How Do You Make a Side-by-Side Bar Chart in Power BI?

You have two numbers per category and you want people to compare them at a glance. Last year against this year. Paid against organic. Desktop against mobile. The chart that does this is the one most people call a side-by-side bar chart, and Power BI does build it — it just does not call it that.

The name in the visualisations pane is Clustered bar chart, or Clustered column chart if you want the bars standing up rather than lying down. Once you know that, the build takes about ninety seconds. The part that takes longer is getting your data shaped so Power BI puts the right things side by side, which is where most people get stuck and end up with one bar instead of two.

This guide covers the build, the data shape behind it, how to read the result honestly, and the cases where a side-by-side chart is the wrong choice even though it looks right.

The short answer

In Power BI, a side-by-side bar chart is the Clustered bar chart visual. Drop your category into the Y-axis field well, then drop two or more measures into the X-axis field well. Power BI draws one bar per measure, grouped together under each category label.

If you want vertical bars instead, use Clustered column chart and swap the axes: category goes on the X-axis, measures on the Y-axis. Everything else in this guide applies identically.

The alternative route, when your comparison lives in a single column rather than in two separate measures, is to put one measure in the value well and the comparison field into Legend. Power BI then splits that one measure into a bar per legend value. Which of the two routes you need depends entirely on how your table is laid out, which is the next section.

Not sure whether your reporting is telling you the right story? Get a free SEO audit and see what your data is actually saying about your traffic.

What a side-by-side bar chart actually is

A side-by-side bar chart — also called a clustered, grouped or double bar chart — shows two or more series against the same set of categories, with each series getting its own bar inside a shared group.

The defining feature is that every bar starts from the same baseline. That is what makes the comparison honest: your eye is judging bar lengths that all begin at zero, which is the one visual comparison humans are reliably good at.

How it differs from a stacked bar chart

A stacked bar chart puts the series on top of each other inside a single bar. The total is easy to read, but the individual segments are not — only the bottom segment starts at zero, so every segment above it is floating at an arbitrary offset and your eye cannot compare it fairly to the segment beside it.

So the rule is simple. Compare the parts, use side-by-side. Compare the totals, use stacked. If you genuinely need both at once, you usually need two charts rather than one clever one.

How it differs from a single bar chart

A plain bar chart has one series. Adding a second series is what turns it into a clustered chart, and it is also what starts eating your horizontal space — every extra series multiplies the number of bars on screen. Two series is comfortable, three is fine, four is pushing it, and beyond that the chart stops being readable.

How to build one in Power BI, step by step

These steps assume Power BI Desktop with a table already loaded. The web service works the same way.

1. Open the report page where you want the chart and click an empty area of the canvas so no existing visual is selected.
2. In the Visualizations pane, click the Clustered bar chart icon. It is the one showing horizontal bars in pairs. For vertical bars, pick Clustered column chart instead.
3. An empty visual container appears on the canvas. Drag its corner handles now to roughly the size you want — clustered charts need more width than single-series ones, so give it room before you judge how it looks.
4. From the Data pane, drag the field you want to compare *across* into the Y-axis well. This is your category: month, channel, region, product, campaign name.
5. Drag your first number into the X-axis well. Power BI aggregates it — usually as a sum — and draws one bar per category.
6. Drag your second number into the same X-axis well, underneath the first. This is the step that creates the clustering. You should now see two bars per category.
7. Repeat for a third series if you need one. Stop at three or four.
8. Check the aggregation on each measure. Click the small arrow beside each field in the X-axis well and confirm it is doing what you expect — Sum and Average tell very different stories, and Power BI's default is not always the one you want.

The legend route, for long-format data

The seven steps above assume your comparison sits in two separate columns — a `Sessions 2025` column and a `Sessions 2026` column, say. Plenty of real tables are not shaped that way. Instead they have one `Sessions` column and a separate `Year` column with the value repeated down the rows.

For that shape, the build is different:

1. Put your category in the Y-axis well as before.
2. Put your single measure — `Sessions` — in the X-axis well.
3. Drag the splitting field — `Year` — into the Legend well.

Power BI now draws one bar per distinct legend value inside each category group. Same chart, different route in.

If you are not sure which shape you have, look at the table. Repeated values down a column mean long format and the legend route. One row per category with several number columns means wide format and the two-measures route.

Want to know which pages are actually earning your traffic before you chart any of it? Get a free SEO audit for a breakdown of what is working.

Making it readable

The default clustered chart is the one Power BI thinks you want before it knows what you are trying to say. It fits the data, and that is roughly where its ambition ends.

A clustered chart built with defaults is usually functional and rarely good. Four adjustments do most of the work.

Sort it deliberately. Click the three dots at the top right of the visual, choose Sort axis, and sort by whichever measure carries the point you are making. An unsorted clustered chart in arbitrary category order forces the reader to do the ranking themselves.

Turn on data labels, then turn most of them off. In the Format pane, switch on data labels. If the numbers crowd the bars, reduce the font size or show labels on only one series rather than removing them entirely.

Set the colours to mean something. Power BI's default palette assigns colours by position. If one series is "last year" and another is "this year", make last year grey and this year your accent colour. The reader should be able to tell which bar matters without reading the legend.

Label the axis, or drop it. If your data labels already show the numbers, the value axis is redundant and you can switch it off to reclaim space. If you keep it, make sure the units are obvious — an axis running to 45,000 with no indication of what is being counted is a question mark, not a chart.

Fix the axis. Check that the value axis starts at zero. A clustered bar chart with a truncated axis exaggerates small differences badly, and Power BI will sometimes truncate automatically when your values are clustered in a narrow range.

When to use it — and when not to

Use a side-by-side bar chart when you have a small number of categories, a small number of series, and the point you are making is a direct comparison between those series within each category. Period-over-period reporting is the classic fit. So is channel performance, A/B results, and before-and-after.

Do not use it when:

• You have more than about eight categories. The chart gets long and the groups blur together. Filter down, or use a small-multiples layout instead.
• You have more than four series. Too many bars per group and nobody can track which colour is which without staring at the legend.
• The story is a trend over many time periods. Twelve months of two series is twenty-four bars. A line chart with two lines says the same thing in a fraction of the space and shows the shape of the change, which bars hide.
• You care about the total, not the split. That is a stacked chart, or a single bar of the total with the breakdown somewhere else.
• Your series are on wildly different scales. Revenue in millions beside conversion rate as a percentage produces one visible bar and one invisible one. Use two charts, or a combo chart with a secondary axis.

The limitations worth knowing

None of these are bugs. They are consequences of what the chart type is, and knowing them in advance saves you from redesigning a report page twice.

Width is the hard constraint. Every series you add multiplies the bars on screen. This is the single most common reason a clustered chart stops working, and no amount of formatting fixes it — you have to reduce categories or series.

Comparison across groups is harder than within them. Readers compare the two bars inside one group easily. Comparing the second bar of group one to the second bar of group five, across the chart, is much harder. If that cross-group comparison is your actual point, sort by that series so the pattern becomes a shape rather than a hunt.

Missing data renders as absence, not as zero. If a category has no value for one series, Power BI simply draws no bar, and a missing bar looks identical to a zero bar. Handle blanks explicitly in your model if the difference matters.

The legend route inherits whatever is in your data. If your `Year` column has a stray value, a null, or an inconsistent format, you get a surprise extra bar in every group. Clean the column before you chart it.

Curious how competitors are performing on the keywords you are tracking? Get a free SEO audit and see where the gaps are.

Side-by-side vs the alternatives

vs stacked bar chart. Stacked wins when the total is the message and the split is context. Side-by-side wins when the split is the message. If somebody asks "how big was it?", stack it. If they ask "which was bigger?", cluster it.

vs line chart. Line wins for more than about six time periods, and for showing rate of change. Bars win for a handful of discrete periods and for categories that are not time at all.

vs small multiples. Power BI's small multiples feature splits one chart into a grid of miniatures, one per category. When you have too many categories for a clustered chart but still need the comparison, this is usually the better answer — each miniature keeps its own clean two-bar comparison.

vs a table. An underrated option. If your audience is going to read the exact numbers anyway, a well-formatted table with conditional formatting communicates more than a chart they have to squint at. Charts are for shape and comparison; tables are for values.

vs a 100% stacked bar chart. This variant normalises every bar to the same length and shows composition as a percentage. It is excellent for comparing mix across categories of very different sizes, and useless if the absolute numbers matter at all.

vs a combo chart. When one series is a count and another is a rate, a clustered chart fails and a line-and-clustered-column chart with a secondary axis works. Power BI has this visual built in.

Final thoughts

The side-by-side bar chart is the right default for comparing a few series across a few categories, and in Power BI the only real trick is knowing it is called a clustered chart and knowing which of the two build routes your data shape calls for. Two measures in the value well, or one measure plus a legend field.

Everything after that is restraint. The chart fails from too many categories, too many series, or an axis that does not start at zero — not from a missing formatting option.

If you are deciding between this and its closest relative, our guide to what a stacked bar chart is in Excel walks through the opposite case: when putting the series on top of each other genuinely is the better choice.

Ready to find out what your data is really telling you about your search performance? Get a free SEO audit and get a clear picture of where you stand.