Every BI project I’ve worked on starts with a version of the same conversation: “We want a dashboard.” The question I always ask first is: what decision do you need to make faster?
If the answer is vague, the dashboard will be too.
Start with decisions, not metrics
The instinct is to gather every available metric and display it. This produces dashboards that are technically comprehensive and practically useless. Leaders scan past them because nothing guides attention.
The better approach: map the three to five decisions that happen regularly in the business — pricing changes, inventory reorders, campaign spend allocation — and design backward from those. Every metric earns its place by supporting a named decision.
The layout hierarchy that works
- KPIs above the fold — one number per key decision, with a trend arrow and period-on-period change
- Supporting charts in the second layer — answer “why is this KPI moving?”
- Drill-down tables at the bottom — for the analyst who needs to investigate
This matches the attention flow of an executive: headline → context → detail on demand.
Automation is not optional
A dashboard fed by manual CSV exports will stop being used within weeks. The pipeline that keeps data fresh is as important as the visualization layer. For the dashboards I built at Rhyton Home, N8N automation kept data synchronized without any manual intervention — which was the reason they stayed in active use.
Key takeaway
Design decisions, not metrics. Automate the data pipeline from day one. Keep the primary view to one screen without scrolling.