Strategic case
Why is change needed, and what should improve?
Economic case
Which options offer the best balance of value and risk?
Commercial case
Which supplier, contractual and sourcing choices are appropriate?
Financial case
Is the investment affordable over its full life cycle?
Uitvoeringscase
Are governance, people, planning and measurement in place?
Start with the decision you need to make
A business case should not justify a tool already chosen. It should help decision makers choose between doing nothing, improving the process without AI, running a limited trial or implementing a broader solution.
First define the business problem, desired outcome and reference scenario. Without a business-as-usual comparison, you cannot judge what the change really adds.
Sources for this section: HM Treasury · HM Treasury
Build on a verifiable baseline
Measure current volumes, processing times, errors and rework, quality, staff effort and relevant risks. Use a representative period and record exceptions. A snapshot from a peak or quiet month can give a misleading starting point.
Then show which figures are confirmed, which still need validation and which remain uncertain. This prevents assumptions from quietly becoming treated as facts later.
Include full costs and achievable benefits
Beyond licences or model usage, account for process analysis, data work, integrations, testing, security, legal review, training, change management, monitoring and maintenance. Internal staff time is a real commitment too, even without an external invoice.
Time saved is not automatically money saved. It creates value when freed capacity is used for work that would otherwise not happen, external costs fall or growth can be handled without additional hiring. Better quality and lower risk may also matter; explain how you will observe them.
Use scenarios and decision points
Prepare at least a cautious, central and favourable scenario. Vary only assumptions that materially affect the decision, such as adoption, achievable time savings, exception rates, management effort and implementation time.
Agree in advance which trial outcomes will lead you to continue, adjust or stop. This keeps the pilot from becoming open-ended and lets you invest in stages as uncertainty falls.
Sources for this section: HM Treasury · NIST
Illustrative example: an AI assistant for document analysis
A fictional team considers an assistant that prepares documents for review. It assumes staff handle similar documents and an authorised employee retains the final decision.
The business case compares doing nothing, improving templates and search, and introducing an AI assistant. The baseline covers handling time, rework and end-to-end time. The trial measures accuracy, acceptance, exceptions and actual time saved. Integration is considered only if the basic version meets the agreed criteria.
When should you delay investment?
Delay investment if no one owns the problem, the baseline cannot be reproduced, benefits exist only as theoretical time savings or essential management and adoption costs are missing.
Start with a concise decision-level business case: the problem, baseline, options, critical assumptions, full costs, measurement plan and next decision point. The level of detail should match the scale, complexity and risk.
Frequently asked questions
How do you establish a baseline?
Choose a representative period and measure volume, time, quality, rework, costs and relevant risks. Record definitions and exceptions so the measurement can be repeated.
Which costs are often overlooked?
Internal process time, data work, integration, testing, security, training, change management, monitoring, maintenance and exit or supplier costs.
How do you determine achievable benefits?
Specify which capacity can actually be reassigned, which costs demonstrably fall and which improvements in quality or risk can be measured.
How do you define decision points?
Link each phase to acceptance criteria, a decision maker and three possible outcomes: continue, adjust or stop.
Sources
The sources below support the indicated factual and regulatory passages. The practical decision frameworks are professional recommendations from DSC Solution.
- HM Treasury — Guidance on developing business cases17 November 2022, updated 10 October 2023Used for: a coherent strategic, economic, commercial, financial and delivery case.
- HM Treasury — The Green Book (2026)5 February 2026Used for: comparing options, making uncertainty explicit, proportionality and evaluation.
- NIST — Artificial Intelligence Risk Management Framework (AI RMF 1.0)26 January 2023Used for: risk management, governance, and the ongoing assessment and management of AI risks.
- European Union — Regulation (EU) 2024/1689 — AI Act13 June 2024Used for: risk-based obligations, documentation, oversight and AI literacy.
