Article 2 of 10 · Data & Decision Intelligence · 10 min read
Why Most Dashboards Fail
Companies spend hundreds of thousands on BI platforms. Dashboards look polished. Meetings keep happening. And decisions don't improve. This isn't a technology problem — it's a philosophy problem.
The scene
It’s Monday morning. The CFO of a mid-sized distribution company opens her Tableau dashboard — 14 charts, real-time data, colour-coded KPIs. Revenue by region. Inventory levels. Return rates. Week-over-week trends. The data is accurate. The design is clean. And she picks up the phone and makes the same call she makes every Monday: “Something’s off in the northern region. I can see the numbers but I can’t figure out what to do about them.”
This is not a one-company problem. It is the norm. And the question worth asking isn’t “why isn’t the dashboard working?” — it’s “what was the dashboard actually built to do?”
A dashboard is not a decision-making tool. It's an information display. The difference sounds subtle. It isn't.
The visibility paradox
We live in the age of data abundance. Organisations have more visibility into their own operations than at any point in history. And yet the evidence consistently shows that management decision quality has not improved proportionally to investment in data technology.
The reason is counterintuitive: more visible data creates more cognitive work for the decision-maker, not less. Every chart on a dashboard is a micro-decision — what does this mean, is this normal, should I act on this? Multiply that by 40 indicators and the dashboard becomes an obstacle, not an aid.
- 74% — of employees say they feel overwhelmed by the volume of data available to them (Tableau Data Culture Report, 2023)
- 58% — of business decisions are still made based on intuition rather than data, despite BI tool adoption (McKinsey Global Survey)
- 40% — of a mid-level manager's time is spent on reporting activities that produce no actionable decisions (IDC Research)
The 5 reasons dashboards stop working
Dashboard failure is rarely dramatic. There's no single moment when it breaks. It degrades gradually — fewer people check it, meetings reference it less, someone builds a parallel Excel file, and eventually the beautiful Tableau build that took three months to deploy becomes an artefact that IT maintains and nobody reads.
The causes follow recognisable patterns, regardless of industry, company size, or platform.
- Built for IT, not for the decision-maker — The dashboard shows everything that can be measured, not everything that needs to be known. The data team delivers completeness. The end user — the regional director, the CFO, the operations manager — receives a tool they didn't design and don't know how to translate into action. Completeness and usefulness are not the same thing.
- Metrics without context — Sales grew 8% this month. Is that good or bad? Without last year's seasonal benchmark, without the original target, without the cost-of-acquisition trend — the number means nothing. A figure without a reference point is noise, not information. Yet most dashboards display raw values and call it insight.
- Too many metrics, no hierarchy — As we discussed in the first article of this series, the human brain can track 7 ± 2 units of information simultaneously. A dashboard with 30 indicators forces the user to do the prioritisation that the dashboard should have done. This is exactly the cognitive work it was supposed to eliminate. When everything is highlighted, nothing is.
- Historical data without direction — Most dashboards answer one question well: what happened? Few answer why it happened. Almost none answer what should happen next. The gap between a reporting tool and a decision support tool lives in that third dimension — the recommended action. Without it, the user is informed but not guided.
- No clear ownership per metric — Who is responsible for each indicator on the dashboard? If the answer is “the analytics team” or “everyone” — nobody is watching it with real accountability. A metric without an owner is a statistic. A metric with an owner is a KPI. The difference is organisational, not technical.
The technical design mistakes that compound the problem
Beyond strategic failures, most dashboards also carry a set of recurring design flaws that make them harder to read and act on — even for users who understand the data.
Reporting vs. decision support — a fundamental distinction
There is a foundational difference between a dashboard that informs and one that supports decisions. It is not a difference in complexity or cost. It is a difference in the question asked at the beginning of the design process.
| Dimension | Reporting Dashboard | Decision Dashboard |
|---|
| Central question | What happened? | What should I do? |
| Number of metrics | Exhaustive — as many as possible | 5–7 per role, maximum |
| Context | Absent — raw values only | Target, trend, benchmark visible |
| Alerts | None, or indiscriminate | Proactive, threshold-based |
| Ownership | Diffuse — “the data team” | Explicit — one owner per KPI |
| Refresh rate | As frequent as technically possible | Matched to decision cadence |
| Audience | General — one dashboard for all | Specific — designed per role |
The same company. Two approaches.
Back to the CFO and the 14-chart Tableau dashboard. The data doesn’t change. The underlying business doesn’t change. What changes is the design philosophy.
The automatic summary at the top of the decision dashboard reads: “Northern region is 18% below target this week. Primary driver: two high-value customers with overdue invoices (INV-1204 and INV-1231, totalling €47,000). Suggested action: accounts receivable follow-up before end of week.”
No additional data. No new technology. The same dataset, organised around a decision rather than around completeness.
The purpose of a dashboard is not to show everything you know. It is to eliminate everything that doesn't matter for today's decision.
Where to start if your dashboard isn't working
- Start from the decision, not the data. Ask the person who will use the dashboard: what do you decide regularly? Design backwards from the answer.
- Remove until it hurts. Take out every metric that doesn't have a named owner and a defined action threshold. Restore only what someone genuinely misses.
- Add context to every figure. No metric should appear without a reference point — target, prior period, or benchmark. A number alone is not information.
- Test with the real user. Not the data team. Not IT. The actual decision-maker. If they can't extract the core insight in 90 seconds, the dashboard is too complex.
Analyse your data. Get insights, not just charts.
Upload any dataset and an AI agent identifies the patterns, anomalies, and relationships in your data — with conclusions in plain business language, not technical output.
The Real Problem of Modern Companies: Too Much Data, Too Little Insight · KPI vs. Metric: The Difference That Costs Companies Millions