What to consider when creating dual-axis charts

Dual-axis charts — also called double-Y charts, combo charts, superimposed charts, or mixed line-and-bar series — might be one of the most controversial chart types out there. And yet, they’re widely used, and for good reason: they’re great at showing how the ups and downs of multiple measures with different units compare. You can read information like “Measure 1 was high in May but low in June, and for measure 2, it was the opposite” out of a dual axis chart easily – easier than out of a small multiple chart, for example, where the series sit next to each other.

So why are they so controversial? Because they invite readers to misread them, as we argued many years ago. Scatter plots, box plots, and other complex chart types both look and are hard to read. Dual-axis charts look easy to understand – like your friendly line chart next door – but they’re not. When your readers see a complex-looking chart, they might think, “This looks too complicated; I can’t get insights out of it.” When seeing a dual axis chart, the risk is that they do get insights from it, but the wrong ones (”The revenue line in May is higher than the sales line, so revenue must be higher!”).

So let's talk about when dual-axis charts are a good choice, when they're not, and how to make sure your readers read them correctly.

When to use dual-axis charts

Dual-axis charts are often used to plot data like stock price vs. volume, temperature vs. precipitation, revenue vs. units sold, and job openings vs. the unemployment rate. Here's what to keep in mind when you do want to use them.

Use dual-axis charts if your audience knows how to read them. If you want to create dual-axis charts for a mainstream audience, they have to be extremely well designed (it can be pulled off). But if “your only readers are regular readers” who are used to dual-axis charts, as Toph Tucker writes in his great article Doing enterprise financial data visualization after data journalism, they’re often fine. “Sometimes [those highly-numerate people] aren’t looking for a statistic or trend or thesis; they are navigating a space,” Toph explains. “When this happened to this instrument, what happened to that? This isn’t a correlation, it’s: when there was a single big move in one, what did the other do?”

Only use dual-axis charts when comparing data series with different units. One way to remind people that two data series in a dual axis chart have little to do with each other is to use measures with different units (like USD and %, or °C and mm). Chart readers who start thinking “Ah, revenue in May was higher than products shipped” will (hopefully) quickly realize that they they can't compare apples and oranges.

For overlapping axes ranges, consider a (multiple) line chart. Sometimes the lines in a simple line chart show the same units (e.g. $), but cover different ranges, so that at least one of them is only a few pixels tall. Using a dual axis chart to make the up and downs of all your lines visible can feel like a good idea. Don’t. Readers are too likely to remember the view you show them, with a wrong insight like “Lesotho had a higher GDP than Uganda at the start of the chart.” Use a simple line chart to give readers the correct big picture. If you want to make the ups and downs visible, consider a small multiple lines chart with different y-axis ranges.

Use dual-axis charts to show different units for the same measure. Here’s the safest use case for dual-axis charts: Showing the same measure in two units, like °C and °F, meters and feet, US dollar and €, or count and share. This is useful to a lot of readers — including those outside (and inside!) the US who are baffled by °F-only charts.

Use dual-axis charts for Pareto charts. Pareto charts are a quality-control tool that show how often something occurs, both in absolute and relative terms. They can reveal, for example, that the three most common defects account for 80% of all quality issues. The columns (and the left axis) show the number of occurrences per issue, while the line (and the right axis) shows the cumulative share of the total number of occurrences.

To make trends more comparable, consider indexed charts. Indexed charts show how much each data series has changed, in percent. Compared to dual axis charts, they make it easier to see if Measure 1 grew more or less than Measure 2. And they give a more truthful picture of your data: In indexed charts, the slopes and angles of all lines can be compared to learn something about the data, while in dual-axis charts, they cannot. To bring the advantages of indexed charts into your dual-axis charts, align the axis scales (see below).

How to make truthful dual-axis charts

You can design better and worse dual-axis charts depending on how you set the range for each of the two axes. It’s a powerful tool with which you can stretch and shrink your data series at will. As we wrote in 2018: “The scales of dual-axis charts are arbitrary and can therefore (deliberately) mislead readers about the relationship between the two data series.” Here’s how to do better.

Align your axis scales. Aligning axis scales means making sure that both axes “grow” at the same rate. If one of your axes grows by 20% (e.g. from 50 to 60), while the second line grows by 80% (from 400k to 720k), then your scales are not proportional. When you align the covered share of your two axes, readers can actually compare the growth of your data series, very similar to an indexed chart (see above) – while keeping the advantages of a dual-axis chart, like being able to see absolute values.

If you start one axis at zero, start the other one at zero, too. If your dual-axis chart includes columns or areas, it comes with a zero baseline. Which is great: it helps readers to understand how big your numbers are, not just how much they change. Zero baselines make it easy to see things like “We doubled AI usage between May and August.” To make sure that all your metrics have a shared reference point and your readers can make absolute statements with the second axis, too, start both axes at zero.

Avoid line crossovers. In a normal line chart, lines crossings tell us important information (”Bread got more expensive than candy! Crazy!”). In a dual axis chart, a line crossing tell us nothing. It’s purely a result of design choices: depending on how you set each axis’s range, the crossing will happen sooner or later or not at all. If the shape of your lines allow, go for “not at all.”

How to make it obvious that you're showing measures with different ranges

A dual axis chart with two lines looks like a line chart – and your readers might read it like one. We want to avoid that. Here's how to make clear that a dual-axis chart has to be interpreted differently.

Use different chart types in one chart. If it suits your data, show one of your data series with lines and the other one with areas or columns. This way, readers are more likely to either recognize that it’s a dual axis chart, or, (if they’re not used to dual-axis charts yet), to pay extra attention to how this differs from what they’re used to. If you do use different chart types, make sure they still let readers see peaks, valleys, and trends. In some cases, a dual-axis chart with lines on both sides shows the overall pattern better.

Label values. Readers notice labels sitting right next to your data points more than they notice the axis ticks on the side of the chart. Consider adding labels (by using the value label feature or by adding annotations) for all data points if your chart allows for it, but at the very least for peaks, valleys, and/or the first and last value. A “7m” below a “400k” is a nice reminder that your data series can't be compared.

Mention the different measures and units in your title or description. Explain both measures in your title or description, not just one. If your chart shows both sales value and units shipped, it becomes clearer if your title “Sales value higher than ever” gets the addition “….while shipped units decreased.” Also consider coloring mentions of measures directly in the title or description, to make it obvious which color shows what.

Mention the different measures and units often in the text on your chart. Try to remind readers again and again of what they’re seeing:

• If you have space, add annotations that mention the measures and units.
• Add a legend and be unmistakably clear in it. Instead of just writing “Sales” (and relying on the little $ sign next to one of the axes to communicate that this is sales value, not units), write “Sales in $.”
• Add prefixes and suffixes directly to your values that make the units clear. Don’t just rely on writing “in USD” in your description. Adding a $ prefix to your numbers makes sure the currency sign shows up on your axis and in tooltips.
• Show tooltips on hover that state the measures. In Datawrapper, tooltips are turned on by default because they’re such a big help for your readers. Only turn them off when absolutely necessary.

How to make it obvious which axis is which

Once you established that you’re showing measures on two different axes, here’s how to make sure your readers quickly understand which one belongs to which.

Use contrasting colors. If both axes' data series come in shades of blue, your chart might be nicely on brand — but your readers will have a hard time understanding which series belongs to which axis. Consider coloring all the series on one axis in a similar hue (e.g. dark orange and red) and all the series on the other axis in a completely different hue (e.g. medium and dark blue).

Use the same color for a data series as for its tick (and axis) labels. To make it very clear which data series uses which units and axis ranges, consider giving your axis labels and ticks the same color as the data series. This works best when you show just one series per side: revenue on the left axis and sales on the right, for example. If you show more than one series with asimilar coloron one of the sides– like three green-ish lines – consider using that color. (Datawrapper will do this automatically for you.)

Show the axis labels and color legends on the axis they apply to. Help readers figure out which axis shows what by aligning the labels with the axis. To be able to do so, consider shortening your axis labels and moving an in-depth explanation into the chart description or footer.

Don’t overload your chart with lots of lines, columns, or areas. Let’s not make this chart type more complicated than it already is. Yes, you can show five different lines on stacked columns — but please think twice before you do. Chances are it will make your readers give up on reading this at all. A split bar chart (for a few values) or a table or small multiple line chart (for many) might be the better choice here. For dual-axis charts, one line on a stacked bar or stacked area chart, or two lines on a simple column or area chart, is often hard enough to decipher.

Think about which y-axis to place where. You can help readers make sense of your chart by deliberately considering which of the two y-axes should show which data series:

• If the reading direction of your readers is left-to-right, consider putting the most important axis on the left. For example, in a chart showing both °C and °F, put the unit you expect most people to use on the left.
• If one data series has especially high values close to one of the axes (by starting or ending high), consider making that the axis that describes it.

Adjust areas and lines to make them work on top of each other. The colors you'd normally use for a simple column, area, or line chart might not work in a dual-axis chart: as soon as you layer those chart types on top of one another, thin lines become hard to read. Make your area and column charts lighter and less saturated, and make your lines darker and more saturated to set them apart. Giving your lines an outline, making them thicker, and adding line symbols can help too.

Example of a dual-axis chart

Go right into the editing process of this chart by clicking on "Edit this chart" in the top right corner (available on desktop devices):

Find more examples in our Academy article Examples of Datawrapper dual-axis charts.

Want to know more?

• Toph Tucker: Dual axes are a great way to show correlation. An interactive explanation of why the correlation in dual-axis charts doesn’t change, no matter how stretched one series is in comparison to another.
• Peter Ellis: Dual axes time series plots may be ok sometimes after all. About good practices when using dual-axis charts, and how aligning the axis scales (”equivalent to drawing two indexed series”) can improve a dual axis chart.