How to visualise data the right way — part 1
Data visualizations often look objective and trustworthy, yet small design choices can dramatically change how we interpret them. This article shows how colors, ordering, scales, and chart types can subtly guide our attention, exaggerate differences, or hide important patterns. By understanding these common pitfalls, we can learn to read charts more critically and create visualizations that inform rather than mislead.
Colour highlighting: what matters should stand out
One of the most basic rules of good data visualization is working with color. If all elements in a chart are shown in the same color, then in practice nothing is highlighted at all. Color should be used only for the key element, such as one company, one country, or one specific result we want to draw attention to.
Imagine a comparison of revenues among technology companies where we want to highlight Walmart. Walmart should be shown in its distinctive dark color, while the other companies remain gray. It does not matter whether Walmart is first or fifth in the ranking. What matters is that it carries the main message. A common mistake is the use of red, which automatically signals a problem or negative emotion and can also feel overly simplistic. A much better choice is to use a brand color when presenting results to a client, such as Spotify green or Amazon orange. Such a chart feels natural and the reader intuitively understands where to look. The difference between a good and a bad example is often just a single color, yet the impact on readability is substantial.
Graph 1: right visualisation (interactive)
Graph 2: wrong visualisation (interactive)
Battery charts
Battery charts are often used to display scores, satisfaction levels, or performance on a simple scale. A typical example is customer satisfaction ratings for brands such as Netflix, Amazon, or Uber. The key issue, however, is not only the battery style itself, but especially the correct ordering of items. If brands are arranged randomly, the chart feels chaotic and the reader must work hard to identify the best and worst results.
The correct approach is to sort the batteries from the most positive to the most negative values. The color logic should be clear and consistent, with green for positive results, yellow for neutral, and red for negative. When sorting by satisfaction, the company with the highest satisfaction score should appear at the top and the most negative at the bottom. The sorting variable is the satisfaction value itself, which determines first place, second place, and so on. For example, if Netflix has very high satisfaction and therefore a large green segment, it should be placed at the top. Poorly ordered charts can create the impression that differences are insignificant even when they are meaningful, or they may simply appear confusing. This is why battery charts are especially effective when a correct and an incorrect version are shown side by side.
Graph 3: battery chart (interactive)
Time series and missing data
Time series charts are among the most common types of visualizations, but they are also where mistakes occur most frequently. If a certain time period is missing from the data, such as financial results during a crisis or a pandemic, we should not try to hide that gap.
Imagine the development of Airbnb revenues before and after Covid, where part of the data simply does not exist. Filling in or smoothing the line would create a false impression of continuous development. The correct solution is to break the line or clearly mark the missing period. This approach openly acknowledges that the data are incomplete. Inspiration can be found in the book How Charts Lie, which shows how dangerous it is to connect points between which no data exist. Such a chart may be less visually appealing, but it is far more honest. The reader then knows that no conclusions can be drawn for that specific period.
Graph 4: time chart (interactive)
Spaghetti charts: the problem of clarity
Spaghetti charts, as I call them, arise when we try to fit too many time series into a single chart. A typical example is comparing the stock performance of several technology companies such as Google, Microsoft, Amazon, and Meta. At first glance the chart looks rich and informative, but on closer inspection it becomes cluttered and difficult to read. The reader does not know which line to follow and the main message gets lost.
One solution is to highlight one key entity, for example Google, using color. The other companies are shown in gray and serve only as context. This approach works well when we want to tell the story of one specific subject. Without highlighting, the chart becomes a tangle of lines with no clear meaning. Good visualization helps the reader immediately understand where to focus.
Graph 5: spaghetti chart (interactive)
Spaghetti charts: splitting into small multiples
If we want to compare the development of multiple subjects on equal terms, it is often better to use so called small multiples. Instead of one overcrowded chart, we create several smaller charts with the same scale. A typical example is a Formula 1 season, where each driver has their own chart showing positions or points across races. This makes it easy to see how each driver performed over time.
The same principle can be applied to companies, countries, or sports teams. The main advantage is much better readability and easier comparison. The reader does not need to decode colors or legends, but simply follows one story at a time. This approach is ideal when no single subject should be prioritized. The charts feel systematic and analytical rather than chaotic.
Graph 6: spaghetti chart split into small multiples (interactive)
Comparing two or more categories: horizontal bar charts
Horizontal bar charts are a very powerful tool for comparing categories. They are often used, for example, to display election results or candidate support. They work equally well when comparing customer satisfaction for brands such as Samsung, Apple, or Xiaomi.
When comparing multiple categories, such as very satisfied, somewhat satisfied, dissatisfied, and very dissatisfied, a horizontal layout is clearer than a pie chart. The reader can directly compare the lengths of the bars, which is more natural for the human eye than comparing angles. This type of chart also handles longer labels very well. As a result, it is suitable for more complex data structures. In many cases, it can replace several pie charts with a single clear visual.
Graph 7: horizontal bar chart with more categories (interactive)
Graph 8: horizontal bar chart with two categories (interactive)
Relationships between two variables: showing the trend
When examining the relationship between two variables, a scatter plot is often used. However, points alone may not be easy for the reader to interpret. For example, when showing the relationship between sugar consumption and the occurrence of tooth decay, individual points can appear chaotic.
Adding a trend line helps reveal the overall direction of the relationship. The reader can then see that as sugar consumption increases, the risk of cavities also rises, even though individual values fluctuate. Of course, a trend line does not imply direct causation, but it greatly simplifies interpretation. Without it, different readers might draw very different conclusions from the same chart. A well used trend does not distort the data, but helps to read them. The goal is to show the relationship, not to overwhelm the viewer with detail.
Graph 9: showing the trend — the wrong way (interactive)
Graph 10: showing the trend — the right way (interactive)