Paste your bank balance, expected receipts, and planned payments into ChatGPT. It builds a week-by-week forecast for a full quarter — with a worst-case scenario that strips out at-risk receipts, so you see exactly which week cash runs tight.
ROLE You are a treasury analyst building a 13-week cash flow forecast in Excel. OBJECTIVE Turn my current bank balance, expected receipts, and planned payments into a week-by-week cash flow forecast spanning one full quarter (13 weeks). DATA CURRENT BANK BALANCE (as of [date]): [Enter your cleared bank balance here, e.g. £85,000] MINIMUM CASH BUFFER: [Enter the lowest acceptable closing balance, e.g. £20,000] EXPECTED CASH INFLOWS (columns: Source, Amount, Expected Week, Confidence): [Paste your inflows data here] PLANNED CASH OUTFLOWS (columns: Category, Amount, Payment Week, Frequency): [Paste your outflows data here — mark Frequency as Weekly / Monthly / Quarterly / One-off] CONSTRAINTS - Week 1 starts on [enter date]. Each subsequent week starts 7 days later. - Opening balance for Week 1 = the bank balance above. Every other week's opening balance = prior week's closing balance. - Closing balance = Opening balance + Total inflows − Total outflows. - For recurring items: replicate the amount into every applicable week based on the stated frequency. - Do NOT invent receipts or payments. Use ONLY the data I have provided. - Flag any week where the closing balance falls below the minimum buffer. OUTPUT FORMAT 1. A 13-week forecast table: Week | Week Starting | Opening Bal | Total In | Total Out | Closing Bal 2. A summary: Total inflows, Total outflows, Net cash movement, Lowest closing balance (and which week), Number of weeks below minimum buffer Format as an Excel-ready table I can paste directly into a spreadsheet.
Replace all bracketed sections with your real data, then paste the prompt into ChatGPT.
My 13-week cash flow forecast includes some uncertain receipts — overdue invoices or unconfirmed contracts — currently treated as confirmed cash. Here is my current inflows list with a Confidence column: [Paste: Source, Amount, Expected Week, Confidence (Confirmed / Probable / At risk)] Please: 1. Keep the base-case forecast exactly as-is, using all inflows at face value. 2. Add a second "worst-case" version of the same 13-week table that EXCLUDES all inflows marked "At risk" entirely. 3. Add a third row showing each at-risk inflow weighted by an assumed probability I provide (e.g. a £20,000 invoice at 50% probability enters the forecast as £10,000). 4. Flag any week where the worst-case closing balance drops below my minimum cash buffer of £[amount], even if the base case does not. Do NOT invent probabilities — use only the percentages I specify per line.
Run this after the base build — it turns one optimistic forecast into two honest ones.
I have a 13-week cash flow forecast and need to extend it to 26 weeks (6 months) for a lender or board review. Here is my existing 13-week forecast: [Paste the full Week | Opening | In | Out | Closing table] New projected inflows and outflows for weeks 14–26: [Paste at whatever level of detail you have — weekly for weeks 14–17, broader monthly totals divided evenly across weeks 18–26 is acceptable] Rules: - Week 14's opening balance = Week 13's closing balance from my existing table. - Keep weeks 1–13 exactly as provided — do not recalculate them. - For weeks 18–26 where I've only given monthly totals, divide evenly across the weeks in that month and say so clearly in a note. - Flag any week in the extended range where the closing balance drops below my minimum buffer of £[amount]. Output the full 26-week table in the same column format as my original.
Treat weeks beyond 13 as directional — accuracy declines the further out you project.
| A | B | C | |
|---|---|---|---|
| 1 | Item | Amount | Timing |
| 2 | Customer invoice A | £36,500 | Week 3 — confirmed |
| 3 | Customer invoice B | £8,500 | Week 4 — at risk |
| 4 | Payroll | −£28,000 | Monthly — week unclear |
| 5 | Supplier batch | −£62,000 | Week 4 |
| 6 | Opening cash | £59,300 | Start of Wk 3 |
| 7 | Minimum buffer | £20,000 |
| A | B | C | D | |
|---|---|---|---|---|
| 1 | Forecast line | Wk 2 | Wk 3 | Wk 4 |
| 2 | Week starting | — | 15 Jun | 22 Jun |
| 3 | Opening Balance | — | £59,300 | £67,800 |
| 4 | Total In | — | £36,500 | £8,500 |
| 5 | Total Out | — | £28,000 | £62,000 |
| 6 | Closing Balance | — | £67,800 | £14,300 ⚠ |
| 7 | Minimum Buffer | £20,000 | £20,000 | £20,000 |
Wk 3 close = £59,300 + £36,500 − £28,000 = £67,800 Wk 4 open = prior week close = £67,800 Wk 4 close = £67,800 + £8,500 − £62,000 = £14,300 £14,300 < £20,000 minimum buffer → flag Week 4. ✓
Use today's cleared balance, not the ledger balance. The difference — uncleared cheques and deposits in transit — belongs as separate line items in week 1, not folded into the opening figure.
Vague timing kills forecast accuracy. If payroll runs on the 15th, put it in the week that contains the 15th. For uncertain receipts, use your best estimate and mark confidence as "At risk".
Recurring items — payroll, rent, subscriptions — repeat across multiple weeks. List them once with their frequency; the prompt replicates them across the right weeks automatically.
Before trusting the base case, look at the worst-case version with at-risk receipts stripped out. If that version still holds above your minimum buffer, you have real headroom — not just optimistic headroom.
Each Monday, replace the actuals for the week just ended, extend the forecast by one week, and adjust any projected figures that have changed. A stale 13-week forecast is barely better than no forecast.