Most community bank boards have been left out of the AI conversation.
Not deliberately. The strategy gets framed as technical, the discussion happens at the operating level, and what arrives at the board is usually a budget request with a high-level slide deck. Directors approve. Management executes. Twelve months later, nobody is sure whether the investment is working.
This is the gap that produces the 95% AI failure rate at large enterprises. Boards approve AI budgets without the framework to evaluate whether the strategy makes sense, and twelve to eighteen months later the institution has spent significant capital with no measurable result.
The good news is that community bank boards do not need to become technical to fix this. They need to ask five specific questions. These questions are diagnostic. They reveal whether management is preparing to fall into the 95% failure pattern or to build the discipline of the 5% that succeed. They are written to be answerable by management at a board level, without slides, jargon, or vendor pitches.
In Pillar 1 of this series, The 90-Second Bank, I argued that community banks should not try to win the AI race the big banks are running. The 95% failure rate at large enterprises is the evidence.
The failure is structural. Big banks fail at AI in predictable ways for predictable reasons: treating AI as a technology procurement problem, funding it as a long bet without intermediate accountability, investing in human review instead of workflow redesign, and trying to deploy AI across the bank instead of process by process. There is a separate piece in this series on these four mistakes that goes deeper.
The board's role is to make sure the institution does not fall into the same pattern. The five questions are the diagnostic framework for that role. Each one tests for a specific element of strategic discipline that separates the 5% from the 95%.
If management can answer all five clearly and specifically, the strategy is probably sound. If management struggles on two or more, the program needs to be restructured before more capital is committed.
This is the foundational question. Every other question depends on a clear answer to this one.
What you want to hear: A specific metric tied to a specific timeline. Cycle time reduced from X days to Y hours. Staff hours per loan application reduced by Z percent. Customer abandonment rate cut in half. Error rate reduced by N basis points. The number is concrete. The deadline is specific. The measurement method is agreed in writing.
What should worry you: Vague language about "transformation," "competitive advantage," "AI capabilities," or "future-proofing." If the answer requires three minutes of context to set up, the strategy is probably not specific enough to succeed. AI strategy that cannot be reduced to a measurable operational outcome is not strategy. It is a budget request dressed as a vision.
Why this matters: Approximately 60% of firms investing in AI have no formal mechanism to track business impact. Those institutions are spending without measuring, which is the textbook entry point to the 95% failure pattern. Ready, Fire, Aim.
This question tests for accountability cadence.
What you want to hear: A measurable result within 90 days. Management can describe the self-funding operating model where each 90-day increment produces enough operational value to fund the next. The cost-to-knowledge ratio is reasonable. The institution is not committing significant capital to find out whether the first step works.
What should worry you: Eighteen months. Twelve months. "We will know after the next quarterly check-in cycle." Any answer that has the institution spending significantly before producing measurable results is the financial pattern that produces the 95% failure rate. Long timelines without intermediate accountability are the actual mechanism of failed AI investment.
Why this matters: The Economist Enterprise reported in 2026 that about 60% of firms take between 7 and 12 months to ship an AI project. By the time most of these projects produce a result, the strategy that approved them has often moved on, and the project either gets quietly defunded or absorbed into a different initiative. Board governance has to be tighter than that.
This is the question most boards never ask, which is why most boards keep funding AI programs that are not working.
What you want to hear: A specific stop condition defined in advance. If the first 90-day increment does not hit the metric defined in Question 1, the second increment does not start. Management has committed to honoring this in writing. Disciplined attrition is built into the program design.
What should worry you: "We will keep iterating." "We will adjust the approach." "We do not anticipate that scenario." Any answer that does not have a clear stop condition signals that management is treating AI as a strategic priority that cannot be wound down, which is the most expensive condition in any investment.
Why this matters: The 5% of firms capturing AI value at scale do not have better technology than the 95%. They have disciplined attrition. They stop programs that are not working before more capital is committed. The 95% keep funding failure. The board's role is to make sure the institution is in the first group, not the second.
This question tests for the distinction between AI-augmented and AI-native thinking.
What you want to hear: Workflow redesign. Role redefinition. The work itself is being rebuilt around what AI can do. Management can describe how specific processes will operate differently after the investment, with concrete before-and-after examples. The technology decisions follow the operational decisions, not the other way around.
What should worry you: A focus on AI tools, vendors, platforms, or "centers of excellence." Any answer that emphasizes what the institution is buying rather than how the work will change is the AI-augmented failure mode. AI deployed onto workflows that have not been redesigned produces marginal improvement and high human review costs. AI deployed into redesigned workflows produces operational transformation.
Why this matters: Approximately 50% of firms cite human review as a top ongoing cost of running AI systems. Only 4% cite employee upskilling. Institutions that have made the AI-augmented mistake are paying for AI tools plus armies of human reviewers checking the AI outputs, and capturing almost none of the productivity gain that justified the investment.
This question tests for strategic awareness beyond operational efficiency.
What you want to hear: Management understands the connection between AI readiness and valuation. If the institution is preparing for a possible sale in the next three to five years, management can describe how the investment shows up in the diligence picture buyers will see. If the institution is committed to staying independent, management can describe how AI readiness is the structural counterweight to PE-backed acquirers and credit unions with deeper digital infrastructure.
What should worry you: "M&A is not our focus right now." "That is a separate strategic conversation." Any answer that disconnects the AI investment from the consolidation environment in community banking signals that management is treating AI as an internal efficiency play rather than a strategic positioning move. There were 181 bank mergers in 2025. The M&A landscape is the strategic context for every community bank, whether buying or selling or fortifying against acquirers.
Why this matters: AI readiness has become a valuation driver in community bank M&A. Institutions that come to a transaction with clean data, modern processes, and quantifiable operational maturity systematically transact at higher multiples. Institutions that do not transact at a discount. The same readiness that drives valuation also drives independent competitive position.
The five questions become more powerful when the board sees the answer patterns side by side.
|
Question |
What You Want to Hear |
What Should Worry You |
|---|---|---|
|
1. Outcome and timeline |
Specific metric + specific date |
Vague language about transformation |
|
2. Cost to know it works |
Measurable result within 90 days |
"We will know in 12 to 18 months" |
|
3. Stop conditions |
Defined in advance, in writing |
"We will keep iterating" |
|
4. Operating model impact |
Workflow redesign and role changes |
Focus on tools, vendors, platforms |
|
5. M&A positioning |
Awareness of readiness premium |
"Not our focus right now" |
If two or more of management's answers fall in the far right column, the program needs to be restructured before more capital is committed. This is not a failure of management. It is a signal that the strategy was developed without the discipline required for a successful AI outcome. Restructuring early is cheaper than restructuring late, and dramatically cheaper than discovering the program did not work after 18 months.
The five questions work in three modes.
Forward the questions to management 48 hours before the AI agenda item. Management will arrive prepared to answer the questions directly, which produces a much more productive board conversation than the typical slide-deck-and-budget-request format.
If the institution already has an active AI program, the five questions can run as an annual checkpoint regardless of where the program is in its lifecycle. The answers will be more concrete or less concrete than they were last year, which is itself diagnostic information.
When a new AI investment proposal comes to the board, the five questions become the gating standard. The proposal does not advance to a vote until management can answer all five clearly.
The framework is intentionally non-technical. Directors do not need to understand AI architecture, machine learning, or vendor stacks to use it. They need to understand strategy, accountability, and operational discipline, which is what boards are designed to govern in the first place.
If you are a CEO or President, forward this article to your board ahead of the next AI strategy discussion. The framework is designed to be that kind of artifact.
If you are a Director, bring the five questions to the next board meeting where AI is on the agenda. Use them. Notice which answers come easily and which ones do not. The pattern itself is the diagnostic.
If management cannot answer two or more of the questions clearly, the program needs work before more capital is committed. That conversation is worth having before the next 90 days of investment, not after.
If management is missing on two or more of the five questions, the next step is a 30-minute conversation.