Your Charts Are Lying

You've seen both kinds of charts. The one where the insight hits you immediately, and the one that looks like colorful spaghetti where everyone nods politely while understanding nothing.

Most people think the difference is design talent, but it's really about knowing how charts lie to people.

Every chart tool ships with defaults that work against you, from truncated axes to rainbow palettes to 3D effects, and most people never change them.

So let's start with the most common problem: noise.

---

Too much noise

Every visual element on your chart makes the viewer work a little harder. Grid lines, gradients, shadows, borders, and decimal points each take a tiny slice of attention. Add enough of them and there's nothing left for the actual data.

Toggle each element below to feel the effect:

When you strip away the decorations, the self-serve channel becomes impossible to miss, which is the whole point: your data should be the loudest thing in the room.

The default Excel chart throws everything at you because Microsoft doesn't know what you're trying to say. But you do, so be ruthless about removing anything that doesn't help your message.

But even a clean chart can mislead if you pick the wrong format...

---

Wrong chart, wrong answer

Every chart answers a question. The format you choose should match that question. This sounds obvious, but I see people using pie charts for trend data all the time.

Click each question below to see which chart format fits:

That's because different questions need different visual tools. A line chart is terrible for ranking, and a bar chart is wrong for showing change over time. The format you choose is part of your message.

Pick the wrong format and your audience has to fight the visualization to understand the data. Pick the right one and the insight jumps out without effort.

Some chart types are just broken, and pie charts are the worst offender.

Can you tell which slice is biggest? Click to guess:

Our brains are terrible at comparing angles and areas but excellent at comparing lengths, which is why bar charts win almost every time.

A few other chart types to watch out for: stacked area charts have a moving baseline that makes comparison nearly impossible, since only the bottom series is accurate. 3D charts distort values because perspective makes the front bars look bigger than identical back bars.

Dual y-axes are easy to manipulate and hard to interpret, because you can make any two trends look correlated by adjusting the scales. And if you have more than 4-5 lines on a single chart, everything becomes unreadable, so use small multiples instead.

So you've decluttered your chart and picked the right format. But your data can still lie in subtler ways...

---

Subtle lies

A truncated y-axis can make a 2% change look like a 200% change, and a rainbow color scale can hide the pattern you're trying to show. Most tools ship with these settings turned on by default.

Toggle the y-axis below and watch the same data tell two completely different stories:

Both charts show identical data, but one looks like a crisis and the other looks like stability. Truncated axes are dangerous in bar charts because your brain reads bar height, not axis labels.

As a rule of thumb, always start your y-axis at zero for bar charts. Line charts get a pass sometimes, because readers follow the shape of the line rather than the height from the baseline.

Color can be just as misleading. Toggle between the palettes below to see what I mean:

Rainbow palettes look appealing but they distort perception. Yellow appears lighter than blue regardless of what value it represents. With a sequential single-hue palette, the mapping is obvious: darker means more, lighter means less.

Roughly 8% of men have some form of color vision deficiency. If your chart relies on the difference between red and green, a significant chunk of your audience can't read it at all. A good test is to check whether your chart still works in grayscale.

You can get all of this right and still end up with a chart nobody remembers...

---

Data without a story is noise

Say someone tells you 847 gigabytes were transferred last month. How much is that? You have no idea, and neither do I.

Now say it differently: we moved 170,000 HD movies worth of data last month.

Which one will you remember tomorrow?

That context is what makes a number stick. But to make people act on data, you also need a narrative.

Here are four data points from a support dashboard: 4.2 hours, 68%, 23%, Mondays.

Now here's a story: our average response time is 4 hours, but Monday mornings hit 12 hours and satisfaction drops to 41%. One extra person on Monday shift fixes it.

The data is the same, but the second version actually tells you what to do about it.

A good data story has three parts: set the context, show the problem, and point to action. The story version above works because it ends with a clear next step: add one person to the Monday shift.

---

The one thing to remember

At the end of the day, the goal is to make charts that actually communicate. Whatever you're presenting should be immediately obvious to someone seeing it for the first time. If you need a paragraph of text to explain your chart, the chart isn't working.

Next time you make a chart, spend 2 minutes checking: is it clean, is the format right, is the axis honest, do the colors work, and does it tell a story? Your audience will spend hours looking at your work, so you owe them clarity.