“What is our AI strategy?” sounds like a strategic question, and it is not. It names a technology rather than an outcome, so nothing in it tells you when you are finished. Compare it with the questions your company answers well: how do we win the mid-market accounts, how do we keep our best installers, how do we stop losing money on rush jobs. Each of those has a shape. You can tell a good answer from a bad one.

A technology question has no shape, so the answers arrive as activity. A tool gets bought. A committee meets. Someone builds a chatbot nobody asked for. Twelve months later the honest summary is that the company spent money and learned a little, which is not the same as having improved anything.

The reframe is not a rhetorical trick. It converts an open-ended technology question into a finite business one, and a finite question can be answered, priced, tested, and abandoned if the answer turns out to be no.

Executives arrive at this subject under one of three pressures, and they call for different responses. Diagnosing which one you are under saves a great deal of wasted motion.

Board or investor pressure is a question about credibility. The board is not usually asking for a system; it is asking whether the person running the company is paying attention. That is answered with a clear point of view and a small, measured programme — not with a large commitment made quickly to look decisive.

Competitive pressure is a question about position. It deserves investigation rather than imitation: what a competitor announces and what a competitor has working are frequently different things, and copying an announcement is how companies buy other people's mistakes.

Staff pull is the most useful pressure and the most commonly mishandled. When people are already using tools without approval, you have free evidence about where the work hurts. Treating that as a policy violation to be shut down destroys the signal. Treating it as a survey gives you a shortlist.

Most improvement exercises begin with a process map, and process maps lead naturally to automation — doing the same steps faster. That is a real benefit but a small one, and in a company of your size the steps are usually not the expensive part. The expensive part is judgement applied with incomplete information.

So inventory decisions instead. Walk one week of your own calendar and one week of the calendar of whoever runs operations, and write down every point where somebody chose between options. Then mark each one with what the chooser knew at the time.

The top-right quadrant is where a company of your size finds its money. A quoting decision made forty times a week with incomplete cost history is worth more attention than a once-a-year strategic choice, however important the strategic choice feels in the moment.

Once you have a ranked list of decisions, a short test removes most of them. It is deliberately severe, because the cost of a well-run programme aimed at the wrong target is a year.

One — would better information change the choice? Some decisions are already correct and merely unpleasant. If your team knows exactly which customers are unprofitable and keeps serving them because you have not decided to stop, that is a management problem wearing an information costume, and no system will resolve it.

Two — does the information exist somewhere? Not tidily, not in a database, not labelled — just in existence. Quotes in an inbox, notes in a job file, a decade of invoices. If the knowledge lives only in one person's head and has never been written down, that is a different and slower project.

Three — would you actually act on it? The most common failure in small companies is not a bad model. It is a good answer that arrives after the decision has already been made, or that nobody has the authority to act on.

You will often need to say something before you have anything working. A position is enough, and a position is stronger than a hastily announced project you may have to unwind.

A defensible position has four parts: where you believe the value is for a company of your size, what you are deliberately not doing and why, what you are testing this quarter, and what evidence would make you spend more or stop. That last part is what distinguishes a considered position from an enthusiastic one, and boards notice the difference.

It is also honest. You do not yet know whether this will pay in your company, and saying so while showing how you intend to find out is a better answer than confidence you have not earned.1