Not Convinced by an AI Strategy Proposal? Separate What to Investigate, Decide, and Test

You have developed a strategy proposal with AI. The explanation is coherent, yet you are not convinced enough to put it into practice in your own business.

Before asking for a more persuasive explanation, put your concern into one sentence. “I do not know whether customers actually need this” and “Even if they need it, I do not want this to become our main business” require different next steps.

This article offers a practical approach for business owners whose decision has stalled around an AI-generated proposal. It does not evaluate a particular AI product's capabilities or the effects of adopting AI.

Separate three kinds of doubt

A question about evidence. Do these customers really have this problem? Are the proposed competitors companies they actually consider? These are questions to investigate. Check the original sources and, where necessary, speak to customers or the delivery team. Do not let a natural-sounding explanation fill a gap in the evidence.

A management choice. Even if demand exists, do you want to prioritize these customers? Would you reduce service to existing customers to do so? More facts alone will not settle these questions. State the commitments you want to keep, the burden you are willing to take on, and the work you will not pursue for now.

A question about delivery conditions. You agree with the direction, but staffing, deadlines, or scope remain unresolved. These require operational decisions. If capacity is insufficient, decide whether to reduce work or change the delivery setup, rather than treating better understanding of the strategy as the solution.

A concern may contain more than one element. The point of separating them is to choose an appropriate next action, not to classify everything neatly.

Suppose the proposal is to compete on short lead times

This is a hypothetical illustration, not a client case.

If the concern is “I doubt customers will pay more for speed,” investigate which delays matter to the intended customers, how they deal with them now, and their budgets and purchasing conditions. Positive feedback alone does not establish willingness to pay.

If the concern is “Prioritizing urgent orders would break our commitments to existing customers,” examine order limits and exceptions before commissioning more research. You may choose not to accept the work even if demand exists.

If the team can deliver quickly under particular specifications, a limited test becomes an option. You do not need to promise short lead times on every order. Connect the proposed distinction to whom you serve, what you deliver, and under which conditions.

Write down what would change your mind before testing

Where customer responses or actual delivery are needed to investigate a premise, limit the scope of the test. Record five things:

• The premise you doubt.
• What you will observe to examine it.
• What result would make you revise the proposal.
• Limits on time, spending, and delivery scope.
• Who will make the next decision, and when.

Where disagreement concerns evidence, ask what findings could change the person's view. If the disagreement concerns commitments or the direction of the business, discuss management priorities rather than expecting test results alone to resolve it. But do not design the test simply to eliminate opposition. Finding that the proposal does not fit also provides material for a decision.

Success in a small trial does not establish that the result will hold at a larger scale. A lack of response may also reflect the people approached, timing, or conditions. If the result leaves the question unresolved, do not force a success-or-failure verdict. Weigh the remaining uncertainty against the burden of another test.

Do not keep adding explanations until everyone agrees

A KPMG seminar report distinguishes AI's support for information search, comparison, structuring, and hypothesis development from final human judgment.[1] This is the firm's practical commentary, not a study demonstrating the effectiveness of the process proposed here.

Whether you adopt or reject the proposal, make the choice your company is prepared to take responsibility for. Being ready to decide need not mean unanimous agreement. Share the basis for the decision, the doubts that remain, and the circumstances that would trigger a review.

The immediate goal is not a completed strategy document. Choose one stalled proposal and decide whether to investigate, make a management choice, or test under limited conditions. If funds or delivery capacity are insufficient, holding the proposal without a trial is also an option.

For deciding whether AI-generated strengths can be used in sales copy, see checking the evidence behind the claims. To incorporate decisions into an ongoing meeting process, see using brand strategy in management meetings.

If your leadership team has not agreed on the premises behind its target customers and value proposition, see HUV's support for defining value, customers, and reasons to choose your business. The connection is the management choice about whom to serve and what to promise, rather than instruction in using AI itself.

[1] KPMG, “AIで進化する戦略策定・実行管理” (AI-enabled strategy formulation and execution management), published October 13, 2026; a report on an August 2026 seminar. Source in Japanese. https://kpmg.com/jp/ja/insights/2026/09/ai-enhancement-of-strategy-formulation-and-execution.html Accessed October 23, 2026.

Service information: HUV's official services page, accessed the same day.

Author: Daisuke Irikura, Representative Director, HUV DESIGN OFFICE