The Five-Second Rule: Designing Dashboards People Actually Read
If a stakeholder can't grasp the state of the business in five seconds, the dashboard has failed. Here's the design system I use to get there.
By Tillmann Kühn · 4 August 2026
Most dashboards fail not because the data is wrong, but because nobody can read them fast enough to act. A dashboard is not a report — it is a decision surface. If it takes more than five seconds to answer "are we okay or not?", it will quietly stop being opened.
Here is the design system I apply to every build.
Start with the decision, not the data
Before I place a single chart, I ask one question: what decision does this screen support? A dashboard that supports "should we shift budget from paid to organic this week?" looks completely different from one that supports "is the sales team on pace for the quarter?"
When you design around the decision, most metrics disqualify themselves. That is the point.
Establish a visual hierarchy in three tiers
- Tier 1 — the verdict. One to three numbers that answer the headline question. Big, high-contrast, top-left. This is what the five-second glance lands on.
- Tier 2 — the drivers. The trends and breakdowns that explain why tier 1 is moving. Mid-sized charts, scannable.
- Tier 3 — the detail. Tables and drill-downs for the person who wants to investigate. Below the fold, deliberately.
If everything is emphasised, nothing is. Restraint is the design.
Colour carries meaning, not decoration
Reserve your accent colour for a single job: signalling attention. When green means "good" everywhere except that one chart where it just means "series two", you have trained people to distrust the colour. I use a neutral base, one accent for the primary metric, and semantic colours (up/down, on/off pace) applied consistently across the whole system.
Kill the vanity charts
Every chart earns its place by changing a decision. Impressions going up feels good; it rarely changes what you do on Monday. Pipeline coverage dropping below 3x changes it immediately. Ruthlessly cut anything that is interesting but not actionable.
Design for the smallest screen it will be seen on
The CEO checks it on a phone between meetings. If the five-second verdict does not survive a 390px-wide screen, redesign it until it does. Mobile-first is not a nicety here — it is where the highest-value glance actually happens.
A great dashboard is an act of editing. The craft is in what you leave out, so the one thing that matters is impossible to miss.
Want a dashboard like the ones I write about?
Book a free discovery call and I'll map your KPIs to a tailor-made build.
Book a call