A cashflow forecasting app for UK households

Most tools give you one time scale and expect you to live with it. The cashflow here has three, and they answer genuinely different questions.

• Daily covers a month at a time, and you step from month to month. This is where you find the trough between the mortgage going out on the 1st and payday on the 28th.
• Monthly covers one forecast year at a time. This is the working view for the next twelve months, and it is where most decisions get made.
• Annual runs the whole way out to the ten-year horizon, one bar a year. Every forecast projects ten years ahead, so this is the long view of the same numbers rather than a second model. Ten years is a deliberate choice rather than a limit, and here is why it is not thirty.

There is no weekly view and no quarterly view. Weekly, fortnightly, four-weekly and quarterly exist as recurrences you can put on a bill or an income, so money lands on the right dates, but they are not separate ways of reading the forecast. You can see the same view in miniature as the cashflow view on the CrestCast overview.

Cash asks two questions and one pill switches between them. Balance is a stock: what is actually there at each period end. Movement is a flow: what came in and what went out over the period. Both carry all three period lenses, so you can ask either question day by day, month by month or year by year. The same two pills sit on profit and loss and on net worth, so the control does not change meaning when you change tab.

A second row underneath decides how far into that answer the chart goes. Summary plots the quantity itself: the running closing balance under Balance, the period's surplus or deficit under Movement. Breakdown groups it, with cash in stacked above the line and cash out below it, in bands that match the statement rows printed underneath. Detail goes one level further and names every account and pot on its own series, so you can tell which one is holding the money instead of reading a single combined line.

Not every pair has a chart behind it, and only the pairs that do are offered. There is no per-account view of cash movements, because a pound that moves between two people's accounts cannot be attributed to one of them without guessing, and a guessed figure is worse than none.

Breakdown carries the balance across it as a line on its own right-hand axis, because a six-figure balance plotted against a few thousand pounds of monthly flow would flatten every bar to a pixel otherwise. Where you hold savings as well as a current account, a second dashed line splits the two, so you can tell how much of the balance is money set aside and how much is the account the bills come from.

Tap one of the bands in the legend and the chart brings it forward with the rest faded behind it. It is a highlight rather than a filter, so nothing leaves the chart and the axis does not move: you can follow one band across the year and still see how big it is against everything beside it.

In the Options sheet, Inc liquid savings decides whether the pots are folded into the running balance at all. Liquid is the operative word: mark a pot as locked away, a fixed term or a notice you have not given, and it stays out of this line, because money you cannot reach this month is not part of what you have to spend this month. High-low adds the highest and lowest point reached inside each period to the balance chart, which is how you catch a month that ends fine having dipped hard in the middle. Real terms restates every figure in today's money, which matters most on the annual view: a balance ten years out is worth reading as what it would buy, not as what it would say on a statement.

A chart on its own is a shape. The statement is where the shape becomes accountable. Under every view sits a table with one column per period, and it does not move when you change the chart. It always reads the same way: the opening balance, everything that moved, and the closing balance those movements arrive at. Cash in, cash out and transfers are the movement rows, with a payroll deductions band beside them on the monthly and annual views, and those two also print the lowest and highest point reached inside the period alongside the balances. It is laid out so you can follow the arithmetic down the column rather than take the closing figure on trust.

Every parent row opens up. Tap cash in or cash out and you get the line items behind them: this salary, that mortgage repayment, the council tax, the car insurance. Tap deductions from pay and you get income tax, National Insurance, the pension contribution and any student loan repayment separately, which is money that never reached the current account and so is shown apart from the cash rather than inside it. Opening and closing balance open too, into the accounts that make them up. Where a household holds money in more than one name, the individual income and bill rows are tagged with whose they are, which is the bit that matters when not everything is joint. Once the columns outnumber a phone screen there is a landscape button that opens the whole grid full screen.

Cashflow is also the only one of the three statements carrying a subtotal column on all three views. Profit and loss has one on the annual view only, and net worth has none, because stocks do not sum.

Take a joint account with the mortgage leaving on the 1st, council tax on the 5th, the smaller direct debits through the first fortnight, and the contributions that fund it all landing on the 28th. On the monthly view that household looks fine: money in covers money out, closing balance steady. Nothing to see.

Switch to daily and a different picture shows up. In the month the £480 annual car insurance premium falls due on the 16th, everything lands before any money arrives, and the account is overdrawn from the 16th until payday on the 27th, which the monthly average never mentions.

A forecast built entirely from estimates drifts. So you record the real figure: in your inputs, under Actuals and one-offs, pick one item and one month and enter what genuinely went in or out. An income, a bill, a joint contribution, a transfer, a planned one-off, a salary, a member's income and a savings contribution can all be recorded. An explicit £0 is a real actual, meaning nothing moved that month, which is different from having none. The same list holds every actual you have recorded, with anything you are still waiting on kept in its own section underneath, because “this is what really moved” and “nothing has moved yet” are opposite statements and a single list would read as though both were settled.

An actual is not a note. It feeds the engine, so the balance for that month and everything downstream of it re-reckons from the real number rather than the estimate.

Two things to know before you go looking. First, actuals are recorded in your inputs, and only there. The cashflow, profit and loss and net worth statements are read-only, so a stray tap cannot change a number you meant to keep. Second, this is not budget versus actual. There is no budget in CrestCast and no variance figure anywhere. An actual replaces the estimate with what happened. It does not report the gap between them, and it will never tell you that you overspent by £73 on groceries.

The other half of the same question is money that has not turned up. Your consulting invoice was due on the 28th and the client has not paid it. Nothing is wrong with the forecast, but the cash is not there, and a balance that still shows it arriving on the 28th will send you into next month too confident.

So in the same list, beside the amount, you can say when you now expect it. That one date is all you enter, because the amount has not changed and typing it twice is how two figures for one payment start disagreeing. The cash moves to the day you named, and your profit and loss stays exactly where it was, because you earned that money in the month you earned it. Your net worth does not move either: the balance sheet carries what you are owed until the money lands. A bill you have not paid yet does the same thing in reverse, and shows as owed by you.

That is the accrual and cash distinction doing something useful rather than sitting in a glossary. The three statements are supposed to disagree in a month like this, and watching them disagree correctly is the fastest way to see they are three documents rather than one number wearing three labels.

Do you have to type everything in? Not all of it. The Bills screen and the Incomes screen both offer a CSV import. Export a statement CSV from your bank and feed it in. CrestCast reads it on your device and looks for the same payee appearing at least twice on a regular schedule and turns those into recurring forecast items: bills, incomes and joint bills.

It is deliberately narrow, and the narrowness is what keeps your data yours:

• If CrestCast does not recognise your bank's layout, you map the Date, Description and Amount columns (or money in and money out) yourself, once. It remembers the mapping for next time.
• Dates are read day-first, the way a UK statement writes them. If the file turns out to be month-first it stops and asks, rather than importing every transaction on the wrong day.
• No OFX and no QIF. It is a CSV import, not an account link.
• It does not store your transactions, and it does not record actuals. It creates forecast items and stops.

You export the file yourself and you choose what goes in, so nothing is ever pulled from your bank.

Plenty of people run a cashflow in a spreadsheet, and for one person with four bills it is fine. The trouble starts when it has to survive contact with real life.

You hand-roll a row per bill and encode the recurrence yourself, so quarterly water and four-weekly something-else each need their own formula. Then a pay date shifts by three days and every date-anchored row downstream needs re-anchoring. Then the mortgage fixed rate ends and the payment changes partway through, so you split the row in two or quietly leave it wrong. Then you want ten years out instead of two, and you rebuild the whole grid, because in a spreadsheet the columns are the horizon.

The real problem is not the effort. It is that a broken spreadsheet keeps working. It carries on producing a confident number long after the number stopped being true, and there is nothing in it that can tell you. A cashflow you cannot trust is worse than no cashflow, because you act on it.

CrestCast does not connect to your bank. No Open Banking, no account aggregation, no automatic transaction sync. Everything the forecast knows, you told it, and what you get for that is a forecast built from your plans rather than your past, with no credential shared and nothing read from your accounts.

It is also not a spending tracker. It forecasts rather than classifying past transactions, so there are no spending categories to police and no variance report. If you want the difference between the two disciplines set out properly, I wrote it up in cashflow versus profit and loss.

Cashflow is one of three statements built from the same set of facts. Profit and loss tells you whether the household is actually making money once interest and depreciation are accounted for. Net worth tells you what you are worth as it changes. Cashflow tells you whether the money is in the account on the day it is needed, which is the one that stops you at the till.

They agree with each other by construction, because they are three readings of one engine rather than three models. Change a bill and all three move. Schedule a pay rise for next April and all three move from April. A spreadsheet almost never manages that, because there the three tabs are three separate acts of bookkeeping.