How to Get Board Buy-In for AI Investment

Why Boards Say No to AI Proposals

In short

Boards reject AI proposals for four recurring reasons: ROI uncertainty, unmanaged risk, an unclear talent plan, and weak governance. None of these are about the technology. They are about whether the proposal is investable. Pre-empt all four and the conversation shifts from defending to deciding.

Most AI pitches arrive in the boardroom optimized for excitement, not for decision-making. That is why they fail.

BCG and MIT Sloan's 2024 AI Survey found roughly 26% of GenAI investments deliver measurable business value. Boards have read that statistic. They are now allergic to proposals that sound like the other 74%.

The four objections that kill most AI proposals:

1. ROI uncertainty. Projections are vague, baselines are missing, value is asserted rather than evidenced.
2. Unmanaged risk. No discussion of model risk, data risk, IP risk, or regulatory exposure under the EU AI Act.
3. Unclear talent plan. Who runs this — and what happens when they leave? No named executive sponsor and no team plan.
4. Weak governance. No policy, no escalation path, no monitoring. The board sees a project, not a capability.

McKinsey's 2024/2025 State of AI research reports a consistent qualitative pattern: companies with a named executive sponsor outperform those without one. Boards know this. Show up with a sponsor.

The 5-Slide Board AI Briefing (Alice Labs Framework)

In short

The Alice Labs Board AI Briefing is a five-slide structure refined across 100+ Nordic enterprise engagements: Why now, Where we'll win, How we'll execute, What it costs, How we'll measure. Five slides force clarity. Boards approve clarity.

Every Alice Labs Board AI Briefing follows the same five-slide spine. The structure is deliberately constrained. Anything longer signals to a board that the case has not been decided yet.

Slide 1 — Why now. Two market facts the board already believes (AI adoption is mainstream; the value gap is real), one company-specific trigger (a strategic priority, a competitor move, a margin pressure point), and one closing line: "Doing nothing has a cost."

Slide 2 — Where we'll win. The single use case you are recommending. Quantified business problem, measured baseline, named beneficiary inside the business. Not three options. One.

Slide 3 — How we'll execute. The named executive sponsor, the delivery partner (internal or external), the build-vs-buy posture, the major architectural choice, and the 12-week milestone plan.

Slide 4 — What it costs. Staged budget. A 12-week pilot envelope, a scale-up envelope contingent on the kill criterion, and the operating cost at steady state. Capex, opex, and people split out.

Slide 5 — How we'll measure. One primary success metric, one secondary metric, the kill criterion, and the review cadence with the board. Make it easy for the board to track without re-asking.

If your case needs more than five slides, the case has not been decided yet. Decide it before you walk in.

Building the ROI Case With Comparable Proof

In short

Boards trust comparable proof more than projections. Anchor the ROI case on three or four real implementations at similar scale, then translate to your context. Alice Labs production cases provide the anchor: Ljusgårda 2.5M SEK/year, a Nordic media client at +2,092% growth, public sector deployments saving 6,400–8,000 hours per year.

Most boards have seen a hundred AI ROI spreadsheets. Almost none of them came true. The credibility cost of an over-promised projection is severe — and lingers.

The structure that works is comparable proof first, projection second. Show two or three real implementations at similar scale, in adjacent industries, with measured outcomes. Then translate those outcomes to your context with a small set of clearly stated assumptions.

Anchor cases we use in Nordic boardrooms:

• Ljusgårda — 2.5M SEK/year in measured operational savings from an AI implementation Alice Labs delivered, with the savings audited against pre-implementation baselines.
• Nordic media client — +2,092% growth in the targeted business metric over the measured period, driven by an AI-enabled workflow change.
• Public sector deployment — 6,400 to 8,000 hours per year of capacity freed across the affected function, with the hours quantified by time-and-motion measurement before and after.

These are not projections. They are audited outcomes. Boards relax when the case is anchored on what happened — not on what could happen.

The Alice Labs Implementation Index 2026 reports a 96% production rate across our enterprise engagements. That is the comparable proof: AI investments structured the way the briefing structures them reach production at materially higher rates than the industry average.

Risk and Governance — Answer Before You Are Asked

In short

Under the EU AI Act (Regulation 2024/1689), board-level governance is no longer optional for high-risk AI. Address risk on slide three of the briefing: classification, named risk owner, monitoring approach, and the link to existing enterprise risk frameworks. RAND's 2024 study identifies a missing business owner as the leading root cause of failure — solve that on slide one.

The fastest way to lose a board is to skip risk. The second fastest is to address risk with generalities. Boards want specifics.

The EU AI Act, Regulation (EU) 2024/1689, places explicit governance obligations on organizations deploying high-risk AI systems — including risk management, data governance, technical documentation, human oversight, accuracy and robustness, and post-market monitoring. For many board members, this is now a fiduciary topic, not an IT topic.

The four risk answers every board expects:

1. EU AI Act classification. Is the use case prohibited, high-risk, limited-risk, or minimal-risk? State the answer and the basis.
2. Named risk owner. One executive accountable for AI risk on this initiative — typically the Chief Risk Officer or General Counsel — with a documented escalation path.
3. Monitoring approach. What is monitored in production, by whom, at what cadence, and what triggers escalation to the board?
4. Failure mode plan. What happens if the model misbehaves, the vendor fails, or a regulatory change requires rework?

RAND's 2024 study (RR-A2680-1, based on 65 practitioner interviews) identifies a missing business owner as the top root cause of AI project failure. Boards intuit this — even without reading the study. Naming the owner on slide one is not optional.

Competitive Positioning — Frame the 26% Gap as Opportunity

In short

The BCG/MIT 2024 finding that ~26% of GenAI investments deliver measurable value is not bad news — it is the strategic argument. Two-thirds of competitors are deploying capital that will not return. A board that approves a rigorously structured initiative buys disproportionate competitive advantage at a moment of broad industry mis-execution.

Boards understand timing. The most powerful framing for the value gap is not as a warning, but as a window.

BCG and MIT Sloan's 2024 research found that roughly 26% of GenAI investments deliver measurable business value. The corollary: a substantial share of competitor AI spend is generating noise rather than outcomes. McKinsey's 2024/2025 State of AI reports 72% of organizations have adopted AI. The combination is the strategic picture.

What this means for the boardroom argument:

• Most AI spend is mis-executed. The bar for outperformance is lower than the headline adoption rate suggests.
• Structural advantage compounds. Companies that get the second use case right learn faster than those still failing the first.
• The window is finite. Stanford HAI's AI Index tracks investment levels that have grown sharply. The opportunity to be in the disciplined minority narrows as the field matures.

The boardroom translation: "Most of our competitors will spend on AI and not see returns. A small minority will, and they will compound the advantage. Our case is built to put us in that minority — here is the evidence why."

The 30-Day Path From Board Pitch to Approved Budget

In short

Boards do not approve in the meeting — they approve in the conversations around the meeting. The 30-day path is: pre-wire the briefing one-on-one, formalize the proposal in the board pack, present and answer, and close with a documented next-meeting decision. Skipping any step adds 60–90 days.

The most common reason a well-structured AI proposal does not get approved is not the content — it is the process. A 30-day path, run deliberately, gets to yes.

Days 1–7 — Pre-wire individually. Meet one-on-one with the Chair, the Audit Committee Chair, and any director with a strong AI or technology background. Share a one-page summary. Capture every objection. The pre-wire is where the proposal actually gets approved.

Days 8–14 — Refine and formalize. Rewrite the briefing to pre-empt every objection captured in pre-wire. Add the risk owner. Confirm the executive sponsor will present alongside you. Submit into the board pack on schedule.

Days 15–25 — Late refinements. Most boards request written follow-ups between pack submission and meeting. Treat these as part of the briefing — not as interruptions. Answer in writing, in the same five-slide structure.

Day 26 — The meeting. Present the five slides in 12 minutes. Spend 18 minutes on questions. Do not extend. Leave with a documented decision — approval, a specific information request with a deadline, or a defined next-meeting agenda. Avoid "we will come back to it."

Days 27–30 — Documented close. Send the minute-confirmation note, the named sponsor, the kill criterion, and the first 90-day milestone. The faster this is documented, the faster delivery begins.

What to Do Next

In short

If you are preparing for a board AI conversation, the next step is structured. Read the Enterprise AI Strategy framework to align the strategic logic. Run a readiness assessment so the slide-three execution plan is credible. Then book the pre-wire with the Chair — that is where approval is won.

The next moves, depending on where you are:

• Strategy not yet defined. Start with the Enterprise AI Strategy Framework. The five-slide briefing rests on a defined strategic posture.
• Strategy defined, execution unclear. Run an AI Readiness Assessment so slide three of the briefing is credible.
• Build-vs-buy unresolved. Use the Build vs Buy AI framework to lock the architectural posture before the board sees the briefing.
• Compliance exposure unclear. Walk the EU AI Act Compliance Checklist before slide-five risk answers are drafted.