Dashboard design: stop a growing total from hiding a falling conversion rate

A larger total can tell the wrong story

This is an evergreen dashboard exercise using fictional training-enquiry data. Imagine that enquiries rose from 100 in August to 160 in September. A headline card reading ‘Enquiries up 60%’ sounds positive. But qualified enquiries rose only from 40 to 48, so the qualification rate fell from 40% to 30%. Both calculations are correct; the total alone hides a change in quality.

The dashboard should not replace one metric with another. Show enquiries, qualified enquiries and qualification rate together, then explain the decision each supports. Volume answers how much work entered the process. The rate answers what share met the stated qualification rule. Neither tells you why the change occurred without further evidence.

Define the numerator and denominator

Write the rate as a sentence before building a chart: qualified enquiries divided by all eligible enquiries received during the same period. Define ‘qualified’, ‘eligible’ and the date used to assign a record to a month. Decide how duplicates, spam, reopened enquiries and records with missing status are treated.

A rate can change because its numerator moved, its denominator moved or both moved. Keep the underlying counts visible so a user can distinguish a genuine shift from a percentage based on a small or changing population. If the definition changes between months, mark the break and avoid presenting the series as directly comparable.

Source: US Census Bureau Statistical Quality Standard E2: Reporting Results

Put quality information near the headline

The UK Office for Statistics Regulation recommends prominent critical caveats, more detailed definitions and limitations in a second layer, and links to supporting methodology for deeper information. A vague warning such as ‘use with caution’ is not enough; users need the exact issue and what it means for interpretation.

For the fictional dashboard, a useful note might say: ‘September includes five working days of delayed status updates; qualification rates may rise after review.’ Place it beside the headline rather than hiding it on an unrelated help page. Record the last refresh time and the period covered so users do not mistake partial-month data for a complete month.

Source: Office for Statistics Regulation: Regulatory guidance for dashboardsGovernment Analysis Function: Testing dashboards for design and accessibility

Complete the two-month metric lab

Create a small spreadsheet or local dataset with one row per fictional enquiry. Include received date, source, status and status-reviewed date. Build a summary for August and September, then review it with a checklist before choosing any chart.

Use the same eligible population and time rule for both months. Recalculate the totals from the raw rows instead of typing headline numbers manually. Add an automated check that the sum of status categories equals the eligible enquiry count, allowing separately documented unknown values where necessary.

• Calculate total enquiries, qualified enquiries and qualification rate for each month.
• Display the numerator and denominator next to every percentage.
• Filter out one duplicate and record why it is excluded rather than deleting it silently.
• Add a missing-status category and explain how it affects the rate.
• Use consistent scales and label every unit, period and data refresh date.
• Write one conclusion the data supports and one tempting conclusion it cannot support.

Source: US Census Bureau Statistical Quality Standard E2: Reporting ResultsGovernment Analysis Function: Data visualisation charts guidance

Choose the visual after choosing the comparison

Office for National Statistics guidance says chart titles, subtitles, axis titles and labels should work together to explain the trend, population, period and units. A title such as ‘Qualification rate fell while enquiry volume grew’ is more informative than ‘Monthly performance’, provided that it accurately describes the most prominent visual pattern.

Use simple bars for the two monthly counts and a separate line or clearly labelled chart for the rate. Avoid a dual-axis design that makes unrelated scales appear to move together. Do not truncate an axis merely to exaggerate a small change; if a non-zero baseline is necessary, make that choice unmistakable and explain it.

Source: Office for National Statistics: Chart text guidanceGovernment Analysis Function: Data visualisation charts guidance

What to include in a data portfolio

Keep the fictional source rows, metric definitions, transformation steps, validation checks and final dashboard. Add a short decision note explaining why the rate accompanies the total and how missing statuses are handled. This lets a reviewer reproduce the result instead of judging only a polished screenshot.

Learners in Chennai or online can use the lab while exploring SPOTHUB’s Data Analytics + AI learning path. Replace fictional data only with information you are authorized to use, apply privacy rules before publishing any breakdown, and do not imply causal explanations from two descriptive periods. Training supports practice but does not guarantee employment or a particular outcome.

Sources and further reading

• Office for Statistics Regulation: Regulatory guidance for dashboards · checked 2026-09-27
• Government Analysis Function: Testing dashboards for design and accessibility · checked 2026-09-27
• Office for National Statistics: Chart text guidance · checked 2026-09-27
• Government Analysis Function: Data visualisation charts guidance · checked 2026-09-27
• US Census Bureau Statistical Quality Standard E2: Reporting Results · checked 2026-09-27

Spot an error? Email info@spothub.in with the article link and correction.