Cash Flow Forecaster

Cash is not revenue. A signed contract is not cash. An invoice is not cash. Cash is money that has actually arrived in the bank account, and the gap between booking revenue and collecting it is where profitable businesses go broke. The skill models cash on the day it actually clears, with payment terms applied per income stream, so the picture you get is what the bank balance will look like — not what the P&L would suggest.

The model runs weekly for the first month and monthly for months two and three. Weekly resolution matters because monthly resolution hides the killer week. A month can net positive while containing a Friday where payroll, rent, and a quarterly tool subscription all clear before a single retainer arrives. The monthly average lies. The weekly minimum tells the truth, and the skill reports that minimum prominently.

Three scenarios always run. Baseline is what you expect. Conservative shifts revenue two weeks late, discounts pipeline to 50%, adds a 10% expense buffer, and puts one major client at 30 days late — because that's what bad months actually look like. Optimistic exists for motivation; conservative is what you plan around. The skill labels the difference clearly and recommends action based on the conservative scenario, not the cheerful one.

Income gets classified before modelling. Recurring (SaaS, retainers) uses predictable monthlies. Project-based work uses a weighted pipeline and a six-month rolling average — never the best month — because freelance feast-and-famine averages mislead. Seasonal models use same-period-last-year as the anchor. Lumpy enterprise contracts get modelled payment by payment, with the dangerous valleys between them shown explicitly. An income volatility score over 0.3 forces the conservative scenario to use the lowest historical month.

You get the period-by-period model across all three scenarios, the explicit lowest-cash-point week with a date, a runway number for each scenario, a what-if table that always includes the hire scenario, the client-loss scenario, and the late-payment scenario, and a recommendation list with quantified impact — "cutting £800/month from tooling extends conservative runway from 4.1 to 5.6 months." Every assumption is stated. Missing inputs get flagged rather than fabricated. The skill refuses to build a model without a real cash-in-bank figure — a forecast off a guessed opening balance is worse than no forecast.