fin-cash-flow

Build a rolling 30/60/90-day cash forecast from AR/AP aging and fixed outflows, flag floor breaches, and model liquidity levers.

SKILL.md

GitHub

Cash Flow Forecast

When to use

Use this to project short-term liquidity, answer "will we make payroll / cover this quarter," or build a rolling 30/60/90-day cash forecast from receivables, payables, and known fixed outflows.

Not for: the historical indirect-method cash flow statement in a reporting package (fin-financial-statements), or actual-vs-plan variance on cash (fin-variance-analysis). This is forward-looking liquidity planning.

Method

  1. Set the starting position. Record today's cash balance, the forecast start date, and any minimum-cash covenant or comfort floor.
  2. Project inflows. From AR aging, schedule expected collections by expected pay date. Decision point: apply collection-probability haircuts by aging bucket (current ~95%, 30–60 ~80%, 90+ ~50%) — do not forecast overdue invoices at face value. Add other known inflows (financing, deposits).
  3. Project outflows. From AP aging and payment terms, schedule disbursements; add recurring fixed outflows (payroll, rent, debt service, taxes) on their due dates.
  4. Bucket into 30/60/90. Roll inflows and outflows into three periods; compute net cash flow and a running ending balance per bucket.
  5. Flag risk. Decision point: mark any period where ending balance breaches the floor or goes negative; record shortfall size and the exact date it hits.
  6. Model levers. Show the effect of accelerating collections, delaying payables, or drawing a credit line.
  7. Recommend actions. Prioritize the levers that close the gap with the least disruption.

Example

Start cash $220k, floor $150k. Days 0–30: inflows $310k (AR, haircut-adjusted), outflows $360k (incl. $180k payroll) → net −$50k, ending $170k (above floor). Days 31–60: inflows $280k, outflows $340k → net −$60k, ending $110k — breaches $150k floor on day ~48, shortfall $40k. Lever: pull a $90k invoice from a current-bucket customer forward 15 days (accelerate collection) → day-48 balance back to $170k. Recommend that plus deferring a $30k non-critical payable one cycle as backup.

Pitfalls

  • Forecasting overdue AR at face value. 90+ invoices rarely pay in full or on time — haircut them or you overstate liquidity.
  • Missing fixed, non-invoice outflows. Payroll, taxes, and debt service don't appear in AP aging; add them explicitly.
  • Netting away timing. A month that nets positive can still go negative mid-period if a big outflow lands before the inflow — track the running daily/weekly balance, not just the bucket total.
  • Ignoring the floor until it's breached. Model levers before the shortfall date, not after.

Output format

Starting cash: <$> | Start date: <...> | Minimum floor: <$>
Forecast:
  | period (30/60/90) | inflows | outflows | net | ending balance |
Inflow detail: | source/customer | amount | expected date | probability |
Outflow detail: | payee/category | amount | due date |
Risk flags: | period | shortfall $ | date floor breached |
Levers: | action | cash impact | timing |
Recommended actions (prioritized): <...>

Reference

The 13-week cash-flow model (layout)

The 13-week direct-method forecast is the treasury standard for short-term liquidity. Columns = 13 weeks (a rolling quarter); rows = receipt and disbursement categories. Every week computes net flow and a running ending balance carried into the next week.

Row Detail
Beginning cash = prior week ending cash (week 1 = today's actual balance)
Receipts
AR collections From AR aging, scheduled by expected pay date × collection probability
Other inflows Financing draws, tax refunds, asset sales, customer deposits
Total receipts Sum of inflows
Disbursements
AP / vendor payments From AP aging + terms, by due date
Payroll Every pay cycle (semi-monthly/bi-weekly) — largest recurring outflow
Rent / lease Monthly, on due date
Debt service Principal + interest per amortization schedule
Taxes Payroll, sales, income estimated payments on statutory dates
Capex / other Planned equipment, one-offs
Total disbursements Sum of outflows
Net cash flow Receipts − disbursements
Ending cash Beginning + net flow
Minimum floor Covenant or comfort floor (constant line)
Headroom Ending cash − floor (flag if negative)

Weeks 1–4 should use near-actual, known items (high confidence); weeks 5–13 rely more on modeled patterns. Roll the model forward each week: drop the completed week, add a new week 13, and reconcile last week's forecast vs. actual to calibrate.

Collection-probability haircuts (AR by aging bucket)

Never forecast overdue AR at face value. Typical assumptions (tune to your own collection history):

Bucket Collection probability Expected timing
Current (not yet due) ~95% On terms
1–30 days past due ~85% +1–2 weeks
31–60 days ~75% +3–4 weeks
61–90 days ~50% Uncertain
90+ days ~25% or write-off review Uncertain

What AP aging misses

Fixed, non-invoice outflows never appear in AP aging — add them explicitly: payroll, payroll taxes, rent, debt service, income/sales tax remittances, insurance, benefits. These are often the largest and most rigid disbursements; missing one is the top cause of a "surprise" shortfall.

Liquidity levers (in order of least disruption)

  1. Accelerate collections — early-pay discounts, pull large current-bucket invoices forward, factor receivables.
  2. Delay non-critical payables — stretch to the terms limit (not past — protect vendor relationships and discounts).
  3. Draw the revolver / line of credit — fast but adds interest and uses covenant headroom.
  4. Defer discretionary capex.
  5. Equity / term debt — slowest, for structural (not timing) gaps. Model the lever before the breach date; a lever applied after the floor is breached is a fire drill.

Timing trap and floor discipline

A period that nets positive can still go negative mid-period if a large outflow (payroll) lands before an inflow. Track the running weekly/daily balance, not just the bucket total. Flag the exact date and dollar size of any floor breach. Minimum-cash covenants are often the binding constraint before insolvency — treat the floor, not zero, as the trigger line. Always run a downside scenario (collections slip one bucket, a large customer delays) alongside the base case.