How to Chart Two Metrics With Different Scales (Without a Misleading Dual Axis)
There are four honest ways to put two differently scaled metrics on one picture. You can use a secondary axis with stated limits, or index both series to a common base. You can also stack two panels that share an x axis, or plot one metric against the other on a scatter. Which one is right depends on the question, not on the data.
The default answer is a dual axis, and it is the only one of the four that can quietly change your conclusion. Move either axis range and two lines that crossed now diverge.
This guide covers why the problem exists and what a careless dual axis costs. It then works through the four options, their honest limits, and where rebuilding by hand starts eating the time.
Why two metrics with different scales break one chart
Revenue sits in the millions. Conversion rate sits between one and four percent. Plot both against one axis and the conversion line becomes a flat mark along the bottom.
Microsoft's own guidance names the situation directly. Its documentation on adding a secondary axis says to use one when values vary widely from series to series. It also names combination charts that mix data types such as price and volume.
So the feature exists for a reason and it is not a hack. The trouble starts with what the second axis implies rather than what it does.
There is a second version of the problem that has nothing to do with units. Two series in the same unit can still differ by three orders of magnitude. A large market plotted next to a small one hides the small one completely.
What a careless dual axis costs you
A correlation you invented. The point where two lines cross is set by the axis ranges you chose. Change either range and the story changes with it, which means the visual relationship is an artefact of formatting.
A reader who cannot check you. Most people read the shape and skip the axis labels. If the two axes have different zero points, the picture says something the numbers do not.
A chart nobody can rebuild. Manual axis limits are invisible six weeks later. When the data updates, the limits stay fixed and the chart slowly stops describing reality.
The workarounds people try
Option 1: Use a secondary axis, with rules
If the two metrics genuinely belong on one picture, a secondary axis is fine. Constrain it.
Use two series and no more. Give them different chart types, usually columns for the level and a line for the rate. A reader can then tell which axis belongs to which. Label both axes with their units.
Set axis limits deliberately rather than accepting the automatic ones, and say what you set. Microsoft documents the controls under changing the scale of the vertical axis.
The limit is honesty rather than mechanics. A dual axis can show that two things moved. It cannot show that one caused the other, and it should never be captioned as though it did.
Option 2: Index both series to a common base
Divide every value by the series value at a chosen starting period, then multiply by 100. Both metrics now start at 100 and share one axis, and the chart reads as relative movement.
This is the cleanest answer when the question is which grew faster. It removes the axis choice entirely, so there is nothing left to manipulate.
It loses absolute levels. A reader cannot recover revenue in dollars from an indexed chart, so pair it with a table when the level matters.
Option 3: Stack two panels sharing an x axis
Draw two charts, one above the other, aligned on the same time axis. Each series keeps its own scale and its own honest shape.
Nothing is distorted, and the eye still compares turning points because the time axis lines up. This is the safest option and the one most people skip because it looks less impressive.
Its cost is space. Two panels take twice the vertical room, which matters on a slide and rarely matters in a document.
Option 4: Plot one against the other
If the real question is whether the two metrics are related, stop charting them against time. Put one on the x axis and the other on the y axis, one point per period.
A scatter answers the relationship question directly, and it will not flatter you. Two lines that looked coupled on a dual axis often produce a shapeless cloud here, which is the useful result.
The limit is that it hides the sequence. Add labels to a few points, or colour by period, so the direction of travel stays readable.
The shared ceiling. All four require you to decide the question before drawing. That decision is fast, and rebuilding the chart four ways to see which reads best is not.
How to chart two metrics with different scales in Powerdrill Bloom
Step 1: Upload your data
Upload the file holding both metrics with their period column. Powerdrill Bloom profiles the columns on arrival, so unit mismatches and missing periods are visible before anything is plotted.
Step 2: Describe the comparison in natural language
Say what you want to compare and why, not which chart to draw. Ask whether revenue and conversion rate moved together over the last eight quarters, and ask for the view that shows it most honestly.
Then ask for the alternatives in the same pass. Request an indexed version and a scatter alongside the time series, so the choice is made by looking rather than by guessing.
Step 3: Export the chart, report, or deck
Take out the chart you chose, a short written read of what it shows, or slides with the comparison already laid out.
Which option fits which question
| Your question | Best option | Why |
|---|---|---|
| Did both move at the same time? | Two stacked panels | No distortion, shared time axis |
| Which one grew faster? | Index both to 100 | One axis, comparable units |
| Are these two related? | Scatter plot | Tests the relationship directly |
| What are the levels and the rate? | Secondary axis, labelled | Keeps both units visible |
The habit worth forming is reading the left column first. Most misleading charts come from choosing the picture before choosing the question.
Common mistakes
Letting the software pick the axis limits. Automatic ranges change when the data changes, so the same chart tells a different story next month without anyone editing it.
Putting three or more series on two axes. A reader cannot map series to axes reliably past two. Split the chart instead.
Using identical chart types on both axes. Two lines in the same style force the reader to check the legend for every observation. Columns plus a line resolves it instantly.
Truncating one axis silently. Starting an axis away from zero is sometimes right and always needs saying. Unstated truncation is the most common way a chart misleads.
Captioning a dual axis as evidence of cause. The chart shows co-movement at best. If causation is the claim, the scatter and the underlying method belong in the same document. Our guide to charts that match your brand colours covers the presentation side.
Comparing a level against a rate without saying so. Revenue and conversion rate answer different questions. Label which is which in the caption, not only in the legend.
Reaching for a second axis before trying indexing. Indexing solves the scale problem outright in many cases, and it removes the argument about axis limits altogether.
Conclusion
Two metrics with different scales are a question about the question. Ask whether you want co-movement, relative growth, a relationship or absolute levels, and each of those has one right picture.
A secondary axis is legitimate, and it is also the option that most rewards discipline: two series, two chart types, labelled units and stated limits. If you would rather see all four views before choosing, try Powerdrill Bloom on the file. See also our guides to turning a CSV into a chart and what a funnel chart is. The line graph creator and AI graph maker pages cover the rendering side.
Frequently asked questions
When should I use a dual axis chart?
When two series have genuinely different units and both levels matter to the reader. Keep it to two series, use different chart types, label both axes, and set the limits deliberately.
Is a dual axis chart misleading?
It can be. The apparent relationship between the two series depends on the axis ranges you choose. The same data can be made to look correlated or unrelated.
What is the alternative to a secondary axis?
Index both series to 100 at a common starting period, or stack two panels that share the time axis. A scatter plotting one metric against the other is the third option. Each answers a different question.
How do I compare growth rates of two different metrics?
Index both to a common base and plot them on one axis. Every value becomes a percentage of its own starting point, which makes the growth rates directly comparable.
Should a chart axis always start at zero?
For bar and column charts, yes, because the bar length encodes the value. For line charts a truncated axis is sometimes reasonable, provided the truncation is visible and stated.