The 13-week cash forecast is the instrument that turns a therapy group's structural cash gap — sessions delivered now, cash collected weeks later, payroll every other Friday regardless — from a recurring surprise into a managed schedule. It is built from data the group already has, refreshed every Monday, and read for one line: the ending balance each week against a floor. How to build it.
The horizon and the grid
Thirteen columns, one per week, starting with the current week; rows grouped into opening cash, collections by source, disbursements by date, net cash, ending cash, and the floor; a fourteenth column is added every Monday and the oldest dropped, so the horizon rolls.
The collections engine, layer one: the receivables aging by payer
The open receivables aging — current, 31 to 60, 61 to 90, over 90 — by payer group as of the forecast date; each payer group's historical collection timing from the remittance history: the share of a week's claims that pays in week two, week three, week four, and so on through week twelve, with a residual that is never collected; the timing curve applied to each aging bucket to project the week in which each dollar of open receivables lands — commercial payer A with a median 24 days to cash puts most of its current bucket in weeks three and four; managed Medicaid at 55 days puts its current bucket in weeks seven through nine; claims already over 90 days are projected at the residual collection rate, which is often 30 percent or less; this layer is the most accurate part of the forecast because the claims exist and the payers' behavior is known.
Layer two: sessions scheduled but not yet billed. The next two to three weeks of scheduled sessions from the practice management system, by payer group, reduced by the group's no-show and late-cancellation rate, multiplied by net revenue per session for that payer, then lagged by the payer's timing curve from the week of service — so a session scheduled in week two with commercial payer A is projected to collect in weeks five and six; beyond three weeks, scheduled sessions are replaced by the group's weekly session run-rate by payer, because the schedule is not yet full.
Layer three: self-pay and time-of-service collections. Self-pay sessions and patient-responsibility amounts collected at the visit — copays, coinsurance, deductibles — projected from scheduled sessions and the group's time-of-service collection rate, landing in the week of service with no lag; this layer is why time-of-service collection discipline is a cash-forecast variable, not just a billing one.
Layer four: other inflows. Patient-balance statements and payment plans at their historical recovery rate; contract payments from school districts, employers, or regional centers on their invoicing and payment cycles — often 60 to 120 days — placed in the week they historically arrive; grant draws; and any line-of-credit draw already planned, shown separately so operating collections are not confused with borrowing.
Disbursements by date, not by average
Payroll on its exact pay dates with the gross-to-net and employer-tax amounts — including the quarters where employer tax resets or a bonus payment lands; rent on the first; malpractice, general liability, and health insurance premiums on their actual due dates, annual or semi-annual as they fall; estimated tax payments on their quarterly dates; software, licensing, and subscription renewals on their anniversaries; loan principal and interest; contractor and vendor payments on the cycle they are actually paid; credentialing and licensing fees; owner distributions only when the policy permits them; and a line for the unexpected at a modest weekly amount, because something always arrives — the forecast fails when any of these is smoothed into a monthly average, because the crunch lives in the week when payroll, rent, and a premium coincide.
Net cash and ending balance
Opening cash plus total collections less total disbursements gives net cash for the week and the ending balance that becomes next week's opening; the ending balance line is the forecast — everything else exists to produce it.
The cash floor
One full payroll cycle plus one month's rent is the minimum for most groups; add the next quarter's known lumpy items and any working-capital need from planned hires during their credentialing gap; the floor is drawn as a line across the thirteen weeks, and every week whose ending balance falls below it is flagged in the report — not as an alarm, but as a decision with a date.
The decisions a flagged week enables. Seen five weeks out: draw on the line of credit for the three weeks needed and repay from the recovery; move a discretionary purchase or a hire's start date; accelerate patient-balance collection on the aged self-pay receivables; delay an owner distribution; or, if the dips recur, size a facility against the receivables cycle rather than against a round number — seen the week of, the same dip is a missed vendor payment and a conversation with the bank the group did not want to have.
The Monday refresh
Replace last week's forecast column with actuals; compute the variance by line — collections by payer, time-of-service, disbursements; roll the horizon one week; update the aging and the schedule; re-run the timing curves; and read the variance, because it is diagnostic: a persistent shortfall from one payer means slower payment or rising denials and belongs in the payer file; a shortfall against scheduled sessions means the no-show assumption is wrong; a shortfall in time-of-service collection means the front desk is not collecting; disbursements over forecast mean something was missed on the calendar; the refresh takes forty-five minutes once the engine is built and is owned by one person.
Scenarios worth running
The largest payer slowing by two weeks; a clinician's departure removing their sessions for twenty-four weeks; a new hire's credentialing gap; a denial wave on a managed-Medicaid plan; a school contract paying 45 days late — each run as a toggle on the base case so the group knows its exposure before it happens.
What the forecast is not. Not the budget — the budget is a twelve-month plan built from capacity, and the forecast is a thirteen-week reality built from claims and the calendar; not the income statement — accrual revenue recognized by date of service will differ from collections every single week, and the forecast is only ever about cash; not a monthly exercise — a 13-week forecast refreshed monthly is a 9-week forecast with a stale first month.
Building it the first time
Two weeks: pull twelve months of remittance data and compute the timing curves and residuals by payer group; pull the current aging by payer; pull the next three weeks of scheduled sessions by payer and the no-show rate; measure the time-of-service collection rate; list every dated outflow for thirteen weeks from the payroll calendar, lease, insurance schedules, tax calendar, loan documents, and vendor terms; agree the floor with the owners; build the grid; run it; and reconcile the first two Mondays against actuals before anyone relies on it.
Why this is CFO work
The timing curves require joining remittance and practice-management data; the disbursement calendar requires knowing every contract and due date in the business; the floor is a policy decision made with the owners; the variance read requires understanding what each line means; and the flagged week is a decision the CFO brings to the owners with options — none of which happens from a bank balance.
Worked example
A 15-clinician outpatient group, US$3.2 million net revenue, payroll every other Friday (US$84,000 gross-to-net plus employer taxes), rent US$14,500 on the first, opening cash US$196,000, floor US$112,000 (one payroll plus rent plus a US$14,000 buffer). Weekly session run-rate 540, no-show 12 percent, mix: commercial A 36 percent (24 days), commercial B 20 percent (41 days), managed Medicaid 19 percent (55 days), self-pay 25 percent (0 days). The thirteen weeks:
| Week | Collections: aging | Collections: scheduled/run-rate | Time-of-service | Other | Payroll + taxes | Rent | Premiums / taxes / renewals | Other disb. | Net | Ending cash | Floor flag |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 41,000 | 0 | 18,500 | 0 | — | — | 2,100 | 9,000 | 48,400 | 244,400 | |
| 2 | 38,000 | 0 | 18,500 | 0 | 98,000 | — | 1,800 | 9,000 | −52,300 | 192,100 | |
| 3 | 36,000 | 6,500 | 18,500 | 0 | — | — | 2,400 | 9,000 | 49,600 | 241,700 | |
| 4 | 31,000 | 14,000 | 18,500 | 0 | 98,000 | — | 1,900 | 9,000 | −45,400 | 196,300 | |
| 5 | 24,000 | 21,000 | 18,500 | 11,000 (district) | — | 14,500 | 2,000 | 9,000 | 49,000 | 245,300 | |
| 6 | 17,000 | 26,500 | 18,500 | 0 | 98,000 | — | 22,600 (malpractice, annual) + 19,400 (est. tax) | 9,000 | −87,000 | 158,300 | |
| 7 | 12,000 | 29,000 | 18,500 | 0 | — | — | 1,700 | 9,000 | 48,800 | 207,100 | |
| 8 | 8,500 | 31,000 | 18,500 | 0 | 104,000 (quarter-start employer tax reset) | — | 2,100 | 9,000 | −57,100 | 150,000 | |
| 9 | 6,000 | 32,500 | 18,500 | 0 | — | 14,500 | 8,900 (software renewal) | 9,000 | 24,600 | 174,600 | |
| 10 | 4,500 | 33,500 | 18,500 | 0 | 98,000 | — | 1,800 | 9,000 | −52,300 | 122,300 | |
| 11 | 3,500 | 34,000 | 18,500 | 0 | — | — | 2,000 | 9,000 | 45,000 | 167,300 | |
| 12 | 2,500 | 34,500 | 18,500 | 0 | 98,000 | — | 11,200 (health premium true-up) | 9,000 | −62,700 | 104,600 | BELOW FLOOR |
| 13 | 1,500 | 35,000 | 18,500 | 0 | — | 14,500 | 1,900 | 9,000 | 29,600 | 134,200 |
Illustrative figures for a hypothetical organization; not a client's data.
Read: the group never runs out of cash, but week 12 breaches the floor by about US$7,400 because three things coincide — payroll, the health-premium true-up, and the tail end of the aging having mostly collected while the run-rate collections are still ramping. Seen on the first Monday, twelve weeks out, the options were ordinary: schedule a US$25,000 line draw in week 11 with repayment in week 13, move the software renewal in week 9 to a monthly plan, or push US$9,000 of aged self-pay balances with a two-week statement campaign in weeks 7 and 8. The group did the third and the first; week 12 closed at US$131,000. Without the forecast, the same week would have arrived as a payroll-Friday call to the bank.