15 Excel Chart Types and When to Use Each One

Most bad charts aren’t badly formatted — they’re the wrong chart type for the data. Excel offers dozens of options under Insert > Charts, and picking between them is a decision about what question your data answers, not about which one looks nicest. This guide covers 15 chart types with the specific situation each one is built for, and just as importantly, when to avoid it.

Quick Decision Table

If your data shows…	Use
Comparison across categories	Column or bar
Change over time	Line or area
Parts of a whole (few slices)	Pie
Relationship between two variables	Scatter
Two different scales or units	Combo
Distribution of values	Histogram or box plot
Running total with increases/decreases	Waterfall
Hierarchy by size	Treemap or sunburst
Stages in a process	Funnel
In-cell trend at a glance	Sparklines

Now the details.

The Core Four

1. Column Chart

Best for: Comparing values across a handful of categories — sales by region, headcount by department. The vertical bars make magnitude differences obvious. A clustered column chart handles a second grouping (sales by region and by quarter).

Don’t use it when: You have more than 10–12 categories (labels collide and the chart turns into a comb) or long category names. And resist the urge to start the value axis anywhere but zero — truncated axes exaggerate small differences.

2. Bar Chart

Best for: The same comparisons as a column chart, but with long category labels or many categories. Horizontal bars give labels room to breathe — survey responses, product names, country lists. Sort the bars by value before presenting; alphabetical order almost never tells the story.

Don’t use it when: The category axis is time. Time should run left to right, which means columns or a line, never sideways bars.

3. Line Chart

Best for: Trends over time with regularly spaced intervals — monthly revenue, weekly site traffic, daily temperatures. Excel’s line graph connects points in category order, which is exactly right for evenly spaced periods. Add a trendline when you want the underlying direction without the noise; it also works on scatter charts as a line of best fit.

Don’t use it when: Your x-values are numeric and unevenly spaced. A line chart treats categories as equal steps, so plotting measurements taken at day 1, 3, and 30 will silently distort spacing — that’s a job for a scatter chart with lines. Also avoid more than 4–5 series; spaghetti charts help no one.

4. Pie Chart

Best for: One snapshot of parts-of-a-whole with 2–5 slices where one or two dominate — market share, budget split. If you do build a pie chart, label slices with percentages directly rather than relying on a legend.

Don’t use it when: You have more than five slices, slices of similar size (humans are bad at comparing angles), or you’re comparing two points in time. Two pies side by side is the classic mistake — a bar chart shows the change far more clearly. And skip 3-D pies entirely; perspective distorts every slice.

Relationships and Composition

5. Scatter (XY) Chart

Best for: Testing whether two numeric variables move together — ad spend vs. revenue, temperature vs. sales. A scatterplot is the only chart in this list where both axes are true value axes. Pair it with a line of best fit to make the relationship explicit.

Don’t use it when: One axis is categories. If you find yourself scattering “Region” against revenue, you wanted a bar chart.

6. Area Chart

Best for: A trend over time where the cumulative magnitude matters — total volume, not just direction. The stacked variant shows how components add up to a moving total (traffic by channel, month over month).

Don’t use it when: Series overlap in the non-stacked version — the front series hides everything behind it. And in stacked area charts, only the bottom series and the total are easy to read; middle bands float on a shifting baseline, so don’t use one if readers need precise values for every series.

7. Combo Chart

Best for: Two related measures with different scales — units sold (thousands) against profit margin (percent). Use columns for one, a line on a secondary axis for the other. This is the standard layout for actuals-vs-rate dashboards.

Don’t use it when: Both series share a scale (just use two lines) or the two measures aren’t genuinely related. Dual axes invite readers to compare line crossings that mean nothing, so reserve combos for pairs with a real connection.

Distribution Charts

8. Histogram

Best for: Seeing the shape of one numeric variable — where exam scores, order values, or delivery times cluster, and whether the distribution skews. Excel 2016+ has a built-in histogram chart that bins values automatically, and you can adjust bin width in Format Axis.

Don’t use it when: You’re comparing categories (that’s a column chart — similar look, completely different meaning) or comparing distributions across several groups, where overlapping histograms get unreadable. For that, use the next chart.

9. Box Plot (Box & Whisker)

Best for: Comparing distributions across groups in one compact view — salaries by department, test scores by class. A box and whisker plot shows median, quartiles, and outliers side by side, which is far more honest than comparing averages alone.

Don’t use it when: Your audience hasn’t seen one before and you can’t explain it, or each group has only a handful of data points — quartiles of five values are nearly meaningless. It’s available natively in Excel 2016 and later.

Financial and Hierarchical Charts

10. Waterfall Chart

Best for: Showing how a starting value becomes an ending value through sequential pluses and minuses — revenue down to net profit, headcount bridges, budget variance. Build the waterfall chart, then double-click your subtotal columns and check “Set as Total” so they anchor to zero.

Don’t use it when: The order of contributions is arbitrary or there’s no meaningful running total. A waterfall implies sequence; if shuffling the bars wouldn’t matter, a sorted bar chart is more honest.

11. Treemap

Best for: Showing relative size across a hierarchy with many items — sales by category and subcategory, disk space by folder. Rectangles sized by value handle 20–50 items where a pie or bar collapses.

Don’t use it when: Precise comparison matters. Readers can’t accurately compare rectangle areas, especially across different corners of the chart. Treemaps answer “what dominates?”, not “is A 12% bigger than B?”. Negative values can’t be plotted at all.

12. Sunburst Chart

Best for: A hierarchy where the levels are the story — how a whole splits into branches and sub-branches, ring by ring. Better than a treemap at making parent-child structure visible.

Don’t use it when: You have only one level (that’s a pie chart wearing a costume) or you need values read precisely — outer-ring segments get thin and hard to compare fast. With more than three levels, labels become unreadable.

13. Funnel Chart

Best for: Sequential stages that shrink — sales pipeline, hiring stages, checkout drop-off. Available natively in Excel 2019/Microsoft 365 under Insert > Charts.

Don’t use it when: Stages aren’t strictly sequential or values don’t consistently decrease — a funnel that bulges in the middle confuses more than it explains. Also note the funnel shape itself encodes nothing; the centered bars are pure decoration, so a plain bar chart carries identical information if your Excel version lacks funnels.

14. Radar Chart

Best for: Comparing 2–3 items across 5–8 metrics rated on the same scale — skills assessments, product feature scorecards. The shape of each polygon gives a fast “profile” impression.

Don’t use it when: Metrics use different units or scales (the chart becomes meaningless), you have many items (overlapping polygons), or order matters — the axis sequence around the circle is arbitrary, yet it changes the shape dramatically. Most radar charts are better as bar charts; use radar only when the profile shape itself communicates something.

15. Sparklines

Best for: Trend-at-a-glance inside a table — one tiny line, column, or win/loss chart per row, living in a cell next to the numbers. Add sparklines from Insert > Sparklines, and use the same vertical axis settings across the group (Sparkline tab > Axis) so rows are comparable. They’re ideal for dashboard tables with dozens of rows where full charts would be absurd.

Don’t use it when: Readers need axis values, exact figures, or comparisons across more than a rough trend. Sparklines have no axes or labels by design — they supplement a table, they don’t replace a chart.

Two Habits That Outrank Any Chart Choice

First, you’re never locked in: select any chart and use Chart Design > Change Chart Type to swap types without rebuilding. Try two or three candidates against real data — the right one is usually obvious within seconds.

Second, when your category counts feed a “which problems matter most” question, consider the Pareto chart — a sorted histogram with a cumulative percentage line that Excel builds natively. It’s the fastest way to show that a few categories drive most of the total.

FAQ

What’s the difference between a histogram and a column chart?

A column chart compares separate categories (East vs. West); a histogram shows how one numeric variable is distributed across bins (how many orders fell between $0–50, $50–100, and so on). They look similar, but a histogram’s bars touch because the bins are continuous, and reordering them would destroy the meaning.

Which chart types require a newer version of Excel?

Histogram, box & whisker, waterfall, treemap, sunburst, and Pareto charts arrived in Excel 2016. Funnel charts require Excel 2019 or Microsoft 365. Everything else in this list — including sparklines (Excel 2010+) — works in any modern version.

Should I ever use a 3-D chart?

Almost never. The perspective distortion makes values nearly impossible to read accurately, and the depth axis adds no information for typical business data. If a flat chart looks plain, fix it with sorting, labeling, and color — not a third dimension.

How many data series is too many for one chart?

For line charts, four or five is the practical ceiling before it turns to spaghetti. Past that, either split into small multiples (several small charts with identical axes), gray out all but the one or two series that matter, or move the detail into a table with sparklines.