How to forecast cash flow when you have 12 open invoices and 3 days of runway
A practical guide for owners. Why weighted aging beats straight-line forecasts. And when to ignore your bookkeeper's spreadsheet.
Most cash-flow forecasts for small businesses are wrong. They're built on straight-line assumptions ("we'll collect 1/30 of receivables each day") and ignore the most important variable: which invoices will actually pay this week, and which ones will sit for another 30 days.
Here's a better way.
The problem with straight-line forecasting
The textbook formula is:
Projected Cash = Current Cash + Expected Receivables - Expected Payables
Where Expected Receivables = Total A/R ÷ Average DSO
For a business with $184K in A/R and a 30-day DSO, that gives you $184K / 30 = $6.1K/day. Over 7 days, that's $42K of expected cash.
But that's not how it actually works. Here's the real distribution of when invoices pay:
- **30%** pay in the first 7 days after the due date
- **25%** pay in days 8-14
- **15%** pay in days 15-30
- **10%** pay in days 31-60
- **10%** pay in days 61-90
- **10%** are written off
If you have 12 open invoices totaling $184K, a straight-line forecast assumes they'll all pay evenly. They won't. A $42K invoice from a 14-day net customer will probably pay next week. A $24K invoice that's already 60 days past due from a 90-day-paying customer probably won't pay for another 30-60 days.
If you treat them the same, your forecast is wrong by 30-50%.
Weighted aging: a better model
The improvement is straightforward: weight each invoice by its probability of payment, based on its age and the customer's payment history.
For each invoice, the probability of paying in week N is:
P(pay in week N) = 1 - P(still unpaid at end of week N)** **P(still unpaid at week N) = 1 / (1 + e^-(k(N - N0)))
Where: - N = days past due (or until due, if not yet due) - N0 = the customer's average days-to-pay - k = a "decay rate" — typically 0.1 for slow payers, 0.3 for fast
For a customer that pays in 14 days on average, an invoice 30 days past due has a low probability of paying in the next week. For a customer that pays in 45 days, a 30-day-past-due invoice is right on schedule.
This is the same math that Gaviti, Growfin, and HighRadius use, but with one big difference: they use 2-3 year customer histories from thousands of invoices. You have 12 invoices. You don't have that data yet.
So for the first 90 days, you use industry defaults. After 90 days, you have real data. The forecast gets sharper.
The 4-week forecast format
Once you have weighted probabilities, you can produce a 4-week forecast that looks like this:
| Week | Expected Cash In | Confidence |
|---|---|---|
| 1 | $24K | Medium (75%) |
| 2 | $32K | Medium (65%) |
| 3 | $18K | Low (50%) |
| 4 | $9K | Low (35%) |
The "confidence" rating drops each week because prediction accuracy degrades with time. By week 4, you're essentially guessing.
**The right way to use this is:** - **Week 1 cash** = what you can confidently spend today - **Weeks 2-3 cash** = what you can plan around (hiring, equipment, etc.) - **Week 4 cash** = directional only, don't make bets
How to make payroll when runway is short
If your forecast says you can make payroll in week 2 but not week 1, here's the playbook:
1. **Stop all non-payroll spending today.** Cut ad spend, software, anything that's not payroll or revenue-generating. 2. **Aggressively chase the 3-5 highest-probability-to-pay invoices.** Not the biggest. The most likely. The 60-day-past-due $93K from a known-slow payer is not your best bet. The 5-day-past-due $8K from a known-fast payer is. 3. **Offer early-payment discounts.** 2/10 Net 30 ("2% off if you pay in 10 days") is standard. For an invoice you'd otherwise wait 45 days to collect, it's a great trade. 4. **Delay non-critical payables by 7-10 days.** Most vendors will tolerate this with a heads-up. 5. **Draw on a line of credit as a last resort.** Better than missing payroll, but expensive.
The order matters. Aggressive chasing comes first because the cost is just your time, and the upside is huge.
The thing your bookkeeper's spreadsheet doesn't do
Your bookkeeper's spreadsheet almost certainly uses straight-line aging. That's why their forecast is always wrong.
The fix isn't complicated — it's just not the default. The math above is what every modern AR tool computes. We built it into Collectly because we couldn't find it anywhere else at this price.
If you want to try it: https://getcollectly.app — 14-day free trial, no credit card, 10-minute setup.
— Davie