Thinking

The Green Line Going Down

Product Design·March 3, 2026·5 min read

There’s a chart in almost every sustainability report. You’ve seen it. A line, trending downward. Usually green. Sometimes a leaf somewhere nearby. It looks like evidence. It isn’t, necessarily.

Here’s what that chart almost never shows: the baseline year it started from, whether the reduction is absolute or just relative to revenue, what happened to production volume over the same period, or whether the methodology changed halfway through. Strip those things out and the line doesn’t tell you anything except that someone wanted you to feel good.

This is the quiet problem at the center of sustainability data design. Not the data itself — the presentation of it. The decisions made before the chart reaches you: what to include, what to leave out, where to start the axis, which metric to foreground. Those decisions are almost never explained. And in a domain where the numbers are supposed to matter — where the whole premise is that measuring environmental performance will change behavior — unexplained decisions are a form of fiction.

The same data — two different charts
What most reports show
10080201820192020202120222023
Y-axis starts at 80. A 14% reduction appears as a near-complete collapse.
The same data, honestly scaled
120020182020202120222023intensity metric, not absolute↑ 2019 baseline
Y-axis from 0. Same data: a 14% reduction over 5 years — intensity-based, 2019 baseline.
Strip the missing context and the line tells you nothing except that someone wanted you to feel good.

I worked on an ESG analytics platform where the users weren’t reading annual reports. They were doing compliance work. Preparing regulatory filings. Cross-referencing supplier emissions data against reduction targets with legal deadlines attached. For them, a mislabeled metric wasn’t an embarrassment. It was a liability.

What that experience made clear: most misleading sustainability visualizations aren’t dishonest. They’re just unconsidered. Alberto Cairo, one of the sharpest thinkers on data visualization, argues in How Charts Lie(2019) that most misleading charts come from oversight, not intent — that even well-meaning designers produce graphics that deceive, because they never stopped to ask what the chart was actually telling someone.

The mechanism is usually structural. A compressed y-axis makes a modest improvement look dramatic — a well-documented manipulation tactic in which truncating the axis creates a deceptive visual-to-data ratio, making small changes appear far more significant than they are. Intensity-based metrics — emissions per dollar of revenue — can show “progress” while absolute emissions climb. A conveniently chosen baseline year makes a target look ambitious when it isn’t. None of this requires bad faith. It just requires not asking the hard question out loud.


The hard question is: does this chart help someone understand what’s actually happening, or does it help someone feel like things are going well? Those are not the same product.

There are now over 600 different ESG scoring systems worldwide, each using different indicators, different weights, different definitions of the same words. A landmark study by Berg, Kölbel, and Rigobon (2022), published in the Review of Finance, found that ratings for the same company from two different agencies can correlate as weakly as 0.38 — barely better than a coin flip.

The same company, rated simultaneously by two agencies — 2023
Chevron Corporation
SustainalyticsSecond worst category out of five — High RiskHigh Risk
MSCIThird best category out of seven — near top-tierA Rating
Berg, Kölbel & Rigobon (2022) found ratings for the same company across agencies correlate as weakly as 0.38 — barely better than a coin flip. Building a product on this data without surfacing the uncertainty presents the appearance of certainty, not analysis.

When you build a product on top of that fragmented foundation and don’t surface the uncertainty to the user, you’re not presenting analysis. You’re presenting the appearance of certainty. That’s a design choice. It’s also an ethical one.

The designers who get this right treat transparency as a feature, not a disclaimer. Methodology visible. Baseline year prominent. Absolute and intensity metrics side by side so users can see the relationship between them. Uncertainty ranges shown, not hidden. As data visualization ethicist Xaquín G.V. puts it: make the scope and limitations of your data clear — note missing data, biases, and assumptions that may affect interpretation. Disclose how data was collected, processed, and filtered. That’s not a footnote. That’s interface design.

None of it is technically hard. All of it requires someone willing to ship the chart that communicates honestly over the chart that communicates confidently.

The green line going down is easy to make. Making it mean something is the work.

References

  1. Data-Nizant (2025). 7 Examples of Bad Data Visualization to Learn From. datanizant.com
  2. CLEVER°FRANKE (2024). ESG Needs Data Design, Part 1. Medium. medium.com
  3. Berg, F., Kölbel, J.F., Rigobon, R. (2022). Aggregate Confusion: The Divergence of ESG Ratings. Review of Finance, 26(6), 1315–1344. academic.oup.com
  4. Pictet Asset Management (2023). Analysing Corporate ESG Ratings. am.pictet.com
  5. Data Europa EU (2025). Honest Charts: Ethics and Integrity in Data Visualisation. Interview with Xaquín G.V. data.europa.eu
  6. Cairo, A. (2019). How Charts Lie: Getting Smarter about Visual Information. W. W. Norton & Company.
Design the uncertainty, not just the trend. 📉 Design the uncertainty, not just the trend. 📉 Design the uncertainty, not just the trend. 📉 Design the uncertainty, not just the trend. 📉