Cash is forecast in weeks, because that is how it runs out

Why the month is the wrong unit

A monthly summary shows that more came in than went out, which is reassuring and can be entirely false as a picture of the month itself. Costs cluster at the beginning — rent, subscriptions, standing commitments — and client payments cluster towards the end.

A month that ends comfortably can contain a week in the middle where the account was empty and something important could not be paid. The summary averages that away, which is exactly the information you needed.

Weeks are short enough to show the collisions and long enough to be worth maintaining. Some businesses with tight positions work in days for a while, and that’s a reasonable temporary response to a bad stretch.

What actually goes in the forecast

Two lists and a running balance. Money expected to arrive, by the week you expect it, and money committed to leave, by the week it leaves. The balance carries forward, and the useful output is not the final figure but the lowest point along the way.

The distinction that matters is between what is committed and what is hoped for. A signed job with an agreed invoice date belongs in the forecast; a proposal that has not been accepted does not, however confident you feel. Optimism in the arrival column is the most common reason these forecasts mislead.

On the outgoing side, include everything, particularly the annual charges. Insurance, professional fees, subscriptions renewed once a year and any obligations to a tax authority. Those are the items that arrive as surprises despite being known months in advance.

Payment dates are not invoice dates

The single most common error is entering income in the week the invoice is issued rather than the week the money realistically arrives. Those are different by whatever your terms are, plus whatever the client actually does with them.

Use observed behaviour rather than agreed terms. If a particular client habitually pays two weeks after they should, forecast on the two weeks, because that is what happened last time and the time before. Recording when payments actually land makes this straightforward and it takes a moment per invoice.

Doing it this way also identifies your slow payers automatically, which is useful information for pricing and for deciding whose next job you want.

The point is the lowest point

A forecast isn’t there to predict the future accurately, which it will not do. It is there to show whether the balance goes below zero at any point in the next couple of months, and if so, when and by how much.

That question has practical answers when it is asked in advance. Chase two specific invoices. Ask a supplier for an extra fortnight. Bring forward an invoice that was going to be issued next month. Delay a purchase that has no deadline. All of those are available when the problem is six weeks away and none of them are available on the day.

This is the entire value of the exercise. It converts a crisis into a scheduling problem, and scheduling problems are solvable.

Keep it small enough to survive a busy week

The forecast that gets maintained is a simple one. A spreadsheet with a column per week and a handful of rows is enough for most small businesses, and it beats an elaborate system that gets abandoned in the second busy month.

Update it weekly, at the same time, alongside the invoicing. Fifteen minutes is usually sufficient once it exists, and the discipline of a fixed slot matters more than the sophistication of the model. Whether your bookkeeping software can produce this automatically is worth asking, though a view you built yourself is often better understood than one that appears at the press of a button.

What it will not tell you

A cash forecast says nothing about whether the business is profitable, whether the prices are right, or whether a particular client is worth having. It’s a timing instrument only.

It is also only as good as the assumptions in it, and the assumption most likely to be wrong is the date a client will pay. Building in some margin — assuming the slower version of each arrival — costs nothing and prevents the specific failure where a forecast said everything was fine right up until it was not.

And it won’t make an underlying shortage go away. A business whose forecast shows the balance dipping below zero every single month has a pricing or a cost problem rather than a timing one, and no amount of rescheduling fixes that. Knowing which of the two you have is itself worth the fifteen minutes, because the remedies are completely different and applying the wrong one wastes a year.