Make the case
AI business case template: fund the first step of a bigger change
An AI business case explains why a change is worth funding. For a service firm, that might mean a new way to sell its expertise. For a large business, it might mean giving more teams access to specialist support. Show how the first project moves you towards that goal, what it costs and what you need to learn.
Be clear about what you want to fund now. Show what it could earn or save, and which later gains still need proof.
Build your AI business case
Use the questions to write a short case for the people funding the work. Link to the facts they need to check. Keep it in one shared document.
Explain the business change before the software
Imagine a firm that helps clients set their strategy. Some clients then leave because they cannot afford to pay the firm to do all the work. It wants to offer regular planning support using its methods and each client’s current goals.
The first build would prepare one type of client plan for an expert to approve. The case must show why clients would buy that help and whether the firm can deliver it well at the proposed cost.
In a large business, the buyer may be an internal team. Show the work it cannot get done today, the delays involved and the effect of changing that. Do not invent internal revenue to make the case look like a new product.
Compare the idea with using tools you already own, changing the steps or adding staff. Judge each choice against the same business goal.
Worked example: a new planning service
All figures below are invented to explain the maths. They are not client results or LeverageAI prices. The example assumes a full year after launch. A real case must also show the build period.
The firm expects 20 clients to pay £1,000 a month. It allows £400 per client per month for expert review, service delivery and customer support. That leaves £600 before the shared cost of building and running the system.
| For one full live year | 20 clients throughout | 10 clients throughout |
|---|---|---|
| Sales | £240,000 | £120,000 |
| Direct delivery costs | £96,000 | £48,000 |
| Money left after delivery | £144,000 | £72,000 |
| Setup, including staff time | £60,000 | £60,000 |
| Shared running costs | £24,000 | £24,000 |
| Net benefit after these costs | £60,000 | −£12,000 |
Both columns assume clients stay for the full year. That is a big assumption. Before funding it, plan when clients are likely to join, leave and pay each month.
The difference shows why a good demo is not enough. Demand and the cost of serving each client can decide whether the whole idea works.
Count what it takes to change the work
Count the build, the work to get data ready, links to other systems, tests and staff time. Add training, software use, support and keeping records up to date.
For a large business, include the effort of bringing each team or region onto the system. Different access rules or ways of working can create extra work.
The example counts client delivery costs separately from shared running costs. Agree where each cost belongs so you count it once. Also check whether the new offer would replace work clients already buy from you.
Give future uses a clear place in the case
The first project may record your methods, link client records and set rules for expert review. A later service could use that work to help clients check their progress.
List what could be reused, who would use it and what would still need building. Keep that possible later return outside the first project’s sales forecast.
Where several projects share a cost, show it once in the overall plan and explain how it is divided. Counting the full benefit in every project would overstate the return.
Time saved also needs a destination. If it helps the team deliver extra paid work, count the money left after doing that work. Do not count the same hours again as wages saved.
A completed funding decision
This made-up case shows what the firm still needs to learn before paying for the full build.
Worked example
A funding decision for a new planning service
Decision requested
Fund a defined demand check and delivery test from an agreed research budget.
Do not approve the full build on forecast sales alone.
Goal
Give strategy clients ongoing help without a full-service engagement.
First build: prepare one type of plan, checked by an expert before release.
Proposed first-year case
20 clients paying £1,000/month for 12 live months = £240,000 sales.
Direct delivery = £96,000. Setup = £60,000. Shared running = £24,000.
Net benefit = £60,000, before tax and financing.
At 10 clients for the same period, net benefit = −£12,000.
All figures are invented assumptions, not quotes or observed results.
Evidence needed
Client interviews and a concrete buying decision for the proposed offer.
A trial showing the full expert review and support effort.
A monthly forecast including slow uptake and displaced existing work.
What carries forward
Client records, agreed planning methods, access rules and review steps.
Possible later service: progress reviews. No later sales counted here.
Owner
Service lead, with finance checking costs and technology checking access.
Next decision:
Review the evidence and costs before committing to the build.
Stop or change the offer if demand or delivery cannot support the case.
Agree the next investment
Ask for a choice people can act on: check one doubt, fund the build or change the plan. Name who makes that decision and when.
Our AI Strategy Workshop helps compare ideas and work through the return behind each one. We agree the scope and price of the build with you as a separate step. The workshop does not commit you to a build.
Further reading
Written by Said Lopez, founder of LeverageAI and author of Lean AI. Our guides draw on the method in the book and our work with clients.