How it works
Three files in, a decision out
No integration and no implementation project. The whole thing takes about a minute.
- 01
Upload three files
A bank transaction history, an AR ageing and an AP ageing — CSV, TSV or XLSX, straight from your accounting system. Currency and date format are read from the files themselves, never assumed. An ambiguous date stops the run and asks, rather than guessing between the 3rd of April and the 4th of March.
- 02
The data is cleaned and checked
Duplicate rows are removed before any figure is computed. Credit notes are treated as credits, not overdue debts. One customer spelled three different ways is folded into a single account — otherwise their payment history splits three ways and their risk score is understated.
- 03
Payment behaviour is measured
For every settled invoice, the lag between the due date and the actual payment date. The median per customer, not the mean — one 200-day dispute would drag a mean far from typical behaviour. Customers with thin history fall back to the whole-book median, and are labelled as such.
- 04
The forecast is built
Direct method: expected receipts and disbursements, week by week, never derived from net income. Each open invoice is placed in the week its customer is actually likely to pay. Recurring commitments — payroll, rent, loans, VAT — are projected from the repeating pattern in your bank statement, and only when they appear at least three times.
- 05
You get four outputs
A one-page dashboard, an Excel workbook with every calculation visible, a ranked collection-risk list with recommended actions, and a written CFO brief naming the three things worth doing this week.
One honest limitation
The forecast covers invoices that already exist. Work you bill next month is not in your AR ageing, so it cannot appear in the forecast — which means the final weeks understate both receipts and payments. The brief says so plainly rather than presenting week 13 with the same authority as week 2.