Guide

How to write a business case

A business case exists to let someone decide. This is a practical guide to writing one that survives the room: what goes in it, how to do the financial analysis properly, and the mistakes that sink otherwise sound proposals. There is a free editable template at the end, and you can grab it now if that is all you came for.

What a business case is, and what it is not

A business case is the reasoning behind spending money on something. It sets out the problem, the options, the recommended one, what it costs, what it returns, and what could go wrong. Its job is not to describe the project. Its job is to give someone enough to say yes, no, or not yet, and to be able to defend that decision later.

That framing matters, because a lot of weak business cases are really project plans wearing a different hat. If your document explains how the work will run before it has established whether the work is worth doing, the order is wrong.

It is also easy to confuse with the documents around it:

DocumentAnswersWhen
Business caseShould we fund this, and which option?Before approval
Project brief or one-pagerIs this worth investigating at all?Before the case
Feasibility studyCan this actually be done?Feeds the case
Project charterWho is accountable, and what is in scope?After approval
Delivery planHow and when will it be built?After approval

When you actually need one

Not every spend deserves a formal case, and writing one where it is not warranted is a good way to get them ignored. You need a business case when at least one of these is true:

  • The spend is above your organization's approval threshold
  • It competes with other initiatives for the same pot of money
  • The benefit is contested, or several teams claim credit for it
  • The payback runs beyond a single budget year
  • Someone will be asked, later, whether it was worth it

You probably do not need one for small, reversible, already-budgeted work. A short brief is enough, and the credibility you save is worth more.

How long it should be

As short as it can be while still answering the questions a reviewer will ask. Most business cases land between five and fifteen pages, with a one-page executive summary that stands on its own. Length is not rigor. A twenty-page case with an invented discount rate is weaker than a six-page one where every number traces to a stated assumption.

Assume the person deciding will read the summary carefully, skim the middle, and go straight to the financials and the risks. Write accordingly.

The sections, in order

This is the structure the template follows. The order matters: each section earns the right to the next one.

1. Executive summary

Write it last. One page: the ask, the problem, the recommendation, the return, and the decision you need. The mistake: introducing a claim here that the rest of the document does not support.

2. The problem or opportunity

What is broken, what it costs today, and the evidence for both. Include what happens if you do nothing, honestly. The mistake: opening with the technology. A case that starts from the capability rather than the problem invites the question you least want.

3. Strategic alignment

Name the specific business objective this serves, not a general aspiration. The mistake: a paragraph of strategy language that would fit any initiative in the portfolio.

4. Options considered

Include doing nothing as a real option with real consequences, plus two or three genuine alternatives. Where it applies, say whether each is build, buy, or hybrid. State the comparison criteria before the verdict. The mistake: two strawmen and the answer you had already chosen. Experienced reviewers spot this instantly, and it damages everything else in the document.

5. The recommended option

Which one, why it beats the others on the stated criteria, and why not the alternatives. The mistake: recommending one option and then costing a different one. More on this below, because it is the single most common inconsistency in business cases.

6. Scope, approach, and requirements

What is in, what is out, the phases, and the requirements that are genuinely load-bearing: performance, availability, security, accuracy, where data is allowed to live. The mistake: discovering that a hard requirement conflicts with the chosen approach during security review, when changing course is expensive.

7. People and change impact

Most initiatives fall over on adoption, not on technology. Set out who has to work differently, what that actually asks of them, and what it will take to get there: communication, training, manager involvement, support through the transition, and the dip in output while people learn. This section is load-bearing in two directions. It creates real cost lines, and it is the only thing that justifies the adoption rate your benefits depend on.

Five questions are worth answering explicitly, because reviewers ask them and most cases have no answer ready:

  • Where the freed capacity actually goes. If the benefit is time saved, name the destination: contractors released, roles not backfilled, more volume through the same team, or work that has been deferred for two years. Capacity that goes nowhere in particular is not a saving, it is a slightly quieter week.
  • Who owns and funds the upskilling. Name the budget holder, not just the activity. Training that belongs to nobody has a way of not happening, and the adoption assumption quietly fails with it.
  • How quality is protected during the ramp. Output usually dips before it improves. Say what you will monitor, what threshold you consider acceptable, and what triggers a pause.
  • Whether the knowledge foundation is ready. New ways of working assume documented process, reliable data, and people who know where things are. Where that groundwork is missing, it becomes part of this initiative whether you planned for it or not.
  • What the reinvestment alternative looks like. If this frees capacity or money, what else could it fund? Naming the alternative makes the recommendation stronger, because it shows the comparison was made rather than avoided.

The mistake: treating change as something to fund later. If the benefit depends on people behaving differently and the case has no plan for making that happen, the benefit is a wish.

8. Costs

Build the cost from its parts. Separate what you pay once from what you pay every year, and say how firm each line is. A signed quote and an educated guess should not look identical on the page. The mistake: a single round number with no visible arithmetic behind it.

9. Benefits

Quantify what can honestly be quantified, and report the rest as outcomes rather than inventing a dollar figure. Apply two discounts to every financial benefit: adoption and attribution. The mistake: assuming everyone uses it and that nothing else contributed.

10. Financial analysis

Cash flows by year, then net present value, internal rate of return, and payback. Then sensitivity. Covered properly in the next section.

11. Assumptions and risks

Number the assumptions so the financial section can point at them, and give every risk a named owner. The mistake: an assumption nobody can find is an assumption nobody can challenge, which is worse than a weak one.

12. Success measures and governance

What you will measure, the baseline as it stands today, the target, and when it gets checked. Then who signs.

Say what happens if performance misses the forecast: at what threshold it gets reviewed, who decides, and what the options are between carrying on and stopping. The mistake: two of them here. Not recording the baseline before you start, which makes the benefit unprovable afterward, including to yourself. And describing only what success looks like, which leaves nobody able to hold the case to account when it does not arrive.

The financial analysis, done properly

This is the section that gets skipped or fudged most often, and it is the one a finance reviewer reads first.

Start with the discount rate

Ask your finance team for the organization's hurdle rate or weighted average cost of capital. Do not choose one yourself. An invented discount rate invalidates every number that depends on it, and it is trivially easy for a reviewer to challenge.

Lay out cash flows by year

Not a single total. Year zero through year five, with one-time costs, recurring costs, gross benefits, net benefits after adoption and attribution, net cash flow, and discounted cash flow. The shape of that curve is what payback depends on.

Then the four numbers

MetricWhat it tells youWhere it misleads
Net present valueWhether the initiative creates value at your cost of capital. Positive is good.Says nothing about scale of effort or risk. A tiny positive NPV is not a reason to proceed.
Internal rate of returnThe discount rate at which NPV hits zero. Compare it to the hurdle rate.Goes absurdly high when the initial outlay is small, and breaks entirely when cash flows change sign more than once.
Payback periodWhen cumulative cash flow turns positive. Executives love it.Ignores everything after the payback point, so it quietly favors short, small initiatives.
Cost of delayWhat a month of not deciding costs you. Underused, and unusually persuasive.Only credible if the benefit itself is credible.

Report more than one. Any single metric can be gamed, and a reviewer who sees all four together can tell quickly whether the case is sound.

Finish with sensitivity

Ask which single assumption, if it turned out to be wrong, would change the decision. Then show the case at a downside, base, and upside value for it. That one table pre-empts most of the questions you would otherwise face live, and it signals that you went looking for the weakness yourself.

Where business cases actually fail

Structure is the easy part. These are the failures that get cases sent back, in rough order of how often we see them.

Hours saved presented as money saved

Twenty minutes back for four hundred people is not a financial benefit until you say what happens to that time. Fewer contractors, more output, or nothing yet. All three are acceptable answers. Silence is not.

Adoption assumed at 100 percent

Almost no benefit lands across the whole population, and almost no case says so. State the adoption rate you are assuming and where the figure came from.

The change effort is nowhere in the numbers

The benefit assumes people work differently, but the cost contains no communication, no training, no manager time, and no allowance for the productivity dip while everyone learns. Adoption does not happen for free. A case that budgets for the system but not for the switch tends to miss on both.

Attribution never stated

When three initiatives all count the same reduction in handling time, the portfolio promises more than the business will ever bank. Say what share of the outcome this initiative is claiming.

The recommended option is not the costed option

The case recommends option two and then finances a number derived from option one, usually because the estimate was built before the options were compared. It is easy to miss internally and very easy to spot in review.

Recurring costs that do not match their own arithmetic

If a line is priced per user per month, the annual figure has to equal unit cost multiplied by users multiplied by twelve. A stated total that disagrees with its own unit economics is a common error and an expensive one, because it usually understates the run cost.

Cost stated with no confidence attached

A firm quote and a rough estimate carry very different risk. Blending them into one number throws away the information a reviewer needs to judge how much contingency the case deserves.

Options that were never real

If the alternatives were assembled after the decision, the options section is doing no work. Reviewers read it precisely to see whether the thinking was genuine.

No baseline, so nothing can be verified

Without a recorded measure of how things stood before, nobody can prove the benefit landed. This is why organizations end up unable to say whether their last five investments paid off, and why the sixth is harder to fund.

The template

An editable Word document covering every section above, with the tables already built for options, costs, benefits, cash flow, assumptions, risks, and sign-off. Guidance notes are included in the file and are meant to be deleted as you write. No email required.

Download the business case template
Common questions

Business cases, answered plainly.

What is a business case?

A business case is the reasoning behind committing money to an initiative. It sets out the problem being solved, the options considered, the recommended option, what it will cost, what it is expected to return, and the risks and assumptions behind those numbers. Its purpose is to let a decision-maker approve, reject, or defer with confidence, and to defend that decision afterward.

What should a business case include?

An executive summary, the problem and what it costs today, strategic alignment, the options considered including doing nothing, the recommended option and why, scope and approach, the people and change impact, costs split between one-time and recurring, quantified benefits discounted for adoption and attribution, financial analysis with net present value and payback, assumptions and risks, success measures with a baseline, and sign-off.

How long should a business case be?

Most land between five and fifteen pages, with a one-page executive summary that stands on its own. Length is not rigor. A short case where every number traces to a stated assumption is stronger than a long one built on figures nobody can check.

What is the difference between a business case and a project charter?

A business case answers whether the work should be funded and which option to pursue, and it is written before approval. A project charter answers who is accountable and what is in scope, and it is written after approval. If your business case is mostly describing how the work will run, it has drifted into charter territory before earning the right to.

How do you calculate ROI in a business case?

Lay out costs and benefits as cash flows by year, discount them at your organization's hurdle rate or weighted average cost of capital, then report net present value, internal rate of return, and payback period together rather than relying on any one of them. Discount the benefit side for adoption and attribution before it reaches the financials, or the return will be overstated.

How do you account for change management in a business case?

Treat it as both a cost and a condition of the benefit. On the cost side, budget for communication, training, manager time, and support through the transition, plus the temporary drop in productivity while people learn. On the benefit side, the adoption rate you assume has to be justified by that plan rather than asserted. A case with no change effort in it has nothing holding its adoption assumption up.

Who writes the business case, and who approves it?

Usually the sponsoring manager or a business analyst writes it, with finance validating the numbers and the affected functions contributing costs and assumptions. Approval sits with whoever holds the budget, which above a certain threshold typically means an investment committee or executive group rather than one individual.

Or have the case built with you.

The template gives you the structure. Arivue drafts the case itself, computes the financials, and checks the whole thing for the inconsistencies above before anyone else sees it.