Choosing Business Measures Before Building a Dashboard

A dashboard is a way to present information, not a decision in itself. If a business starts with charts, it can spend money displaying whatever its systems already count rather than what managers need to know. Begin with the decisions you make each week or month, then choose a few measures that could improve them. The first result may be a simple report rather than a new dashboard.

Begin with a real decision

Ask what action a manager could take after seeing the information. A service firm may need to decide whether to add staff on busy days, investigate delayed jobs, or change how appointments are scheduled. An order business may need to identify which products create repeated fulfillment problems. Each decision needs a different measure. A total number of website visits is unlikely to answer any of them unless the question concerns customer acquisition.

The GOV.UK service standard on defining success links measures to the problem a service is meant to solve and calls for revising them as needs become clearer. Its reporting obligations belong to public services, but the principle helps a small business avoid collecting numbers merely because they are easy to display.

Define each measure precisely

Decide what counts, when it counts, and which record is authoritative. Does a completed order mean it was packed, delivered, or accepted by the customer? Does an active customer mean a paid account or recent use? Write down the definition beside the report. If two teams use a word differently, the chart may appear precise while its input is inconsistent.

Check a sample of records against the measure. Ask staff where source information is entered and whether it is corrected when work changes. A visual design cannot repair unreliable entries. The GOV.UK guidance on measuring service success recommends combining quantitative measures with user research, support contacts, and cost information. Numbers can show where to look; conversations and case records often explain why.

Show enough context for action

Choose a comparison period that fits the business cycle and show the underlying count. A percentage based on only a few cases can swing sharply and invite a false conclusion. Separate branches or customer groups only when the comparison is meaningful. Include known data gaps and when the report was last updated. Give the reader a path from an unusual figure to the relevant cases rather than presenting a chart that cannot be investigated.

Ask how the measure could be misread. A rising average order value may reflect a different customer mix rather than better selling. A lower support volume may mean customers found the product easier, or that reaching support became harder. Note alternative explanations beside important trends and check them with staff or customers before making a costly change. A dashboard should invite sound questions, not create false certainty.

Before adding a new data field, decide who will enter it and what they will do when the answer is unknown. A measure that depends on guesses will not become reliable because it is displayed attractively. If the information is expensive to collect, compare that effort with the value of the decision it supports.

Set an owner for each measure and a review date. If nobody uses it to decide anything, remove it. If it repeatedly raises the same question, improve the source or add the missing context. A small set of trusted figures can support a better conversation than a crowded dashboard. Start with a basic report, observe how people use it, and add interaction only where it helps a real decision.

Holism can help an owner define the business questions and have a developer use AI and reusable foundations to produce reliable, traceable information. The owner determines what the figures mean and how to act on them. Build the dashboard after the measures have earned their place in decisions, not before.