How to Build a Business Case for a Product Bet

A business case has a bad reputation, and mostly it is deserved. In many companies it is a spreadsheet built after the decision, to justify it, with a hockey-stick revenue line and assumptions nobody examined. The people reading it know this, so they discount everything in it, and the document fails at the one thing it exists to do: help someone decide.

A good business case is a different object. It is a structured argument that makes the assumptions behind a decision visible, shows how the answer changes if those assumptions are wrong, and gives the decision-maker a clear way to commit, to commit partially, or to say no. It is more useful when it is honest about uncertainty than when it is confident.

I have built these for decisions of several kinds. At a NYSE-listed edtech company, I took a new B2B offering from market research and a pricing model through A/B-tested landing pages to a soft launch, and it grew into a new business division. I also sized markets for international expansion, built a US location and demand study that fed marketing recommendations, and delivered the research behind a global expansion plan taken to C-level. The method below is what I use. It works for a feature, a new product, a market entry or a build-buy-partner choice.

It pairs with from research to recommendation, which covers how to present the case, and with build, buy, or partner, which uses the same cost logic. If you want to try the arithmetic as you read, the product business case calculator runs the worked example from this post and lets you change every input.

What a business case is for

A business case answers four questions for the person who has to decide:

1. What are we proposing, and why now?
2. What will it cost, and what do we get back, and when?
3. How sure are we, and what could make us wrong?
4. What exactly are we asking for, and how will we know whether it is working?

Everything in the document should serve one of those. If a section does not help answer one, remove it.

It is also worth being clear about what a business case is not. It is not a forecast in the sense of a prediction, because early-stage numbers cannot be predicted. It is a model of the logic: if these assumptions hold, here is what follows. Its value is in forcing the assumptions into the open.

Step 1: Frame the decision and the options

Start with the decision, in one sentence, and the options on the table. “Should we build a B2B tier for our product, and if so, how much do we commit now?” is a decision. “Evaluate B2B” is not.

Always include the do-nothing option as a baseline. The case for any investment is the difference between that investment and the alternative, and doing nothing is rarely free: customers leave, competitors move, a window closes. Describe what happens to the business if you do not act, and use it as the comparison throughout.

State the strategic fit in a few lines. How does the proposal connect to the company’s strategy and its guiding choices? A project that is financially attractive but contradicts the strategy deserves a hard look, and the way to judge that is in how to write a product strategy that actually gets implemented.

Step 2: Size the opportunity from the bottom up

The top of the case is the size of the prize. The most common error is a top-down figure: “the market is 50 billion, we only need one percent”. That number says nothing about whether you can win a single customer.

Build the opportunity bottom-up instead. Count the customers you could realistically reach, estimate how many you can convert at what price, and let the total emerge from the build. The structure is:

Reachable customers (a number you can source or name) times the conversion you can defend (from comparable evidence, not hope) times the price you can charge (from deals, tests or credible comparisons) times retention over time.

Every term should be grounded. Use real counts, not estimates of the whole world. Use conversion rates from analogous launches, early tests or pilots. Use prices from customer conversations or competitors. The method in market sizing for expansion covers the full build, and the market sizing calculator handles the TAM, SAM and SOM arithmetic.

For the B2B offering I mentioned, the evidence stack was market research, a pricing model and A/B-tested landing pages. That mattered because each stage replaced an assumption with a measurement before the full commitment. A business case gets stronger every time a number moves from “we believe” to “we tested”.

Step 3: Model the ramp, not the end state

A business case that jumps straight to a mature run rate is hiding the hard part, which is the path. Revenue arrives gradually, customers take time to sign and costs often arrive first.

Model the ramp explicitly, period by period, usually by year for a first pass and by quarter or month for the early stage. Questions to answer:

• How many customers or accounts do we add in each period, and why that many?
• How does acquisition cost behave as we scale: does it fall with learning or rise as the easy customers run out?
• What is the lag between spending and revenue?
• How does retention change the base? Model churn, or work in net terms and say so.

Be wary of straight lines and sudden acceleration. If the plan needs customer growth to double each year, ask what has to be true for that, and who has done it.

Step 4: Count all the costs

Benefits attract attention and costs get shortchanged, so be systematic.

One-time investment. Build or buy costs, set-up, migration, launch, legal and compliance work. Use fully loaded people costs.

Running costs. Hosting and infrastructure, support, maintenance, licences, and the people who run it. Maintenance is chronically underestimated.

Acquisition costs. Sales, marketing and partner commissions to win each new customer. Tie this to the ramp: a cost per new account, multiplied by the accounts you add.

Cost of variable delivery. Anything that grows with usage, such as payment fees or compute. For an AI feature this can dominate, and the LLM cost calculator helps you estimate it.

Opportunity cost. What the same people and budget would produce elsewhere. It is rarely in the spreadsheet and often the largest cost. At the least, name it in words.

Cost of delay and of change. Training, migration and the adoption effort that comes with any new way of working, as in change management for new martech.

The same total-cost thinking applies to the options in build, buy, or partner, where it matters even more because the options are priced so differently.

Step 5: Choose the right financial measures

You do not need a finance degree, but you do need to know what each measure tells you and what it hides.

Net cash flow by period. Revenue less costs, each period. The foundation for everything else.

Payback period. How long until cumulative cash flow turns positive. It is intuitive and tells you how long the money is at risk, and it ignores everything after payback and the time value of money.

Net present value, NPV. The sum of each period’s cash flow discounted back to today at a rate that reflects the cost of capital and the risk. A positive NPV means the investment returns more than the required rate. The formula is the sum of cash flow in each period divided by one plus the discount rate, raised to the power of that period. Choose the discount rate deliberately and state it. Riskier ventures deserve higher rates.

Return on investment and internal rate of return. Useful for comparing projects of different sizes. IRR is the discount rate at which NPV is zero.

Break-even. The level of volume, price or conversion at which the project stops losing money. It is the most useful number in a conversation about risk, because it answers the question “how wrong can we be?”.

Unit economics. Contribution per customer and the relationship between lifetime value and the cost of acquiring them. I cover how to compute and interpret these in LTV and CAC, and the LTV:CAC calculator does the arithmetic. If unit economics do not work at small scale, they will not rescue the case at large scale.

Two measures together are usually enough for a first discussion: NPV to show value creation and payback to show exposure.

Step 6: Run scenarios and sensitivity

A single set of numbers is a claim. A range is an argument. Present at least three scenarios.

Base case. Your honest best estimate.

Downside case. What happens if the main assumptions turn out materially worse. I usually define it concretely, for example volumes at half the plan, not vaguely as “conservative”.

Upside case. What happens if things go better. Include it, but spend less time on it.

Then run sensitivity analysis: change one assumption at a time and see how much the result moves. The assumptions that move the result most are the ones that deserve the most evidence. This is how you decide where to spend the next week of research, and which assumptions to test first with a pilot or an experiment, using a rigorous approach such as in experiment prioritization.

A useful discipline from forecasting research is to compare your assumptions with a reference class, the outcomes of similar projects. Studies of large projects show that costs and timelines tend to be underestimated and benefits overestimated, so if your plan is more optimistic than comparable ones, you should be able to say why.

A worked example

The numbers below are illustrative, not a client. They come from the product business case calculator, where you can change any of them.

The proposal. A software company with a consumer subscription is considering a B2B tier.

Assumptions.

• Build cost: 400,000 euros, spent up front.
• Price: 12,000 euros per account per year.
• Accounts at the end of each year: 40, then 110, then 200 (net of churn).
• Gross margin: 75 percent.
• Fixed running cost: 150,000 euros per year.
• Acquisition cost: 3,000 euros per new account.
• Discount rate: 12 percent.

Cash flow by year. Year 0 is minus 400,000. Year 1: revenue of 480,000 at 75 percent gives 360,000, less 150,000 fixed and 120,000 acquisition (40 new accounts), so 90,000. Year 2: 990,000 less 150,000 less 210,000 (70 new accounts), so 630,000. Year 3: 1.8 million less 150,000 less 270,000 (90 new accounts), so 1.38 million.

Result, base case. NPV of about 1.16 million euros. Payback in about a year and a half. Cumulative cash of about 1.7 million by the end of year three.

Downside case. If the company wins only half the planned accounts, year 1 turns negative at about minus 30,000, NPV falls to roughly 200,000 euros and payback stretches to about two and a third years. It is still positive, which says the case is resilient.

Sensitivity. Hold everything else at base and change one input at a time:

• Price falls to 10,000 euros: NPV about 770,000.
• Gross margin falls to 65 percent: NPV about 850,000.
• Acquisition cost rises to 5,000 euros: NPV about 850,000.
• Build cost rises to 600,000 euros: NPV about 960,000.

Break-even. NPV reaches zero when the company wins about 40 percent of the planned accounts.

What this tells the decision-maker. The case does not depend on any one favorable assumption. The biggest risks are on the revenue side, so the right first step is not full funding but a pilot designed to test willingness to pay and conversion. That leads directly to a staged ask.

Step 7: Define the ask, the stages and the kill criteria

A business case should end with a specific request, not a general endorsement.

Stage the investment. Instead of “approve 400,000 euros”, propose a first tranche to answer the biggest uncertainty, such as a pilot with a handful of design customers, with a second tranche released if defined evidence appears. Staging turns a large irreversible bet into a series of smaller reversible ones, and options have real value when uncertainty is high.

Set success metrics. A small number of leading indicators you can read early (pilot conversion, activation, willingness to pay) and the lagging outcomes you care about (revenue, retention, margin).

Agree kill criteria in advance. Write down the signal that would tell you to stop: if pilot conversion is below a stated level after a stated period, the project ends or changes shape. Agreed beforehand, this is a decision rule. Left until later, it becomes a political argument about sunk costs.

Name the owner and the review date. A case without an owner is a document. Set when the numbers will be revisited against actual results.

Step 8: Present the case so it gets read

Most of this guidance is covered in from research to recommendation, and a few points are worth repeating here.

• Lead with the decision and the headline numbers. One page, then the detail.
• Show the range. Base, downside and upside, with the break-even.
• Make the three most important assumptions explicit. Say how you know them and how you will test them.
• Include non-financial value, separately. Strategic benefits such as learning, option value or defense against a competitor are real, but do not disguise them as financial figures. State them honestly and let the decision-maker weigh them.
• Keep the model open. Share the spreadsheet. A model people can change builds trust. A closed one invites suspicion.

Common mistakes

Building the case to justify a decision already made. If everything points the same way and nothing is uncertain, the reader will notice.

Top-down sizing. “One percent of a huge market” is not a plan.

Hockey-stick ramps. Ask what has to be true, and who has done it before.

Forgetting maintenance and opportunity cost. Both are real and both are usually missing.

One scenario. A single outcome is a prediction you cannot defend.

Ignoring the do-nothing baseline. The comparison should be against what happens otherwise.

No kill criteria. Without them, projects continue long after the evidence says otherwise.

False precision. Reporting NPV to the nearest euro on guesses is theater. Round, and show the range.

The short version

• A business case is a structured argument that makes assumptions visible and shows how the answer changes if they are wrong. It is not a prediction.
• Frame the decision, include the do-nothing baseline, and check strategic fit.
• Size the opportunity bottom-up: reachable customers, defensible conversion, supported price and retention.
• Model the ramp period by period, count all costs including maintenance and opportunity cost.
• Use NPV for value and payback for exposure, and find the break-even.
• Run base, downside and upside scenarios and a sensitivity analysis to learn which assumptions matter most.
• End with a staged ask, success metrics, kill criteria, an owner and a review date.
• Present it on one page with the range, the key assumptions and an open model.

Where to go next

To turn the case into a recommendation people act on, read from research to recommendation. To apply the cost logic to a specific choice, build, buy, or partner is the companion. And for the strategy that decides which bets deserve a business case in the first place, return to how to write a product strategy that actually gets implemented.

This post is part of the product strategy guide, a reading path through all seven guides and the free tools that go with them.

---

I am Deepanshu Grover, a Senior Product Strategy Consultant in Paris. I take ambiguous, high-stakes product and business questions, run the analysis, and see the recommendation implemented. If you are building a case for a product bet, connect on LinkedIn or get in touch.