You’ve seen the chart. A line trending down, usually green, sometimes with a leaf somewhere nearby. It looks like evidence of progress.
What it almost never tells you is which year it started from, or whether emissions actually fell or just grew more slowly than revenue. Take those away and all the line really says is that someone wanted you to feel good.
I worked on an ESG analytics platform where nobody was reading glossy reports. Users were preparing regulatory filings and checking supplier emissions against targets with legal deadlines attached. For them, a mislabelled metric was a legal risk.
Most misleading sustainability charts aren’t made by liars. Alberto Cairo makes this point in How Charts Lie: they’re usually made by people who never stopped to ask what the chart is actually telling someone. A squashed y-axis turns a small improvement into a dramatic one. An intensity metric, like emissions per euro of revenue, can show lovely progress while total emissions go up.
Add the missing context, one switch at a time
Emissions per euro of revenue
Tonnes of CO₂e per €1 million of revenue, 2021 to 2024
| Per €m of revenue | Total emissions | |||
|---|---|---|---|---|
| Year | tCO₂e | vs 2021 | kt CO₂e | vs 2021 |
| 2018Left out | 71 | 380 | ||
| 2019Left out | 78 | 405 | ||
| 2020Left out | 92 | 430 | ||
| 2021 | 100 | Baseline | 455 | Baseline |
| 2022 | 88 | −12% | 470 | +3% |
| 2023 | 79 | −21% | 488 | +7% |
| 2024 | 70 | −30% | 503 | +11% |
Changes are compared with 2021. Every value is ± 12%. Illustrative data for a fictional company.
Illustrative data
So the question I ask of every chart is whether it helps someone understand what’s happening, or just helps them feel like things are going well.
This matters even more because the data underneath is shaky. A 2022 study by Berg, Kölbel and Rigobon found that ratings for the same company from different agencies can correlate as weakly as 0.38. In 2023, Chevron was rated high-risk by Sustainalytics and fairly near the top by MSCI at the same time. If your product shows a number like that with no sign of the uncertainty, it’s presenting confidence it doesn’t have.
The fix isn’t technically hard. Make the baseline year obvious, and put absolute and intensity numbers side by side so people can see how they relate. Show uncertainty ranges instead of hiding them, and say plainly what’s missing and how the data was collected. Xaquín G.V., who writes about ethics in data visualisation, makes the point that disclosing scope and limitations is part of the chart itself. I’d go further and call it interface design. It just takes someone willing to ship the honest chart over the reassuring one.
References