← All Kits · Excel Kit

Dual Axis Charts: When Two Scales Help, and When They Manufacture a Relationship

Michael Nocito · Updated August 2026 · Every number on this page was worked before it was published

A dual axis chart puts two series on one plot with two different scales. It is the fastest way to make two unrelated lines appear to move together, and it is also, in three specific cases, exactly the right chart. Knowing which of those you are looking at is the whole skill.

What you do: default to two panels sharing an x axis, and reach for a second y axis only when the two series are genuinely two views of the same thing.

The short version. The point where two lines cross on a dual axis chart is set by you, not by the data.

The demonstration

Six months of marketing spend, in thousands: 40, 44, 41, 47, 45, 50. Six months of support tickets: 610, 655, 600, 690, 670, 735. Two unrelated things, both drifting up because the business is busier.

Put spend on the left axis from 0 to 60 and tickets on the right from 0 to 800, and the lines sit apart with roughly similar slopes. Now set the left axis from 35 to 55 and the right from 580 to 760. The two lines lie almost on top of each other, rising and dipping in the same months. A reader takes away that spend drives tickets.

Nothing was computed. Two axis ranges were typed. The actual correlation of those two series is high, 0.99, but so is the correlation of both with the calendar, and the chart provides no way to see that. This is the shape of the argument in correlation vs causation, drawn instead of stated.

The three cases where a second axis is right

In all three, the second series is not an independent thing you hope is related. It is arithmetic on the first. That is the line: a second axis to show two aspects of one measure, not to suggest a link between two.

If you do use one, make it readable

1. Different chart types. Bars on the primary, line on the secondary. Two lines in two colours force the reader to keep checking the legend.
2. Colour the axis labels to match the series. This is the single highest-value formatting change on a dual axis chart.
3. Put the units in each axis title. Not "Value" on both sides.
4. Pick round ranges and leave them alone. If you find yourself nudging a bound to make the lines meet, stop: that is the moment the chart became an argument.
5. Never truncate the axis under bars, even on a combo chart. The bar rule from the axis page still applies to the bar half.

The two better defaults

Stacked panels, shared x

Two small charts, one above the other, same date axis, each with its own honest scale. The reader can still see whether the peaks line up, which is the real question, and nobody has to be told which line belongs to which side. This is the answer roughly three quarters of the time.

Index both to 100

Divide every point by the first point and multiply by 100. Both series now start together on one scale, and the chart answers the question people usually mean by "do these move together": which one grew faster from the same starting line.

Indexed = value / value_at_start * 100

On the numbers above, spend indexes to 100, 110, 102.5, 117.5, 112.5, 125 and tickets to 100, 107.4, 98.4, 113.1, 109.8, 120.5. Spend has grown faster than tickets over the window, which is a real finding that the dual axis version hid entirely.

Doing it in Excel

1. Select both series, Insert, Combo chart.
2. In the Combo dialog, choose Clustered Column for one series and Line for the other, and tick Secondary Axis on the line.
3. Click each axis, Format Axis, and set Bounds to round numbers.
4. Format the axis label colours to match the series, and add both units to the axis titles.

How to apply this to your own work

1. Find every dual axis chart in your reporting and ask whether the second series is arithmetic on the first. If it is not, it is a candidate for splitting.
2. Rebuild one of them as two stacked panels and compare how long each takes you to read.
3. For any pair you actually want to compare, index both to 100 and see whether the story survives.
4. Compute the correlation as a number rather than leaving it to the eye, and check it against a third variable such as time or volume.
5. If a chart's conclusion moves when you nudge an axis bound, do not publish it with that conclusion in the title.

The one habit to keep

Ask what the crossing point means. On a real dual axis chart, where the lines cross means nothing at all, because it is a product of two arbitrary scales. Anybody who points at the crossing as if it were an event has been misled by a chart you drew.

Would your dual axis chart survive somebody else choosing the ranges?