wunder beta📘 How do you design an honest dashboard?
A dashboard exists to drive a decision or an action, not to look impressive. Stephen Few, in Information Dashboard Design, defines a dashboard
What you’ll learn
- What a Dashboard Is ForExplain the purpose of a dashboard and distinguish it from reports and exploratory analysis.A dashboard is the most important decision-relevant information consolidated on one screen for at-a-glance monitoring, as Stephen Few defines it. Its job is to support decisions and actions, not to decorate. It differs from a detailed report and from open-ended analysis. Build it question-first, naming the decisions and the people who own them. Because attention is the scarce resource, defend every element by the decision it enables and cut the rest.
- Choosing Metrics That MatterSelect decision-relevant metrics by distinguishing north-star, supporting, leading, lagging, and vanity metrics.Choose one north-star metric that captures core value, then supporting metrics that explain its movement. Distinguish leading indicators (early predictive signals) from lagging ones (confirmed outcomes), and avoid vanity metrics—impressive numbers that change no decision—per Eric Ries's actionable-vs-vanity distinction. Prefer rates and ratios over raw totals so performance stays comparable as you scale, and pair every primary metric with a target or prior-period comparison so it is interpretable.
- Designing for the AudienceDesign audience-appropriate dashboard views that match granularity and layout to the viewer's decisions.Different audiences need different views of the same data. Executives need summarized, outcome-focused scorecards with targets and exception flags, scanned in under a minute; operators need granular, fresh breakdowns to act on today's anomalies. Layout should follow priority—most important metric in the high-attention upper-left, related metrics grouped. Use progressive disclosure to layer a clean summary over on-demand detail so one product can serve multiple audiences.
- Data-Viz Best Practices: Charts, Color & Honest AxesApply Tufte's data-ink principles, choose the right chart and color for a question, and avoid misleading axes.Edward Tufte's The Visual Display of Quantitative Information (1983) introduced the data-ink ratio and the term chartjunk: maximize the ink that encodes data and remove decoration. Match chart type to the question—lines for trends, sorted bars for category comparison, histograms for distribution, scatter for relationship. Cleveland and McGill's 1984 study shows position and length are decoded more accurately than angle and area, which is why bars beat pies. Use preattentive attributes deliberately: reserve saturated color for emphasis, mute the rest, choose colorblind-safe palettes (Okabe-Ito/Wong, ColorBrewer) and never rely on color alone. Keep axes honest—start bar baselines at zero, since bars encode value by length.
- Data Hygiene & a Single Source of TruthEstablish trustworthy metric definitions, a single source of truth, and data-quality practices for a dashboard.A dashboard is only as trustworthy as its data. Define every metric precisely—definition, formula, source, window, owner—so viewers do not argue over what a number means. Draw every dashboard from one governed single source of truth so the same metric reads identically everywhere. Catch data-quality issues upstream and surface freshness on the screen. Keep definitions stable and comparable over time, annotating changes, because trust is fragile: one wrong number can discredit the whole dashboard.
- Guided Project: Dashboard Spec & Mockup (Peer Critique)Produce a dashboard spec and mockup, then evaluate it against a peer-critique checklist.Walk through building a dashboard spec end to end: frame the brief (audience, decisions, time horizon, key questions); choose one north star plus supporting metrics with leading/lagging flags and comparisons; map metrics to chart types and sketch a priority-driven layout with purposeful, colorblind-safe color; and document data hygiene—single source of truth, definitions, refresh cadence, freshness. Finish by applying a structured peer-critique checklist and revising the spec from the feedback received.
Questions this course answers
According to Stephen Few's definition, what most distinguishes a dashboard from a detailed report?
Few defines a dashboard as the most important information consolidated on a single screen to be monitored at a glance. Interactivity (a) is not the defining trait; a dashboard typically holds less, not more, data than a report (c); and authorship (d) is irrelevant to the definition.
A team wants to add a chart 'because we already collect that data.' What is the best test for whether it belongs on the dashboard?
A dashboard exists to drive decisions, so the test is whether an element informs an action. Update frequency (a) and source reliability (d) matter for quality but not relevance, and visual balance (b) is aesthetics, not purpose.
Why does adding 'just in case' metrics tend to hurt a dashboard?
Attention, not data, is the scarce resource; extra metrics raise cognitive load and dilute the important signals. Server load (a) and export (d) are not the core issue, and extra metrics do not necessarily cause data errors (c).
Which pairing correctly distinguishes a leading from a lagging indicator?
Leading indicators are early signals (trial signups) believed to predict later outcomes; lagging indicators (revenue) confirm results after they occur. Option (a) reverses them, (c) misclassifies signups, and (d) confuses the concept with refresh rate.
A startup proudly tracks 'cumulative total registered users,' which only ever rises. Why is this likely a vanity metric?
Per Eric Ries's contrast of vanity vs. actionable metrics, a cumulative total that can only rise masks the rate that matters and seldom guides action. It is easy to calculate (a), can fit on a screen (b), and being leading is not what makes it vanity (d).
Traffic tripled this quarter but conversion rate fell. Which dashboard design would have surfaced the real problem soonest?
Rates normalize for scale and a comparison makes the value interpretable, so the falling conversion rate would stand out. Total conversions (a), cumulative visitors (c), and uncontextualized revenue (d) all grow with traffic and hide the decline.
Grounded in trusted sources
- Stephen Few, Information Dashboard Design
- Edward Tufte, The Visual Display of Quantitative Information — chart integrity
- Google Data Studio / Looker Studio documentation patterns
- Cole Nussbaumer Knaflic, Storytelling with Data
- Kimball BI / dimensional reporting primers on single source of truth
Every Wunder lesson is built from real, reputable sources — never invented.
Related courses
Wunder is a personalized learn-anything platform — tell it any topic and it builds a beautiful, fact-checked course in minutes, with narration, a knowledge check, and a college-style University track.
© 2026 Wunder Learning LLC · Terms & Privacy