Business Case Example: A Filled-In Cost and Benefit Case over Three Years
A business case sets out what a project costs, what it is expected to return, and which assumptions those numbers rest on. The example below is a fictional replacement of an on-premise IT service management platform with a subscription service: eight cost lines, six benefit lines, three years, each line owned and tied to a numbered assumption.
The example
An organisation of roughly 3,000 staff replaces an ageing on-premise service management platform with a subscription product. Year 0 is the delivery year; years 1 to 3 are the run years used for the return. Costs are shown as positive numbers, benefits separately, and the two are netted at the bottom.
Each line carries an assumption reference. The assumptions themselves live in their own log — A1 through A14 here — which is where the case is actually argued. Figures are illustrative and the organisation is invented.
| Ref | Line | Type | Year 0 | Year 1 | Year 2 | Year 3 | Owner | Assumption |
|---|---|---|---|---|---|---|---|---|
| C01 | Subscription licences | Cost, recurring | 0 | 84,000 | 86,500 | 89,100 | IT Service Owner | A1: 750 agent and approver licences, 3% annual uplift |
| C02 | Implementation partner | Cost, one-off | 145,000 | 20,000 | 0 | 0 | Programme Manager | A2: fixed-price statement of work plus four weeks hypercare |
| C03 | Internal delivery effort | Cost, one-off | 96,000 | 12,000 | 0 | 0 | Programme Manager | A3: 640 person-days at a blended internal rate |
| C04 | Data migration and cleansing | Cost, one-off | 38,000 | 0 | 0 | 0 | Data Lead | A4: six years of ticket history, two dry runs |
| C05 | Integration build | Cost, mixed | 42,000 | 6,000 | 6,000 | 6,000 | Integration Lead | A5: three interfaces on vendor connectors, annual maintenance |
| C06 | Training and change | Cost, mixed | 27,000 | 4,000 | 4,000 | 4,000 | Change Lead | A6: 210 agents trained, 1,400 self-service users briefed |
| C07 | Legacy platform run-off | Cost, recurring | 61,000 | 15,000 | 0 | 0 | IT Service Owner | A7: parallel run to month 4, then decommission |
| C08 | Contingency | Cost, one-off | 34,800 | 0 | 0 | 0 | Programme Manager | A8: 10% applied to C02 to C06 only |
| B01 | Legacy licence and support avoided | Benefit, cash | 0 | 96,000 | 98,000 | 98,000 | IT Service Owner | A9: current support contract ends month 12 |
| B02 | Server and hosting withdrawal | Benefit, cash | 0 | 31,000 | 31,000 | 31,000 | Infrastructure Manager | A10: nine virtual machines across two environments |
| B03 | Agent handling time reduction | Benefit, non-cash | 0 | 44,000 | 88,000 | 88,000 | Service Desk Manager | A11: time released, not headcount removed |
| B04 | Self-service deflection | Benefit, non-cash | 0 | 18,000 | 55,000 | 62,000 | Service Desk Manager | A12: adoption ramps over 18 months from go-live |
| B05 | Audit remediation avoided | Benefit, cash | 0 | 25,000 | 0 | 0 | Head of IT Risk | A13: quoted remediation on an existing audit finding |
| B06 | Shorter major incident duration | Benefit, non-cash | 0 | 0 | 40,000 | 40,000 | IT Operations Manager | A14: baseline measured during year 1, benefit claimed from year 2 |
| — | Total cost | Cost | 443,800 | 141,000 | 96,500 | 99,100 | Programme Manager | Sum C01 to C08 |
| — | Total benefit | Benefit | 0 | 214,000 | 312,000 | 319,000 | Sponsor | Sum B01 to B06 |
| — | Net position | Net | −443,800 | 73,000 | 215,500 | 219,900 | Sponsor | Benefit less cost |
| — | Cumulative net | Net | −443,800 | −370,800 | −155,300 | 64,600 | Sponsor | Payback in month 33 |
Reading the example
Cost lines
Eight lines, split between one-off and recurring, because the two behave differently after approval. One-off cost is what the delivery budget is measured against; recurring cost is what somebody inherits. Two entries are easy to leave out and expensive to forget: C07, the year of parallel running before the old platform can be switched off, and the tail on C05 and C06, where integration maintenance and refresher training carry on after the project has closed.
Contingency
C08 is a single visible line at 10%, applied only to the one-off delivery lines and not to the subscription. Holding it as its own row rather than spreading it across estimates means the release of contingency can be tracked. It is also the line most often argued about, which is easier when it is one number rather than an unstated margin inside six others.
Cash and non-cash benefits
The split matters more than the totals. B01, B02 and B05 are cash: a contract ends, machines are switched off, a remediation spend does not happen. B03, B04 and B06 are time released rather than money removed, and A11 says so directly. On cash benefits alone this case does not pay back inside three years — cumulative net stays at about −34,000 at the end of year 3. Whether that is acceptable is the sponsor decision the case exists to inform, and stating both figures is what allows the decision to be made honestly.
Owners
Every line has a name, including the benefit lines. This is the column that survives longest: the project ends, the benefits do not, and the person named against B04 is the one who will be asked in eighteen months whether deflection reached the assumed level. Benefit owners sitting outside the project team is normal and intentional.
Assumptions
Every figure points at a numbered assumption. A11 and A12 carry most of the weight in this case: a ramp profile and a claim about released time. When a case is challenged, it is almost always the assumption that is challenged rather than the arithmetic, so keeping them numbered and separate makes the argument reviewable. The Business Case Template keeps the assumptions log on its own tab for that reason.
What this example leaves out
There is no discounting. Net present value, internal rate of return and a discount rate set by finance would change the payback month and are usually required for capital approval above a threshold. The case here shows undiscounted cash flows, which is enough to frame the decision and not enough for a finance committee that asks for NPV.
It shows one option. A complete case sets the preferred option against at least a do-nothing baseline and one alternative — extend the current platform, or move to a different product — and the comparison is often where the real argument sits. Where the alternative is a supplier choice, the scoring behind it belongs in a vendor selection record rather than in the case.
It has no sensitivity analysis. What happens if the deflection ramp takes 30 months instead of 18, or if the implementation partner cost lands 20% over, is the question a reviewer asks second. The single-point figures here do not answer it.
And it carries no realisation record. The case ends at approval; the benefits register that measures B01 to B06 against these numbers runs for years afterwards, usually owned by someone who was never in the project. Project closure is where that handover happens, and the budget and vendor tracker is where the cost side is followed during delivery.
Questions
Should non-cash benefits be included in a business case?
They can be, as long as they are labelled. Time released is real but does not appear in a budget, so cases that mix it with cash savings without saying which is which tend to be challenged later.
How many years should a business case cover?
Commonly three to five, matched to the life of the thing being bought. A subscription with a three-year term and a three-year case is easier to defend than a ten-year projection built on the same assumptions.
Where does contingency belong?
As a visible line, applied to the estimates that carry uncertainty rather than to everything. Spreading it inside individual estimates makes it impossible to track how much has been released.
Who owns the benefits after approval?
Usually an operational manager rather than the project, since the project will have closed before most benefits are measurable. Naming that person in the case itself is what makes the handover at closure possible.