Free tool

Chart Honesty Checklist

Put a chart next to this list — one from a report, a pitch deck, a news story, anywhere. Tick each check it passes. The score updates as you go and tells you how much weight the chart can actually hold.

The Y-axis starts at zero Or, if it’s truncated, the break is clearly marked. A cropped axis turns a 3% change into a cliff. Axis labels and units are stated Percentages or absolute numbers? Daily or monthly? A chart without units can be read to mean almost anything. The time range is wide enough for context A three-month window showing steep growth may be cherry-picked. What does the two-year view look like? Every comparison names its baseline “Up 40%” compared to what — last month, last year, a competitor? A number with no reference point isn’t a result. Sample size (n) is shown for any segment claim A claim about a subgroup can be pure noise if the subgroup had 11 people in it. Small samples hide inside clean-looking charts. The population is clear Who’s counted and who isn’t? “80% of customers prefer X” means less if it only covers the ones who answered a survey. It shows correlation without implying unproven causation Two lines moving together is not proof one caused the other. Look for a third factor driving both. The source is disclosed and has no strong incentive to skew it A chart in a company’s earnings deck or a vendor’s ROI page was built by people paid to make it look good.

This checklist is the short version of How to Read a Chart Without Being Fooled, which walks through why each of these traps works. For the related question of whether the underlying number even measures what it claims to, see What Is a Metric, Really?

Get the signal, skip the noise