Most reporting projects fail quietly because nobody ever writes down the decision the report is supposed to improve.
A dashboard can be technically correct and commercially useless at the same time. That usually happens when the team built pages, visuals, and filters before it built a decision brief.
At Data Disruption, we do not start with chart types. We start with five pieces of information that sound simple but change the entire engagement.
- The decision owner. Name the person who will act differently if this works. A board pack, an ops manager, and a commercial lead do not need the same brief.
- The moment of use. Monthly close, Monday morning operations, price review, service exception triage. Timing changes design.
- The trigger question. "What changed?", "why did margin move?", "is this an anomaly or normal noise?", "what should we do next?"
- The acceptable proof. What evidence makes the answer trustworthy enough to act on — cited transactions, tied-out measures, named business rules, or a human reviewer.
- The action window. If the answer arrives late, the decision is already gone. That sets the bar for automation.
When a client cannot answer those five things, the project usually needs diagnosis before delivery. That is often why the right first offer is a readiness sprint, not a bigger build.
When a client can answer them, the work becomes sharper immediately. The semantic model gets cleaner. The prompts get narrower. The eval set becomes possible. The sign-off path becomes obvious.
That is the practical difference between "we need a better dashboard" and "we need a better decision system."