How budgets actually work

Most lost enterprise deals are not lost to a competitor. They are lost to a budget cycle nobody mapped.

The annual cycle. Strategy and targets are set three to six months before the year begins. Then the budget build — departments submit bids, and your deal has to exist as a line item here or spend the year fighting for scraps — then challenge and cuts, allocation, in-year management, and reforecast, where money genuinely moves. Two windows make new money easy: the build and the reforecast. Between them you are asking someone to find money already assigned elsewhere.

Budget holder, approver, signatory — three different people. The budget holder owns the money and commits within a limit. The approver authorises above it under a scheme of delegation, a documented ladder of thresholds often published by NHS trusts. The signatory executes the contract, frequently with no view on the merits. Hence the most useful question a beginner can learn: "What's the approval path for something of this size, and where does it stop?"

Capital versus revenue in the public sector. CDEL — capital departmental expenditure limit covers things lasting over a year; RDEL — resource (revenue) departmental expenditure limit covers running costs. They are separate cash limits; transfers require Treasury agreement and have become rare. So "we've got capital but no revenue" (or the reverse) is structural, not an excuse — and a vendor offering only one commercial structure loses these deals invisibly.

Use-it-or-lose-it, honestly. At cost-centre level the pressure is real: unspent allocations usually do not carry over, and an underspend reads as evidence you did not need the money. At departmental level strict annuality has been softened since the late 1990s, with multi-year settlements and limited carry-forward. So "they must spend it by March" is often true of the manager you are talking to and untrue of the department above them. Treat a year-end scramble as a chance to close something the buyer already wanted, not licence to sell them something they don't.

The dates that matter here. The NHS financial year runs 1 April to 31 March. Planning guidance is published the preceding autumn — the Medium Term Planning Framework for 2026/27 to 2028/29 appeared on 24 October 2025 — plans are built over winter and finalised in spring, and from 2026/27 operational capital allocations flow directly to providers. Practically: October to January is when next year's money is argued over, the highest-value window for a long-cycle seller. Pharma mostly runs the calendar year — AstraZeneca, Pfizer and their large-cap peers have 31 December year ends — with the exception that major Japanese pharma (Takeda, Astellas, Daiichi Sankyo) runs 1 April to 31 March, like the NHS. Biotech runs on funding rounds and grant windows (section 7).

Procurement thresholds as a budget artefact

Above certain contract values, UK public bodies must run a regulated procurement. Not a preference — law, and months of elapsed time. Under the Procurement Act 2023, for procurements started on or after 1 January 2026, inclusive of VAT:

Worked example 6. You propose £120,000 a year excluding VAT. A one-year contract is £144,000 including VAT — below the £207,720 sub-central threshold, so an NHS trust may buy it without a full regulated competition, though a central-government body could not. A three-year contract is £432,000 including VAT — over both.

Two rules sellers get wrong. The estimated value must include options and extensions, so "one year plus two optional years" is valued at three. And contract-splitting to duck a threshold is prohibited — never advise a public buyer on avoiding a procurement rule. One complication: healthcare services commissioned by NHS bodies run under the Provider Selection Regime, in force since 1 January 2024, which replaced competitive tendering for clinical services with five routes including three direct-award processes. Software, devices and goods stay under the Procurement Act.

Field note

The most expensive assumption a new rep makes is that a signed-off business case means signed-off money. A case can be approved in principle and then wait for the next planning round, or for a capital committee that meets quarterly, or for a procurement route that has not started. The question that surfaces this — "is the funding already in an approved budget line this year, or does it need to go into next year's plan?" — takes eight seconds and routinely saves a quarter of forecasting.

5. The business case your buyer must write

At some point your champion defends spending money on you in a room you are not in. Whatever you gave them is what they have.

What is in one. The problem, in their language, with evidence. The options, always including do nothing — the counterfactual is not zero, it is the cost of carrying on. The recommendation. The costs at full total cost of ownership, internal staff time and exit included. The benefits, split into cash-releasing and non-cash-releasing (section 7). The appraisal plus affordability year by year. The risks. The plan.

In the UK public sector this is formalised. HM Treasury's Green Book, updated 5 February 2026:

Voice from the field

"The Five Case Model is a framework for developing business cases. It sets out five 'cases', or 'dimensions', that are distinct but closely linked perspectives on the same proposal" — the strategic, economic, commercial, financial and management cases. And on the relationship between merit and money: "Options that do not deliver a proposal's objectives cannot represent value for money", assessed separately from the affordability of the proposal for the public bodies involved.
— HM Treasury, The Green Book (2026)

For larger NHS capital schemes this arrives in sequence: Strategic Outline Case → Outline Business Case → Full Business Case, each taking months and each with an approval committee. A deal at SOC stage is not closing this quarter, and saying so in a pipeline review is a mark of competence. Note too that value for money and affordability are separate tests: a proposal can be excellent value and unaffordable this year, which is why phasing, billing structure and start date are commercial levers rather than admin.

Who signs it. The budget holder builds it; a finance business partner — an accountant embedded in the department, and often your most important invisible stakeholder — checks the numbers; then an executive group, and above a threshold a board. Sellers who never meet the finance business partner are gambling.

Payback, ROI and NPV — one case, three lenses

Illustrative: a trust considering a digital pathology workflow platform. Year 0 implementation plus first-year licence £150,000. Years 1–3 licence £60,000 a year. Gross annual benefit £140,000 from year 1 (reduced outsourced reporting, fewer locum sessions). Net annual benefit = £80,000 in each of years 1–3.

Payback period — time to recover the outlay, and the measure non-financial approvers grasp fastest.

£150,000 ÷ £80,000 = 1.875 years ≈ 22.5 months.

Weakness: it ignores everything after the payback point, so a project returning nothing in year 4 looks identical to one returning millions.

ROI — return on investment — total net gain as a percentage of spend.

Net benefit £80,000 × 3 = £240,000; cost £150,000; net gain £90,000. ROI = £90,000 ÷ £150,000 = 60% over three years.

Weakness: silent on timing, and quotable over any horizon — which is why vendors quote five years and buyers ask about three. An ROI without a stated period is marketing, not a number.

NPV — net present value — the whole cash-flow stream in today's money, discounted at the Green Book's social time preference rate of 3.5% in real terms for years 1 to 30. Discount factors: year 1 = 0.9662, year 2 = 0.9335, year 3 = 0.9019.

PV of benefits = £80,000 × (0.9662 + 0.9335 + 0.9019) = £80,000 × 2.8016 = £224,131. NPV = £224,131 − £150,000 = £74,131.

Against the undiscounted net gain of £90,000, discounting removed about £16,000 — roughly 18% — purely because benefits arrive after the cost. A positive NPV means value is created. It is the most rigorous and least intuitive measure, which is why good cases lead with payback and support it with NPV.

Arming your champion. Give them a model, not a number — a spreadsheet with visible, editable assumptions. A single confident figure invites "where did that come from?"; a model invites "let's test that assumption", which your champion can win. Flag conservative assumptions, use their volumes and costs, include the do-nothing cost, and keep cash-releasing separate from non-cash-releasing. Add a one-page summary and an honest risk section.