The Dashboard You Build vs. The Dashboard They Need
Here’s the gap:
What you think they want: Comprehensive analysis. All data. Rich interactions. Room to explore.
What they actually want: An answer to one specific question, fast. And if it’s wrong, context for why.
Non-technical users aren’t explorers—they’re decision-makers on deadline. They need dashboards that tell them what to pay attention to, not dashboards that let them explore if they find time.
Design Principles That Actually Work
1. One primary metric at the top
Put the number they care about most above the fold. Not buried in a chart. Big, clear, unavoidable.
- “Are we on track?” — show it first.
- Make it red or green if possible. Make it a number, not a trend line.
2. Lead with the answer, then the why
Don’t make users think.
- ✓ “Revenue: $2.3M (↑12% vs. target)”
- ✗ “Revenue by week, plotted against forecast, adjusted for seasonality”
One shows you’re on track. The other requires interpretation.
3. Remove everything non-essential
Every axis label, legend, and gridline is visual noise. Does it change the decision? If no, delete it.
4. Use colour sparingly
Resist the rainbow. Use colour to highlight what matters:
- Green for good
- Red for bad
- Grey for context
Everything else muddles the message.
5. Make comparisons obvious
“Revenue this month” means nothing without context.
- Revenue this month vs. last month
- Revenue vs. target
- Revenue vs. same month last year
Show the comparison in the same chart, not in separate sheets.
6. Write for clarity, not cleverness
Axis label: “MAU by Cohort” — fine. Axis label: “First-time user population segment by acquisition month” — now they’re confused.
If someone needs five seconds to parse your label, your dashboard already lost.
Common Mistakes
Mistake: Too many chart types
Line charts, bar charts, pie charts, scatter plots—all on one dashboard. Your brain is trying to learn a new visual language for each one.
Use one or two chart types. Bar charts and line charts cover 90% of cases.
Mistake: Interactivity for its own sake
“Let’s add filters!” — fine, but every filter adds cognitive load. Can someone without filters understand the default view?
Mistake: Forgetting to define “good”
What does success look like? Show me where the target line is. Show me where “healthy” begins and “concerning” ends.
If I have to guess whether +5% is good or bad, you’ve failed.
Mistake: Assuming business logic is obvious
“Why are we tracking new user cohorts?” — because retention matters for LTV. “So which cohort do I act on?” — the one that’s churning fastest.
Don’t make them reverse-engineer your thinking. Tell them explicitly.
The Test
Before you ship a dashboard, show it to someone unfamiliar with it and ask three things:
- “What’s the main thing this is telling you?” (Did they get it?)
- “If this number turns red, what do you do?” (Does it drive action?)
- “What would you change?” (Honest feedback)
If they struggle with #1, go back to design. If #2 has no clear answer, it’s chart for chart’s sake. If #3 has real insights, take them seriously.
The best dashboard is the one that’s used. Not the one that’s featured. Not the one with the most data. The one that answers a real question, clearly and fast.
Make that, and people will come back. Make a masterpiece nobody understands, and they’ll keep using their Excel spreadsheet from 2019.