Why do most business dashboards fail?
Most dashboards fail not because the data is wrong but because nobody can tell, in a few seconds, whether things are good or bad and what to do next. They are usually built by listing every available metric and arranging them in a grid, so the owner, the sales manager and the accountant all get the same crowded screen.
The symptoms are familiar: people export to Excel to “see it properly”, managers still ask for the numbers on WhatsApp every morning, and the dashboard that took weeks to build is opened once a month. The fix is a design problem, not a data problem.
Who is the dashboard for, and what will they decide?
Before choosing a single chart, write down the audience and the decisions the dashboard should support. Different roles need different views of the same data:
- Owner or director: Are we on track this month? Where is cash tied up? Which branch or product needs attention?
- Sales manager: Which leads are stuck? Who is behind target? Which customers have not reordered?
- Accounts: What is overdue, from whom, and for how long? Is GST liability for the period in line with sales?
- Store or branch manager: How are today’s sales against the same day last week? What stock is running low?
If one dashboard tries to answer all of these, it answers none of them well. Build one focused view per role, and let people move between them. Access control often follows naturally from this split, since a branch manager rarely needs to see company-wide margins or bank balances.
How many metrics should a dashboard show?
There is no magic number, but a useful working limit is five to seven headline figures on the first screen, with detail available one click away. Short-term memory is limited, and every extra tile competes for attention with the ones that matter.
Arrange the screen as a hierarchy. The top row holds the few key numbers in large type. The middle holds the charts that explain them, such as trend over time or breakdown by branch. The bottom or a separate tab holds tables for people who need to dig into individual records.
Which chart should you use for which question?
Choose the chart by the question, not by what looks impressive. Simple, familiar charts are read faster and more accurately than novel ones.
| Question | Best choice | Avoid |
|---|---|---|
| What is the number right now? | Big number with comparison (vs target or last period) | Gauges and speedometers |
| How is it changing over time? | Line chart | 3D charts, many overlapping lines |
| Which item is biggest or smallest? | Horizontal bar chart, sorted | Pie charts with many slices |
| What share does each part take? | Stacked bar, or a pie with two or three slices | Donuts with ten segments |
| Which records need action? | Table with highlighted rows | Charts that hide the names |
| How close are we to a target? | Progress bar or bullet chart | Decorative dials |
How do you give numbers context?
A number on its own means little. “Sales: ₹18.4 lakh” is only useful next to something: the monthly target, the same period last year, or the trend over the past twelve weeks. Every headline figure should carry at least one comparison, and the comparison should be stated in words, such as “+12% vs October”, not just shown as an arrow.
- Use colour for meaning only. Reserve red and green for good or bad against a target. Everything else can be neutral, so the warnings stand out. Do not rely on red and green alone, since some users have colour vision deficiency; add an icon or label.
- Format numbers for your readers. Indian audiences read ₹12.5 lakh and ₹3.2 crore faster than 1,250,000 or 32 million. Use Indian digit grouping (1,25,000) in tables and round headline figures.
- Label directly. Put labels on lines and bars instead of a separate legend people must match by colour.
- Show the period and freshness. State clearly “This month, updated 10:30 am”, so nobody acts on stale data.
Should dashboards work on mobile?
For most Indian owners and managers, yes. Many check business numbers on a phone between meetings or on the road, often after a WhatsApp nudge. A dashboard that only works on a wide office monitor will be ignored on the move.
Design the phone view deliberately rather than letting a desktop grid squash itself. Stack the headline numbers vertically, show one chart per screen width, swap wide tables for lists of the top items, and keep tap targets large. A scheduled daily summary sent by email or WhatsApp, linking to the full dashboard, often gets more use than the dashboard itself. See Mobile-first UX: Designing for Thumbs, Not Cursors for the underlying principles.
How do you test a dashboard before rolling it out?
Test with the people who will use it, using their real data. Sit with three or four users and ask them to answer specific questions from the screen, such as “Which branch is furthest behind target?” or “How much is overdue beyond 60 days?”. Time how long each answer takes and note where they hesitate.
- Can each user state, within about five seconds, whether things are on track?
- Can they find the answer to their three most common questions without help?
- Do any numbers contradict figures they see elsewhere, such as in the accounts package?
- Is anything on the screen they never look at? Remove or demote it.
- After two weeks, check usage: who opens it, how often, and which views are ignored.
Mismatched numbers destroy trust faster than poor design. Agree definitions in writing before building, for example whether “sales” includes GST, whether returns are deducted, and which date (invoice or payment) a sale belongs to.
