Your Charts Are Lying to Your Stakeholders (And You Don't Know It)

The Dashboard That Cost a Budget Decision

Last quarter, I reviewed an executive dashboard that showed two lines on a dual-axis chart: marketing spend and revenue. The lines moved almost in lockstep. The implied message was clear: "every dollar we spend on marketing drives revenue."

The VP almost approved a 40% budget increase based on that chart.

Then I re-plotted it with a proper correlation analysis. The actual correlation coefficient was 0.23 — barely meaningful. Both lines happened to trend upward over time because the company was growing. Marketing spend and revenue were almost independent.

That dual-axis chart nearly cost the company six figures in misallocated budget.

Every chart is an editorial decision. The moment you choose a chart type, axis scale, colour, or annotation, you are telling a story. The question is whether you're telling the truth.

Lie #1: The Truncated Y-Axis

This is the most common visual lie, and it's the easiest to commit accidentally.

Take this revenue data across four quarters: $4.1M, $4.2M, $4.15M, $4.25M. The actual range of variation is about 3.6%.

If you start the y-axis at $0, the line looks almost flat — a gentle 3.6% wobble. Accurate, but visually boring.

If you start the y-axis at $4.0M, the same data looks like a dramatic roller coaster with huge swings. Same numbers. Completely different story.

| Y-Axis Start | Visual Story | Actual Change | |-------------|-------------|---------------| | $0 | Stable, flat revenue | 3.6% | | $4.0M | Volatile, dramatic swings | 3.6% | | $4.1M | Explosive growth | 3.6% |

The fix: Always show a y-axis that starts at zero for bar charts. For line charts, if you truncate the axis, add a visual break (a zigzag line) to signal that you've cut the axis. And ask yourself: "Would I be comfortable if my stakeholder only saw the chart and not the numbers?"

Lie #2: Pie Charts with Too Many Slices

Pie charts work for one thing: showing parts of a whole when you have 2-4 categories. That's it.

I once saw a pie chart with 14 slices showing regional revenue breakdown. The smallest slices were indistinguishable. The legend was bigger than the chart. And the key insight — that three regions accounted for 72% of revenue — was completely invisible.

If you have more than 4-5 categories, switch to a horizontal bar chart sorted by value. Your stakeholders will read it in half the time and actually remember the key takeaway.

Lie #3: Dual Axes That Imply False Correlation

This is the sneakiest lie because dual-axis charts feel rigorous. Two metrics, two scales, one chart. Very analytical.

The problem: you can always adjust two arbitrary scales until any two trending lines look correlated. Revenue going up? Marketing spend going up? Set the scales right and they'll appear to move together perfectly.

The fix: If you want to show correlation, use a scatter plot with a calculated correlation coefficient. If you want to show two trends over time, use two separate small charts stacked vertically (small multiples). Don't combine them into one chart unless you've verified the statistical relationship.

Lie #4: Colour That Implies Meaning Where There Is None

I've seen dashboards where red means "bad" in one visual and "segment A" in another visual on the same page. The human brain can't override pattern recognition — if red means "danger" in your revenue chart, your brain will read it as "danger" in your segment chart too.

Consistent colour semantics matter. In every dashboard I build at AlCircle:

• Green = on target or positive variance
• Red = below target or negative variance
• Amber = within 10% of target (caution zone)
• Blue = neutral informational metric
• Grey = inactive, historical, or comparison baseline

This isn't decoration. It's a visual language that lets your stakeholders scan a page in 3 seconds and know what needs attention.

The Ethics of Data Visualization

Every chart you build is a persuasion tool. That's not inherently bad — data should drive decisions. But when you accidentally (or deliberately) mislead through chart design, you erode trust.

The test I use before publishing any dashboard: "If someone only saw this chart and not my explanation, what would they believe?"

If the answer differs from reality, the chart needs fixing.

Your stakeholders trust you with the data. Don't betray that trust with a truncated y-axis.

What's the worst chart you've ever seen in a meeting? I'm collecting horror stories for a follow-up post.