a denials problem instead.

Arrival rate versus visits per clinician day. With a fixed schedule, every point of arrival rate is a point of productivity. The APTA definition treats an advance cancellation as a refillable slot; the strict definition treats it as lost time, which changes how urgently a falling number should worry you.

Revenue per visit versus payer mix. As shown above, payer mix moves revenue per visit independent of coding, so track it by payer class, not as one blended number.

KPIs for cash-pay and hybrid clinics

Cash-pay and membership models need revenue per visit split by payer class from the start: insurance visits against expected allowed amount, cash visits against the price charged (per visit or package visit), membership visits against monthly membership revenue divided by member visits. A single blended number across all three tells you almost nothing about where the revenue comes from.

Package completion rate (visits used divided by visits purchased, per cohort month) matters because unused package visits sit on the books as a liability, and prepaid-package rules vary by state; see our cash-pay package rules guide.

Membership retention is members active at month end who were also active the prior month end, divided by members active the prior month end (monthly, or as a 12-month cohort). Membership visit utilization is member visits divided by total members, per month; a low number means members are paying but not using the benefit, a churn risk even while revenue looks healthy.

Medicare beneficiaries paying cash for PT carry their own mandatory-claims rule; see our Medicare cash-pay guide rather than treat cash-pay PT as outside Medicare’s reach. Cash clinics carry near-zero AR days by design, but should still track days to payment on installment-billed packages, and refund rate.

A monthly KPI review routine

A fixed weekly cadence keeps the review from becoming a single overwhelming month-end pull.

Week 1, close the prior month’s schedule (every appointment needs a final status) and post all outstanding charges (watch HFMA’s PB-4 charge lag). Run arrival rate, cancellation rate, no-show rate, visits per clinician day, units per visit, and evaluation conversion.

Week 2, run the revenue cycle numbers: AR days, AR over 90, the remittance denial rate from 835 files, and net collection rate on the trailing, lagged window described above. Work the over-90 bucket by payer, since Medicare’s payment floor and a commercial payer’s slow adjudication look identical in one aggregate AR-days number but need different fixes.

Week 3, run an episode review: cases at visit 8 or beyond without a progress report on file, plans approaching day 90, and Medicare cases approaching the KX threshold.

Week 4, review payer mix and revenue per visit by payer class, comparing against the same month a year earlier rather than the prior month, since visit volume and payer mix carry seasonal patterns a month-over-month comparison will misread as a trend.

The standing rule underneath all of it: fix the definition in writing (which cancellation counts, which day starts the AR clock, whether home care is in or out) before comparing your number to anyone else’s.

A worked example month

The table below is an illustration: a single set of made-up, internally consistent numbers, not a benchmark and not sourced from USPH, APTA Private Practice, or the studies above. It exists only to show how the formulas above chain together across a month.

Two full-time PTs, 22 business days each, for 44 clinician days.

Notice the two arrival-rate rows: 93.0 percent under the APTA Private Practice definition, 87.0 percent under the strict definition, for the same 1,100 scheduled appointments. That six-point gap is the definition, not a different month.

Every table above pulls from a different system: scheduling for arrival rate and visits per day, documentation for evaluation conversion and discharge counts, claims for units per visit and revenue by payer. PrismEHR puts financial reporting and analytics against that same data on its physical therapy side, tracking provider productivity, clinical KPIs, and condition-level analytics from one system, so the routine above, and the denial-rate check inside it, doesn’t require stitching a scheduling export, a documentation export, and a claims export together by hand.