How Do You Build an AI Business Case the Board Will Approve?
Expert review: · Content updated:
An AI business case for the board answers four questions in a fixed order: Why now? What does it deliver? Can we execute? What does it cost, and when do we see results? In a 2025 Deloitte survey of 1,854 executives in Europe and the Middle East, only 6% reported payback in under a year — promising six months anyway costs you credibility at the next budget request.
This guide was reviewed by an expert on 23 July 2026 and its content was updated on 31 October 2026.
Which four sections does the board expect?
An AI business case follows different logic than a classic project proposal. It answers four questions in this order — and no other: strategic imperative, business value projection, readiness & risk, investment & roadmap.
Why now and not later?
Market data, competitive position, the current window of capability — the frame that justifies urgency.
What does it deliver concretely?
Value drivers, KPI targets, ROI model and benchmarks — in numbers the board can verify themselves.
Can we execute this?
Readiness score, the three biggest risks and their mitigations — honest, not dressed up.
What does it cost, when do we see results?
Investment plan, milestones, quick wins, and a realistic ROI timeline.
Which KPIs fit which archetype?
A one-size-fits-all KPI framework fits nobody. AI Starters primarily measure readiness — whether the foundation holds. AI Scalers measure leading indicators — whether scaling is on track. AI Transformers measure lagging indicators — whether AI measurably contributes to results.
Only 6% report payback in under a year (Deloitte 2025)
13% even of the most successful projects: returns within 12 months (Deloitte 2025)
2–4 years to satisfactory ROI on a typical use case, for most respondents (Deloitte 2025)
| KPI type | AI Starter | AI Scaler | AI Transformer | Review cadence |
|---|---|---|---|---|
| Readiness KPIs | Data quality score, governance bootstrap | API coverage, MLOps maturity | Knowledge graph quality, agent governance maturity | weekly |
| Leading KPIs | First use case live, readiness score | Use cases in production, adoption rate | Use cases enterprise-wide, automation level | monthly |
| Lagging KPIs | EBIT impact, cost reduction | ROI per use case, time-to-value | Revenue lift, market share | quarterly |
Two design rules apply regardless of archetype: every KPI needs a baseline value — no measuring without a before-value. And a maximum of five active KPIs per use case; more creates reporting overhead, not insight. Readiness KPIs are reviewed weekly, leading KPIs monthly, lagging KPIs quarterly.
Board rule: communicate the ROI timeline honestly. A business case promising payback in six months loses credibility — and with it the next budget request. An honest timeline with clearly named quick wins convinces a board more reliably than an optimistic projection that gets disproven at the first quarterly review.
Glossary
- Lagging Indicator
- A trailing metric that measures the outcome only after the fact — for example EBIT impact or ROI per use case.
- Leading Indicator
- A forward-looking metric that shows early on whether an initiative is on track — for example adoption rate or the number of use cases in production.
- AI Value Canvas
- A template in AI Navigator that guides you through the four sections of an AI business case.
- Business Case
- A decision document that sets out the benefits, costs, risks, and timeline of an initiative so that the investment can be decided on.
Sources
- Deloitte: "AI ROI: The paradox of rising investment and elusive returns", 22 October 2025 (survey of 1,854 executives in Europe and the Middle East) — source of the 6%, 13% and 2–4 year figures — https://www.deloitte.com/global/en/issues/ai/ai-roi-the-paradox-of-rising-investment-and-elusive-returns.html
Fill in the business case using a template
The AI Value Canvas in the product guides you through all four sections and automatically maps KPIs to your archetype.
Open the Use-Case Scoring →FAQ
How long should an AI business case for the board be?
Three pages total: half a page imperative, one page business value, half a page readiness & risk, one page investment & roadmap. Anything longer rarely gets read.
What if ROI really only shows up after 3 years?
Name that honestly — and back it up with concrete quick wins in year one. In Deloitte's 2025 survey, most respondents reached satisfactory ROI on a typical AI use case only within 2 to 4 years; that's not a red flag, it's the norm.
Do KPIs change as a company moves from Starter to Scaler?
Yes, deliberately. Readiness KPIs fade into the background, leading KPIs move into focus — the framework grows with the archetype, not the other way around.
How many KPIs should a single use case show in the business case?
A maximum of five active KPIs. More per use case mostly creates reporting overhead, not more decision-relevant insight.
Daniel Ostner
From Chapter 3 of the Enterprise AI Guide book
Related guides
AI Maturity
Three archetypes summarize AI maturity: Starter, Scaler, Transformer. Plus a 5-level maturity model with measurable criteria instead of gut feeling.
Building AI Governance
Solid AI governance runs on three levels: strategic, tactical, operational. A minimal model with 5 roles and 6 core decisions is enough to get started.