Sometimes you shouldn't. A business with genuine reserves, one entity, one currency and predictable payment timing can manage its cash competently without a forecasting tool. But a reserve only measures how much cash you hold. It cannot tell you whether cash will be in the account on the dates your obligations fall due, and timing is what catches reserve-rich businesses out.

When a reserve alone is genuinely enough

The honest boundary looks like this. A reserve-only approach is defensible when nearly all of the following are true: you operate one entity, or cash moves freely between entities; you trade in one main currency; customers pay at the point of sale, in advance, or on stable recurring schedules; no single customer can materially move a month's receipts; payroll, tax, rent and debt payments are routine and sit well within the buffer; seasonality is low; and growth is modest enough that working capital needs are not expanding quickly.

If that describes your business, a well-managed reserve plus the disciplines covered later on this page is a reasonable strategy, and you should not let anyone scare you into buying software. It is worth being direct about the state of the evidence here, because almost nobody else is: no study has ever classified established businesses into "reserve is enough" and "forecast required" groups and tested the outcomes. The boundary is a decision framework built from well-evidenced mechanisms, not a proven formula. The same honesty cuts the other way, though. The most-repeated reserve rule in circulation, that a business should hold three to six months of operating expenses, has no primary evidence base either. It is advisory convention, repeated from bank to adviser to blog, and it was never a research finding.

What the research does establish is the mechanism underneath. Corporate finance work on the precautionary motive, from Opler and colleagues in 1999 through Bates, Kahle and Stulz in 2009 and Han and Qiu in 2007, consistently finds that the volatility of a firm's cash flows, not its average balance, is what drives how much cash it rationally needs to hold. Where inflows are predictable and outflows are smooth, a buffer alone can bridge shocks. As volatility and lumpiness rise, the buffer stops being a substitute for visibility.

Why a healthy balance can still miss payroll

A reserve is a stock: a quantity of cash at a point in time. Payroll, supplier runs, VAT and customer receipts are flows: movements that land on dates. The gap between the two is where reserve-rich businesses get caught, because your outflows and your inflows behave in opposite ways.

Your largest outflows are fixed and dated. In the UK, VAT for a quarterly filer is due one calendar month and seven days after the period ends, and PAYE and National Insurance are due by the 22nd of the following tax month for electronic payers. Payroll lands on the date in your employment contracts. None of these moves because a customer is slow.

Your inflows are variable and habitually late. Xero's Small Business Insights, the company's measurement of its own customer base of around 440,000 UK small businesses, put the average wait to be paid at 29.0 days in the March 2026 quarter, with invoices settled 8.2 days late on average. That lateness sits on top of agreed terms, so an invoice issued on 30-day terms realistically pays out at around five and a half weeks. The Federal Reserve Banks' 2023 Small Business Credit Survey found the same friction in the US: 39% of employer firms reported slow-paying customers as a challenge, and firms invoicing after delivery felt it most.

Put fixed-date outflows against variable, late-arriving inflows and the conclusion follows: a balance that is adequate measured monthly can still be short on a specific due date. That is the failure a stock number cannot see and a dated view of flows can.

What UK payment data shows about timing risk

The timing problem is well documented, and it is improving without going away. The government's Payment and Cash Flow Review reported that smaller businesses' overall payment times fell from 81 days in 2010 to 36 days in 2020/21, yet SMEs were still owed around £22,000 each on average in late payments in 2022. The Department for Business and Trade's 2025 payment statistics show large businesses now paying suppliers in a median of 32 days, with 15% of invoices still paid late.

The law is moving too. In March 2026 the government announced its largest late-payment reforms in over 25 years, and the Bill has been before Parliament since May. The package includes a 60-day cap on payment terms where large firms pay smaller suppliers, mandatory statutory interest at 8% above the Bank of England base rate, and new powers for the Small Business Commissioner to investigate and fine persistent late payers. None of it is law yet, and even once it is, a cap on terms does not put cash in your account on payroll day. Timing risk shrinks; it does not disappear.

One more number is worth knowing for what it says about how thin buffers typically run, provided it is read carefully. The JPMorgan Chase Institute's transaction research found the median US small business held 27 cash buffer days in 2015, and its 2026 update puts the current median at 17.6 days. Those samples skew far smaller than a mid-sized finance-led business (median annual revenue in the 2025 sample was around $125,000), so neither figure is a target for you. What they show is dispersion: even among businesses that feel comfortable, measured cover is often shorter than assumed.

A worked example: the month the timing bites

Take a UK business with around 25 staff and roughly £3 million in turnover, running a single entity on monthly payroll. It starts November with £150,000 in the bank, about three weeks of typical outflows, which feels comfortable.

The figures are illustrative; the timing inputs are not. The VAT and PAYE dates are HMRC's, and the eight-day slippage is the current UK average in Xero's data, applied to invoices on ordinary 30-day terms. Both columns end the period in the same place. Only one of them makes payroll. Nothing went wrong in the late column that would show up in a monthly management pack: no lost customer, no bad debt, no fall in sales. The month nets out. The 22nd and the 28th do not.

That is the whole argument for forward visibility in one table. A reserve answers "how much". It cannot answer "will the cash be there on the 28th", because that is a question about dates.

The signals that a reserve has stopped being enough

The boundary from the first section erodes in recognisable ways, and each one is evidenced rather than hypothetical.

Customer concentration. When one customer's payment can move your whole month, receipt timing stops being statistical and starts being a single point of failure. A useful caution line is 10% of revenue from a single customer: it is the threshold at which US accounting rules require public companies to disclose reliance on a major customer, and credit-risk practice treats the same line as the point to start managing concentration actively, with 20 to 25% widely treated as serious. These are conventions rather than laws of nature, but they exist because concentrated receivables and a fixed payroll date are a dangerous pairing.

A second entity, currency or bank. Group cash is not the same as available cash. Money held in one entity is not automatically there to meet another entity's payroll on Friday, and reconciling positions across accounts introduces exactly the timing blind spot a single balance cannot represent.

Seasonality and growth. A buffer sized for an average month understates the peak-date requirement in a seasonal one. Growth is quieter but just as corrosive: it consumes cash ahead of collection, so a reserve that is fixed in pounds silently shrinks in weeks of cover as payroll and payables expand.

Exception-handling becoming routine. If your finance team is repeatedly moving money between accounts to make dates, chasing specific receipts to cover specific payments, or discovering obligations late, the business has already crossed the boundary. The workaround is the forecast; it is just being run by hand, under pressure, every month.

How to manage a reserve well without forecasting software

For a business genuinely on the simple side of the boundary, good practice needs a bank statement, a calendar and a spreadsheet, and it is worth doing properly.

Set a minimum balance floor and name it. Tie it to your largest predictable near-term obligations, typically one payroll run plus the next tax payment, and treat breaching the floor as a trigger to act rather than a number to feel uneasy about. Keep a dated obligations calendar covering payroll, PAYE, VAT, rent, debt service and major supplier runs, and review the reconciled bank position against it on a fixed cadence, weekly as a defensible default and more often around payroll and tax dates. Hold VAT and PAYE money apart from working cash, because cash you have collected for HMRC only looks like yours. Keep an aged receivables list and watch concentration, since a £200,000 receivables ledger is not £200,000 of cash, least of all when half of it sits with one customer. Recalculate the floor after any structural change: a large customer won or lost, a new entity, new debt, a step up in headcount. And arrange a credit line before you need it, sized to your largest plausible timing gap; an unused facility is cheaper insurance than an oversized idle buffer, which quietly pays an inflation cost for sitting still.

None of this requires software, and for a genuinely simple business it may be all that is ever needed. The disciplines also degrade gracefully: the day one of the escalation signals appears, you already hold the obligations calendar and the floor that a proper forecast formalises.

How Float fits

Float is built for the businesses on the other side of the boundary: finance teams on Xero or QuickBooks Online whose timing risk has outgrown a balance check.

Float imports your reconciled bank transactions, invoices and bills from your accounting platform every 24 hours, or on demand, so the forecast reflects your accounts as at the last sync without anyone rebuilding a spreadsheet. The 13-week rolling view is built for exactly the dated questions in the worked example, mapping every expected receipt and payment to the date it actually lands, while the monthly view extends up to 36 months for board planning. You can set a cash threshold, the software equivalent of the minimum balance floor, and see the date any scenario would breach it. Scenario planning lets you model a slower-paying customer, a new hire or a delayed project alongside your base forecast rather than inside it. And for groups, multi-entity consolidation shows the combined position with the ability to drill into any single entity, which is precisely the visibility a second entity takes away from a bank balance.

If the escalation signals on this page read like a description of your last quarter, you can start a free 14-day trial and see your own dates in a forecast the same afternoon. Float connects to Xero or QuickBooks Online in about three minutes.

Frequently asked questions

How much cash should a business keep in reserve?

There is no evidenced universal answer, and you should be sceptical of anyone offering one. The widely repeated three-to-six-months rule is advisory convention with no primary research behind it, and the best transaction data, from the JPMorgan Chase Institute, measures much smaller businesses than a mid-sized finance-led firm. The defensible approach is to size a floor from your own figures: your fixed monthly outflows, your largest near-term obligations and your own worst historical month.

Is a cash reserve the same as a cash flow forecast?

No, and the difference is the point. A reserve is a stock: how much cash you hold today. A forecast is a dated view of flows: what is due in and out, and when it lands. A reserve protects against shocks; a forecast protects against timing surprises. Most established businesses eventually need both.

Can scenario planning help us avoid unexpected cash shortfalls?

Yes, because most shortfalls are timing events rather than genuine surprises. Modelling a slower collections month, a delayed project or a new hire against your base forecast shows you the date a problem would arrive and how deep it would run, while there is still time to act. In Float, scenarios sit alongside the base forecast, so you can compare outcomes without disturbing the version you trust.

Do profitable businesses still run out of cash?

Yes, regularly, because profit and cash are different measurements on different timelines. A business can be profitable on paper while the cash from those profits is still sitting in unpaid invoices when payroll falls due. We cover that failure mode in detail in our guide to why profitable businesses run out of cash, which pairs with this page: profit that has not yet become cash, and a cushion that hides timing risk, are the two ways a healthy-looking business gets caught.

What is changing in UK late payment law?

In March 2026 the government announced its largest late-payment reforms in over 25 years, and a Bill has been before Parliament since May 2026. It would cap payment terms at 60 days where large firms pay smaller suppliers, make statutory interest at 8% above the Bank of England base rate mandatory, and give the Small Business Commissioner powers to investigate and fine persistent late payers. The measures are not yet law, and a cap on terms still would not guarantee cash arrives before your own fixed dates.

Does Float connect to our bank account?

No. Float retrieves bank transactions, invoices and bills through your accounting platform rather than connecting to your bank, so it works with any bank that Xero or QuickBooks Online already syncs with. Data refreshes every 24 hours, or on demand, and reflects what has been reconciled in your accounts.

Is Float secure enough to satisfy our IT team?

Float's connection to your accounting platform uses the official API and is one-way and read-only, so Float can see your data but can never edit, add or delete anything in your accounts. Access within Float is controlled with user roles, two-factor authentication is mandatory for Xero-connected users, and you can disconnect the integration at any time.

How much does Float cost?

Pricing depends on the plan and the number of entities you connect, and every plan starts with a free 14-day trial with no credit card required. Current plans and what each includes are on our pricing page.

Reserves and forecasts answer different questions, and the businesses that get caught are usually the ones that let a comfortable answer to "how much" stand in for an answer to "when". If your dates have started to matter, start a free 14-day trial and see this month's cash, mapped to the days it actually moves.