Stop buying dashboards. Start buying decisions.
Most BI initiatives fail before the first chart is drawn, because they optimize for visibility instead of decisions. Here's the question that fixes it.
Walk into any scale-up and you'll find the same graveyard: forty dashboards, four of them opened last month, one of them trusted. The company didn't fail at building dashboards. It succeeded at building the wrong thing.
The visibility trap
Most analytics projects start with a reasonable-sounding goal: "we need visibility into X." The problem is that visibility is not an outcome. Nobody's bonus depends on having seen a chart. The chart only matters if it changes what someone does on Monday morning.
When you optimize for visibility, you get breadth: every metric anyone ever asked about, arranged in tabs. When you optimize for decisions, you get depth: the three numbers that drive next quarter's plan, defined precisely, refreshed reliably, and owned by someone.
The question that fixes it
Before building any dashboard, chart, or metric, ask:
"What decision will change based on this number, and who makes it?"
If there's no answer, don't build it. If the answer is vague ("leadership will keep an eye on it"), park it. If the answer is specific ("the ops lead reorders stock every Tuesday based on this forecast"), you've found a real analytics product. Build that one first.
What this looks like in practice
- Start from the decision calendar, not the data model. List the recurring decisions in the business (weekly ordering, monthly pricing reviews, quarterly planning) and work backwards to the data each one needs.
- One owner per metric. A metric without an owner drifts. Definitions fork, trust erodes, and six months later two VPs are arguing about whose "active users" is right.
- Certify a small core. Mark a handful of metrics as certified (governed definitions, tested pipelines, guaranteed freshness) and let everything else be explicitly exploratory.
- Measure adoption ruthlessly. If a dashboard hasn't been opened in 30 days, archive it. The graveyard sends a worse signal than the empty page.
The payoff
Teams that make this shift ship fewer analytics assets and get more value from them. The stack gets cheaper, the arguments get shorter, and (the real prize) data starts showing up inside decisions instead of alongside them.
That's the difference between a reporting function and an analytics capability. Only one of them compounds.