Strategy

What Questions Should the Board Ask About AI Investment?

AI investment has long since left the technology department’s agenda; it now sits on the board table as a strategic topic in its own right. Yet many directors feel they lack the footing to ask the right questions about it. You do not need to know the technical details, but asking the right questions is a core responsibility of the board.

In this article we cover 10 critical questions the board should ask about AI investments, an investment evaluation framework, and the risk management approach. Our AI consulting services include giving boards an independent perspective on exactly these matters.

The Board’s Role in AI

The board’s job on AI is not to make technical decisions but to set the strategic direction and keep an eye on risk. As with cybersecurity or financial risk, AI belongs on the board’s standing agenda.

The Board’s Three Fundamental Responsibilities

Strategic Direction: Making sure AI investments line up with the company’s overall strategy. Pressing on exactly how a given AI project will strengthen competitive advantage. Approving resource allocation and ranking the priorities.

Risk Oversight: Weighing the risks that AI use creates—ethical, legal, reputational, and operational. Setting the risk appetite and asking whether the existing risk management mechanisms are genuinely up to the job.

Performance Tracking: Watching whether AI investments actually produce the value that was promised. Reading the progress reports and, when the numbers call for it, making the decision to change course.

Board AI Literacy

Directors are not expected to know deep learning algorithms or data science techniques. What they are expected to do is grasp a handful of concepts at a working level:

• What AI can and cannot do
• The role data quality plays in AI projects
• Where AI projects part ways with traditional IT projects
• The broad shape of the ethical and regulatory framework
• AI trends in the sector and where the competition stands

10 Critical Questions to Ask

Question 1: What Concrete Business Problem Does This Investment Solve?

This question keeps the project tied to business value from beginning to end. Lines like “we will innovate with AI” or “we will accelerate our digital transformation” sound good but say nothing. The board needs to see clearly which business process the project will improve, by how much, and what that is worth in money.

Example of a satisfactory answer: “We aim to reduce our excess inventory costs by 2 million TL annually by increasing our demand forecasting accuracy from 70% to 85%.”

Example of an unsatisfactory answer: “We will optimize our supply chain with AI.”

Question 2: What is the Expected Return and How Will It Be Measured?

Every investment should carry an expectation of return. With AI investments, it helps to read that return across three layers:

• Layer 1 — Direct Cost Savings: Labor reduction, lower error costs, shorter cycle times. This is the easiest layer to measure.
• Layer 2 — Revenue Impact: Better customer experience, new product opportunities, market expansion. Harder to measure, and it takes time to show.
• Layer 3 — Strategic Value: Competitive advantage, organizational learning, risk reduction. Long-term, and the hardest of all to pin down.

The board should ask which layer the project is aiming at and how, in practice, the value will be measured.

Question 3: Is Our Data Infrastructure Sufficient to Support This Project?

AI projects are built on data. Without data—or with poor-quality data—even the best algorithm produces no business value. The board needs to understand the current state of the data infrastructure and what investment it will take.

Sub-questions:

• Is the data the project needs actually available, and in digital form?
• Is the data quality (consistency, completeness, accuracy) good enough?
• Does the data infrastructure need further investment, and if so, how much?
• Is data governance (access control, privacy, retention) sufficient?

Question 4: Do We Have the Organizational Competence, or How Will It Be Acquired?

These projects pull together several different skill sets—data scientists, data engineers, business analysts, project managers. Will those come from inside the company or from outside? If from outside, how does the knowledge transfer back to us?

The board should also put the long-term competence strategy on the table: will we stay dependent on outside parties for every project, or will we build our own capacity?

Question 5: What Are the Risks and How Will They Be Managed?

AI projects carry four main risk categories:

• Technical Risk: The model failing to deliver the expected performance, data quality issues, integration difficulties
• Operational Risk: Disruptions when integrating into the business process, user resistance, maintenance difficulties
• Ethical and Legal Risk: Biased model outputs, data privacy breaches, regulatory non-compliance
• Reputational Risk: A flawed or unethical AI decision spilling into public view

Each category needs both a mitigation strategy and a fallback plan defined up front.

Question 6: What Is Our AI Ethical Framework and Governance Model?

Using AI carries ethical and social responsibility dimensions. Are the model’s outputs fair? Do users know when they are dealing with an AI decision? How will the model be audited? These answers cannot be left to chance—they have to be structured within an ethical framework and a governance model.

Topics the governance model should cover:

• Model development standards and processes
• Model approval and deployment procedures
• Continuous monitoring and performance evaluation
• Definitions of responsibility and accountability
• Ethical principles and the red lines you will not cross

Question 7: How Does It Affect Our Competitive Position?

An AI investment cannot be judged in a vacuum. What are our competitors doing? How fast is adoption moving in our sector? What is the opportunity cost of not making this investment? The board has to read the competitive context and weigh that “not doing it” carries its own risk.

That said, this question must not curdle into a reflex of “everyone is doing it, so we must too.” The competitive read has to stay cool-headed and measured.

Question 8: What Is the Total Investment Amount and How Is It Phased?

The cost structure of AI projects does not look like that of traditional IT projects. Beyond the initial investment come the recurring items: data infrastructure, model maintenance, retraining, monitoring tools, cloud resources.

The board needs to see the total cost of ownership (TCO) and how that cost spreads over time. For managing risk, a phasing strategy—start small, prove it works, then scale—is almost always the better route.

Question 9: What Is Our Exit Strategy in Case of Failure?

Every investment can fail. The board should settle in advance under what conditions the project gets stopped, what the fallback plan is, and how much sunk cost is at stake. The “this project continues no matter what” stance is the most dangerous one you can take.

Project gates need to be defined, with a structure that forces a go/no-go decision at each one.

Question 10: How Does Our AI Strategy Integrate with Our Overall Business Strategy?

The last question, and maybe the most important, tests how well AI initiatives fit the company’s overall strategic direction. Does the project support the growth strategy? Does it serve the cost leadership goal? Does it line up with the customer experience strategy?

Rather than a standalone “AI strategy,” AI use woven into the business strategy produces far more durable results.

Investment Evaluation Framework

Once you put the answers to the questions above into a structured framework, the board’s decision-making gets noticeably easier. The framework below can be used to evaluate AI investment proposals.

Evaluation Dimensions

Gate Model

Manage AI investment not as one big decision but as a process that moves through phased gates:

1. Gate 1 — Feasibility Approval: Is the business problem clear, is the data ready, are the resources there? Projects that fail this gate never get started.
2. Gate 2 — Pilot Evaluation: Do the PoC/pilot results meet expectations? Is production feasibility proven? Projects that fail this gate are stopped or redesigned.
3. Gate 3 — Production Approval: Is the production environment ready, is the change management plan complete, is the monitoring mechanism in place? Projects that clear this gate go into production.
4. Gate 4 — Scaling Decision: Are the production results satisfactory, is scaling feasible? Projects that clear this gate get expanded.

Risk Management

With AI investments, risk needs to be managed at two levels: project-based and portfolio-based.

Project-Based Risk Management

Each AI project should have its own risk assessment. Keep a risk register, and for every risk write down the probability, the impact, and the mitigation strategy.

Common project risks and how to mitigate them:

• Data quality risk: Assess data quality at the start of the project and define a minimum acceptable threshold
• Model performance risk: Set minimum acceptance criteria and compare against a baseline model
• Integration risk: Test compatibility with existing systems early, and start in shadow mode
• User acceptance risk: Bring end-users in early and prepare a training plan
• Budget overrun risk: Apply a phasing strategy and check the budget at each phase

Portfolio-Based Risk Management

In organizations running several AI projects at once, the risk spread across the whole portfolio also has to be managed. When every project sits at the same risk profile—all high-risk, or all low-value—that points to an imbalanced portfolio.

A balanced AI project portfolio looks roughly like this:

• 50-60% low-risk, proven use cases (quick wins)
• 30-40% medium-risk, medium-to-high value projects (growth projects)
• 10-20% high-risk, high-potential projects (discovery projects)

Regulatory and Compliance Risks

AI regulation is moving fast. The board should track regulatory developments in the markets where the company operates and settle on a compliance strategy. Getting ahead of it costs far less than scrambling to comply after the fact.

Providing boards with independent assessments on these matters falls within the scope of our AI project management services.

Conclusion

The board’s biggest mistake on AI is handing the topic off entirely to the technical team. AI is no longer a technology decision; it is a strategic business decision. Asking the right questions, making the right investments, and managing the risk sit with the board.

The 10 critical questions we shared here give boards a structured way to interrogate a proposal. When these questions cannot be answered satisfactorily, that is a clear signal the project needs more preparation.

And it is worth remembering that the decision not to invest in AI can be a deliberate strategy too. What matters is that the decision is made on good information, with clear eyes on the risk, and within a strategic frame.

“The board’s role on AI is not to understand the technical details but to ask the right questions and set the strategic direction.”