You've found a genuine AI opportunity. The technology works, the data is there, and your team is ready. But none of that matters if you can't get the investment approved.
Most AI business cases fail not because the idea is bad, but because the case is poorly constructed. Technical teams write proposals that read like research papers. They focus on model architectures and accuracy metrics while executives want to know three things: How much will it cost? What will it return? And what happens if it doesn't work?
Here's how to build an AI business case that actually gets approved.
Understanding the Executive Perspective
Before you write a single slide, understand what your decision-makers care about. Different executives evaluate AI investment through different lenses:
- The CEO wants strategic advantage. How does this AI initiative strengthen competitive positioning? Does it align with the company's growth strategy? Will it be a talking point with the board and investors?
- The CFO wants financial rigour. What's the total cost of ownership? When does it break even? What are the ongoing costs? How does the ROI compare to other investment options?
- The COO wants operational impact. How does this change day-to-day operations? What's the disruption during implementation? How does it scale across the organisation?
- The CTO/CIO wants technical feasibility. Does it integrate with existing systems? What are the security and compliance implications? Is the organisation technically ready?
Your business case needs to address all four perspectives. Lead with the one that matters most to your primary decision-maker, but cover them all.
Structuring the Business Case
A strong AI business case follows a clear structure. We use the AI Business Case template as a starting framework, but the core sections are:
- Problem statement: Define the business problem in commercial terms, not technical ones. "Our customer onboarding process takes 14 days and costs £340 per customer" is far more compelling than "we need to implement NLP for document processing."
- Proposed solution: Describe what the AI system will do in plain English. Avoid jargon. Executives don't need to understand transformers — they need to understand that the system will read documents, extract key information, and populate forms automatically.
- Financial model: Total investment required (development, infrastructure, data preparation, change management), expected returns (broken down by value type), and timeline to break even. Use conservative estimates — overdelivering beats underdelivering every time.
- Risk assessment: What could go wrong and how you'll mitigate it. More on this below.
- Implementation plan: A phased approach with clear milestones and decision points. No executive wants to approve a £500K commitment with no off-ramps.
ROI Modelling That Survives Scrutiny
The financial model is where most AI business cases collapse. Here's how to build one that holds up:
- Use three scenarios: Conservative, expected, and optimistic. Lead with the conservative case. If the project makes sense at conservative estimates, it's a strong investment. If it only works at optimistic estimates, it's a gamble.
- Include all costs: Development, infrastructure, data preparation, testing, deployment, training, change management, ongoing maintenance, and model retraining. Undercounting costs is the fastest way to lose credibility when the real bills arrive.
- Show time to value: Executives care about when they'll see returns, not just how much. Plot the cumulative cost and benefit over 12-24 months. Mark the break-even point clearly.
- Benchmark against alternatives: What would it cost to solve this problem without AI? Hiring more staff? Outsourcing? Buying an off-the-shelf tool? Your AI solution needs to be better than the alternatives, not just better than doing nothing.
Framing Risk Effectively
Many AI business cases either ignore risk entirely or bury it in an appendix. Both approaches erode trust. Smart executives know AI is uncertain — they want to see that you've thought about what could go wrong.
Frame risk proactively:
- Technical risk: Model accuracy may not meet targets. Mitigation: phased approach with go/no-go gates after the pilot phase.
- Adoption risk: Users may resist the new workflow. Mitigation: change management plan, user involvement from day one, gradual rollout.
- Data risk: Data quality may be insufficient. Mitigation: data assessment in the first sprint before committing full budget.
- Regulatory risk: AI governance requirements may change. Mitigation: compliance review as part of the implementation, flexible architecture that can adapt.
The key principle is: every risk should have a mitigation, and the phased approach itself is your biggest risk mitigator. You're not asking for a leap of faith — you're asking for permission to take the first step, with decision points along the way.
Presenting to the Board
When it's time to present, keep these principles in mind:
- Lead with the problem, not the technology. Start with the business pain point and the cost of inaction. Make the board feel the problem before you propose the solution.
- Keep it to 10 slides or fewer. Executive summary, problem, solution, financial model, risks, implementation plan, team, and ask. Everything else belongs in the appendix.
- Show a working demo if possible. A 60-second demo of a prototype is worth more than 20 slides of explanation. Even a rough prototype built during a discovery phase makes the opportunity tangible.
- Make the ask specific. "We need £150K over 6 months, with a go/no-go decision at the 8-week mark after the pilot phase" is far better than "we need budget for AI."
- Prepare for the hard questions. "What if it doesn't work?" "Why can't we just hire more people?" "How is this different from the last tech project that went over budget?" Have clear, honest answers ready.
Need help building an AI business case? We work with leadership teams to develop investment cases that are financially rigorous and strategically compelling. Get in touch for a confidential discussion about your AI opportunity.