Dual Axis Charts: When They Mislead and What to Use Instead

Dual-axis charts are common in dashboards because they fit two measures into one space. But the two scales are independent, so the chart can show crossings and trends that the data does not support. This guide shows the problem with real numbers, and four clearer ways to show the same data.

What a dual-axis chart is

A dual-axis chart shows two measures on the same time axis, each with its own y-axis: one on the left, one on the right. A common example is sessions as bars (left axis) and conversion rate as a line (right axis).

It saves space and puts two things side by side. The problem is that the two y-axes are independent, so their ranges are a choice, and that choice decides what the chart appears to say.

How axis ranges change the story

Take six weeks of data (sample numbers): sessions rise from 10,000 to 16,000, conversion rate falls from 3.0% to 2.4%.

• Left axis 0–20,000 and right axis 0–4%: the bars rise gently and the line drifts down. Readers see "traffic up a bit, conversion down a bit".
• Left axis 9,000–16,000 and right axis 2.3–3.1%: the bars shoot up and the line plunges; the two lines cross in week 3. Readers see "a crisis that started in week 3".

Same data, opposite impressions. The crossing point has no meaning at all; it is an artefact of the two ranges. Readers also tend to read the gap between the lines as meaningful, and to confuse which line belongs to which axis.

The same data, four ways

The six weeks behind both versions above are listed in the chart below. Week 6 vs week 1: sessions +60%, conversion rate −20%, conversions +28%.

1. Dual axis: shows both, but the story depends on the ranges chosen.
2. Two panels, one above the other: sessions on top, conversion rate below, sharing the time axis. Each has an honest scale; the alignment of weeks is kept.
3. Index to 100: divide each series by its week 1 value. Sessions end at 160, conversion rate at 80, conversions at 128. One axis, one unit, direct comparison of change.
4. The count that matters: if the question is "did the extra traffic pay off?", chart conversions (300 → 384). It answers the question directly: yes, but less than traffic suggests, because the new visitors convert worse.

Which to use

The table below compares the options.

When in doubt, chart the outcome the reader cares about.

If you must use a dual axis

• Use it only when the two measures are closely related and the audience knows both well.
• Start both axes at zero, or explain why not.
• Use different mark types (bars and a line) and colour each axis label to match its series.
• Label the axes with units, and put the measure names in the legend and on the axes.
• Never draw conclusions from where the lines cross or how far apart they are.

Doing it in TapCub

In All-platform analytics › Overview › My dashboards, put each measure on its own card from a saved analysis, with the same date range and granularity, and place them one above the other so the weeks line up: the two-panel approach. For a direct comparison of change, save a custom metric that divides one measure by another in All-platform analytics › Behavior › Metrics & SQL, or chart the outcome count itself. For telling the story around the chart, see data storytelling; for dashboard layout, see KPI dashboard.