From reporting factory to decision engine

Many data teams have quietly become factories. Requests come in, reports go out, and the volume is treated as a sign of health. But ask how many of those reports actually changed a decision, and the room goes quiet. Output is not the same as impact.

The trap of the request queue

A reporting factory optimises for throughput. Whoever asks loudest gets served, the backlog grows, and the team's success is measured in dashboards delivered rather than decisions improved. It feels productive and creates almost no leverage, because nobody upstream ever asked whether the report was the right thing to build.

Start from the decision, work backwards

A decision engine inverts the flow. It begins with the decisions the organisation needs to make, then asks what evidence each one requires, and builds only that. The unit of work is not a report; it is a better decision. Everything that does not serve one is a candidate for deletion.

• List the recurring decisions the business actually makes.
• For each, define what evidence would change the answer.
• Build the smallest thing that delivers that evidence reliably.
• Retire reports that no decision depends on.

What changes

The team shrinks its surface area and grows its influence. Fewer artefacts, each tied to a real decision, each with an owner who acts on it. Analysts move from order-takers to partners in the decision itself — which is both more valuable to the business and considerably more satisfying to do.

The shift is not primarily technical. It is a decision about what the data function is for. Choose “decisions” over “requests” and the factory becomes an engine.