A Good Question Is a Piece of Infrastructure

“How do we improve engagement?” is a question with excellent professional posture. It stands upright in a presentation and appears to know where the conference room is. Unfortunately, it has not explained what engagement means.

Does it mean more visits, longer reading, useful replies, purchases, or return readers? Do we want all of those? What happens if they disagree?

A question can organize attention or scatter it. The difference is not literary elegance. It is whether people can use the question to make progress together.

Give the question a job

A useful working question usually points toward a decision. “Why are people leaving?” becomes more actionable when paired with “Which part of the signup process should we change first?”

The broader question remains interesting. The narrower one tells us what information would help. We might inspect where people stop, talk to a few recent visitors, or test clearer instructions. We now have a route from curiosity to action.

This is especially useful with AI. A broad prompt can produce a broad answer full of sensible possibilities. The difficulty is that sensible possibilities are available in almost unlimited quantities. Time to test them is not.

Separate the observation from the explanation

“Readers hate long articles” contains both a claim about behavior and a theory of motive. Perhaps visitors leave because the opening is weak. Perhaps the page loads slowly. Perhaps the right readers never reach it.

Start with what you actually observed: fewer people reached the end of this article than that one. Then list competing explanations. This small separation prevents an early guess from becoming the foundation of a confident strategy.

It also makes disagreement less personal. People can disagree about explanations while sharing the observation. That is a much better starting point than dividing the room into those who understand readers and those who apparently hate evidence.

Ask what an answer would change

Before commissioning another report, ask what you would do differently depending on its result. If every possible result leads to the same action, the report may serve another purpose: reassurance, documentation, or delay.

Those purposes should be named honestly. Sometimes documentation is necessary. Sometimes reassurance is valuable. But neither should wear a fake mustache and introduce itself as research.

A decision-focused question creates a stopping point. Once the evidence is sufficient for the decision, you can act instead of continuing to collect information indefinitely.

Build a reusable question

For your next uncertain project, complete this sentence: “Given what we observed, which change should we try, for whom, over what period, and what result would make us keep it?”

That sentence is not appropriate for every kind of inquiry. Art, philosophy, and open-ended science deserve room beyond immediate utility. But for practical work it gives a group shared coordinates.

A good question is infrastructure because it supports work after the person who asked it has left the room. It makes the next observation easier to interpret. It helps the next person contribute. It gives an answer somewhere useful to land.