We publish our math
The recovery numbers, with the assumptions attached.
Every vendor claims savings. We built a deterministic, seeded simulator of a 65-client multi-payer clinic, ran 200 Monte Carlo years over literature-anchored assumptions, attacked it with adversarial scenarios, and published the intervals instead of the ceiling.
| 65-client clinic, one year | Conservative (P5) | Median | Strong (P95) |
|---|---|---|---|
| Gross revenue recaptured / year | $365,097 | $437,373 | $530,218 |
| Contribution after labor and overtime | $177,237 | $220,435 | $276,447 |
| Authorized hours actually delivered | 94.8% | 95.8% | 96.4% |
| Hours recovered / year | 5,522 | 6,627 | 8,139 |
Modeled projections from a seeded simulation, not guarantees. Rates assumed: Medicaid $60/hr, commercial $75, private pay $85, blended by a 70/20/10 mix. Contribution nets out roughly $31/hr loaded RBT labor and overtime premiums.
Calendar basis: the published model schedules its simulated 65-client clinic Monday through Friday, and weekend operating days are not part of the model. Makeup and coverage completion are sampled as availability odds, not placed on specific calendar days, so a clinic’s real operating calendar, weekday-only or weekend-running, shifts what is recoverable. Your operating calendar changes this number. The walkthrough runs the model on yours.
Now make it your clinic.
Scale the published intervals to your caseload. Same model, your numbers.
Linear scaling from the published 65-client model (30 clients x 25 hrs = 750 client-hours vs the model’s 2,112). Modeled projections, not guarantees; payer mix and rates vary by state. Calendar basis: the published model runs a 65-client clinic scheduled Monday through Friday; weekend operating days are not part of the model.
What the model told us that marketing would have hidden
Only 17 percent of auto-extension attempts fit
Our own engine discovered that high-intensity kids sit at daily-hour caps and cannot absorb extended sessions. We report that instead of hiding it, because the waterfall’s other stages (makeups, the coverage marketplace, telehealth fallback) are what hold the number up.
57 percent of late cancellations arrive in clustered weeks
Sick weeks and vacations wipe out whole stretches, which is exactly when naive recovery math overpromises. Our model clusters cancellations the way real families do.
Winter belongs to staff coverage, not extensions
In a January-to-March run of a 70-client clinic, the substitute-coverage marketplace recovered four times what session extensions did. The system’s redundancy, not any single trick, is why the quarter still delivered 96.8 percent of authorized hours.
The alert loop alone is worth about $35,000 a year
The same simulated year with family notifications switched off delivers 510 fewer hours. That is why the caregiver app installs to a phone’s home screen and speaks Spanish.
How the model works, in one paragraph
Every client gets a real Monday-to-Friday schedule grid, weekend days never scheduled; every RBT gets a back-to-back route. Cancellations arrive as episodes (illness runs, whole-week vacations, one-offs, true no-shows, staff callouts, staff departures at the industry’s 40 to 65 percent annual turnover), seasonally weighted. The Recovery Waterfall then executes against physics: extensions must fit the therapist’s gap, the child’s daily cap, and the family’s consent; makeups expire; coverage needs an available substitute; telehealth needs BCBA capacity. Behavioral assumptions are triangular distributions with stated evidence tiers, sampled across 200 runs. Same seed, same year, every time, and when clinics run on the platform, their real telemetry replaces the assumptions.