How Can a Dashboard Avoid False Precision?
From the earlier reference collection. Read the current software blog.
Display only as many digits as the measurement and calculation can support, and explain how aggregation changes a value.
Match digits to the measurement
Extra decimal places can make a number appear more accurate without adding useful information. If a sensor records temperature to the nearest tenth of a degree, showing 18.437°C implies a resolution the instrument did not provide. Rounding to 18.4°C better reflects the source.
The right precision depends on the use. A tiny difference may matter in a laboratory record but not in a regional overview. Choose a display rule for each kind of measure, and use it consistently across the headline, chart, tooltip, and download documentation.
Understand precision after calculation
Averages and totals can produce more digits than their inputs, but arithmetic does not automatically improve measurement quality. Suppose three readings are 10.1, 10.2, and 10.4°C. Their computed mean is 10.233…, yet a display such as 10.2°C may communicate the evidence more honestly than three decimal places.
Aggregation should also be explained. “Daily average of hourly readings” tells readers what 10.2 represents. If only some expected observations were present, report the count or coverage. Otherwise, a precise-looking average may disguise an incomplete sample.
Separate exact counts from estimates
Counts can be exact within a defined system while still incomplete in the real world. “17 reports received” may be a literal database count, but it does not necessarily mean exactly 17 events occurred. Label the measure as reports, cases, or estimated people affected as appropriate. For rates, include the denominator and period so a value such as 3.2 has a clear meaning.
For model outputs, use a precision that fits the uncertainty and identify the result as an estimate. If a forecast is 43.8 units but the plausible range spans several units, displaying 44 or a range may be more useful than a decimal.
Keep detail available without overstating it
Rounding does not require throwing information away. A chart may show rounded values while a data table offers source values with the original resolution, alongside clear metadata about their meaning. Avoid silently inventing precision during conversion or export.
The goal is not fewer digits everywhere. It is a display whose apparent certainty matches the underlying measurement and method. Readers can then compare values without mistaking formatting for evidence.