Overview
A segmented bar chart is a bar chart in which each bar represents a whole—usually 100%—and is divided into colored segments showing the share of each category within that whole. It is the right choice when you need to compare composition across groups, and the wrong choice when readers need exact counts or absolute totals, because normalizing every bar to the same height hides differences in group size.
What a segmented bar chart shows
A segmented bar chart (often called a segmented bar graph) displays categorical data by splitting each bar into parts, where every part represents one category’s contribution to the group’s total. As Littlegate Publishing puts it, a segmented bar graph is a graphical representation of data in which the bars are divided into segments, and it can act as a relative frequency table in visual form.
Picture a survey asking employees in three departments whether they prefer email, chat, or phone. Each department gets one bar, each bar reaches the same full height, and the segments inside show what fraction of that department chose each option. ThoughtSpot’s bar chart guide describes this pattern directly: each category’s bar represents 100% but is split by the percentage of respondents preferring each option. Reading the chart means comparing segment sizes within a bar and across bars, not comparing bar heights.
The short answer: percentages, not raw totals
Segmented bar charts typically normalize every bar to the same total. Statistics How To defines it as one kind of stacked bar chart where each bar shows 100% of the discrete value. That normalization is the chart’s strength and its main risk: it makes composition easy to compare, but two bars of identical height can represent very different underlying totals—one group might have 400 responses and another only 25, and the chart alone will not tell you. Keep that trade-off in mind throughout this article; it drives most of the decision guidance below.
Segmented bar chart vs. stacked bar chart vs. clustered bar chart
Readers searching for a segmented bar chart often actually need one of its neighbors, so it helps to see the options side by side. The core distinction: segmented bars show proportions of a whole, regular stacked bars show composition plus total magnitude, and clustered bars show individual values side by side. The table below summarizes when each chart type—and a few adjacent alternatives—fits best.
| Chart type | What each bar/mark shows | Best when you need | Weakness |
|---|---|---|---|
| Segmented (100% stacked) bar | Percent composition of a group | Comparing composition across groups | Hides absolute totals and group sizes |
| Regular stacked bar | Counts stacked to the group total | Composition and total magnitude together | Middle segments are hard to compare precisely |
| Clustered (grouped) bar | One bar per category per group | Direct side-by-side value comparison | No part-to-whole reading; gets crowded fast |
| Pie chart | Shares of a single whole | One group, few categories | Poor for comparing multiple groups |
| Mosaic / Marimekko | Shares plus group-size width | Composition and relative group size at once | Less familiar to general audiences |
| Simple table | Exact values | Precise numbers and small differences | No visual pattern recognition |
Use the table as a first filter, then read the notes below for the judgment calls that a matrix cannot capture.
Segmented bar chart
Choose a segmented bar chart when the analytical question is about relative composition: “what fraction of each group falls into each category?” ThoughtSpot notes segmented bar charts are best when you’re trying to show the breakdown of a certain category as part of the whole. Because every bar spans the same range, differences in mix jump out even when group sizes differ—provided you disclose those group sizes somewhere on the chart.
Regular stacked bar chart
A regular stacked bar chart stacks raw counts or values, so bar heights differ and the total for each group stays visible. That makes it the better option when readers need both the mix and the magnitude—for example, revenue by product line per region, where a small region with a distinctive mix should still look small. The cost is that comparing proportions across groups becomes harder, because segments start at different heights in different bars.
Clustered bar chart
Clustered (grouped) bars place each category’s bar next to the others within a group rather than stacking them. This is the stronger choice when the reader’s question is “which category is bigger, and by how much?”—every bar starts at zero, so lengths are directly comparable. Reach for clustered bars when exact value comparison matters more than part-to-whole framing; reach for segmented bars when composition is the story.
Pie chart, mosaic plot, and Marimekko chart
A pie chart handles a single whole with a small number of slices, but it compares poorly across multiple groups—a row of pies is usually harder to read than one segmented bar chart with a shared axis. A mosaic plot extends the segmented idea by also varying bar widths, which helps show association between two categorical variables along with group sizes. Marimekko-style charts do something similar in business reporting, encoding both share and total size. These alternatives are worth knowing, but for most composition-across-groups questions, the segmented bar chart remains the most familiar and legible option.
When to use a segmented bar chart
The decision comes down to one question: does your audience need composition or magnitude? If the answer is composition—how the mix differs between groups—a segmented bar chart fits. If the answer is magnitude, or both, pick a different chart from the comparison above.
A quick fit check before you build one:
- Your data is categorical, with mutually exclusive segment categories
- The question is about shares or mix, not absolute volume
- You have a manageable number of segments (roughly six or fewer reads best)
- You can disclose group totals or n-values alongside the bars
If most of these hold, the segmented format will serve you well; the two subsections below show the boundary cases.
Good fit: comparing composition across groups
Segmented bars excel at comparing conditional distributions—the distribution of one categorical variable within each level of another. Statistics instruction commonly uses multiple segmented bar charts to compare conditional distributions and to look for association between categorical variables: if the segment pattern looks similar across bars, the two variables show little association; if the pattern shifts noticeably, they are related. Practical examples include survey responses by department, product mix by region, and budget share by cost category.
Poor fit: comparing exact counts or many small segments
When the audience needs to know that Region A sold 12,000 units and Region B sold 3,000, a segmented bar chart actively obscures the answer, since both bars look identical in height. It also degrades when there are many small segments: thin slices become indistinguishable, and readers cannot compare them across bars. In those cases, a clustered bar chart, a regular stacked bar chart, or simply a table of counts and percentages will communicate more honestly. When precision matters—say, distinguishing 4% from 6%—a table beats visual estimation.
Data requirements for a segmented bar chart
Before opening any charting tool, make sure your data can actually support the chart. A segmented bar chart needs two categorical variables and one numeric measure, plus a derived percentage.
Categories, segments, totals, and percentages
At minimum, prepare these fields:
- Group category — the variable that defines each bar (e.g., department, region, year)
- Segment category — the variable that splits each bar (e.g., response option, product line)
- Segment count or value — the raw number for each group–segment combination
- Group total — the sum of all segment values within each group
- Segment percentage — segment value ÷ group total × 100
Segment categories must be mutually exclusive within a group, so that percentages sum to 100. If a respondent or record can belong to multiple segments, decide up front how to handle it—either restructure the categories or use a different chart, because overlapping categories break the part-to-whole logic.
Unequal group sizes, missing values, and zero values
Three edge cases deserve attention before you chart. First, unequal group sizes: the chart will render a group of 20 and a group of 2,000 as equally tall bars, so plan to annotate each bar with its total or n-value. Second, missing values: decide whether missing responses become their own segment (often the honest choice for surveys) or are excluded from the group total—and state which you did, because the percentages change either way. Third, zero values: a category with zero observations simply produces no visible segment, which is fine, but keep it in the legend if it appears in other groups so colors stay aligned across bars. None of these are reasons to avoid the chart; they are reasons to prepare the data deliberately.
Worked example: from raw counts to percentages
The fastest way to understand a segmented bar chart is to build one on paper. This worked example takes a small survey dataset from raw counts to percentages to interpretation, with no software required.
Step 1: Start with a frequency table
Suppose two teams answered a one-question satisfaction survey with three options: Satisfied, Neutral, and Dissatisfied. Team A returned 40 responses: 24 Satisfied, 10 Neutral, 6 Dissatisfied. Team B returned 10 responses: 3 Satisfied, 3 Neutral, 4 Dissatisfied. Note the constraint already visible in the raw data: Team A has four times as many respondents as Team B, which the finished chart will not show unless we label it.
Step 2: Convert counts to percentages
The formula is: segment percentage = segment count ÷ group total × 100. Applying it to each cell:
- Team A Satisfied: 24 ÷ 40 × 100 = 60%
- Team A Neutral: 10 ÷ 40 × 100 = 25%
- Team A Dissatisfied: 6 ÷ 40 × 100 = 15%
- Team B Satisfied: 3 ÷ 10 × 100 = 30%
- Team B Neutral: 3 ÷ 10 × 100 = 30%
- Team B Dissatisfied: 4 ÷ 10 × 100 = 40%
Check that each group’s percentages sum to 100 (60 + 25 + 15 = 100; 30 + 30 + 40 = 100). If rounding pushes a group to 99% or 101%, either display one decimal place or note the rounding.
Step 3: Read the finished chart
Plotted as a segmented bar chart, Team A’s bar is 60% Satisfied while Team B’s bar is 40% Dissatisfied—a striking difference in composition. The chart supports the claim “a larger share of Team B reported dissatisfaction.” It does not support “more people on Team B are dissatisfied”: Team B’s 40% Dissatisfied is 4 people, fewer than Team A’s 15%, which is 6 people. It also cannot tell you why the teams differ—composition shows association at most, never causation. Labeling each bar “n=40” and “n=10” keeps readers from over-trusting the smaller group’s percentages.
How to make a segmented bar chart in common tools
Most software calls this chart a “100% stacked” bar or column chart rather than a segmented bar chart, so knowing the naming saves time. The general workflow is the same everywhere: arrange groups as rows, segment categories as columns, and let the tool stack and normalize.
Excel
In Excel, arrange your data with group names in the first column and one column per segment category, then insert a chart and choose the 100% stacked column (vertical) or 100% stacked bar (horizontal) option. Excel normalizes each bar to 100% automatically, so you can chart raw counts directly. Exact menu labels vary by version, so a practical checklist:
- Lay out groups as rows, segment categories as columns
- Select the range including headers
- Insert → chart → 100% stacked bar or column
- Add data labels and, ideally, group totals in the axis labels or title
If you have already converted counts to percentages by hand, a regular stacked chart of those percentages produces the same visual.
Google Sheets
Google Sheets follows the same pattern: prepare the table with groups in rows and segments in columns, select it, and insert a chart. In the chart editor, choose a stacked bar or column type and set stacking to 100% where the option is available. As in Excel, verify that segment colors and order match any related charts in the same report, and add n-values to the group labels when sample sizes differ.
BI and programming tools
Dashboard and programming tools map the same three ingredients to their own vocabulary: the group variable goes on one axis, the segment variable becomes the color or series encoding, and the value is either a pre-computed percentage or a count with a “100% stacked” or “normalize” setting enabled. If your tool lacks a normalized-stacking option, compute the percentages in your data preparation step and stack those instead. Whichever tool you use, the data-preparation logic from the worked example above stays identical—the software only automates the drawing.
How to read a segmented bar chart accurately
Making the chart is half the job; reading it without over-interpreting is the other half. Three habits prevent the most common misreads.
Compare outer segments first
Segments anchored to the bottom or top of each bar share a common baseline, which makes their lengths directly comparable across bars. Segments floating in the middle start at different heights in different bars, so judging their relative size is genuinely harder. When your message hinges on one particular category, place it at an edge of the stack—or switch to a clustered bar chart where every value gets its own baseline.
Check the base size behind each 100% bar
Because every bar reaches the same height, a segmented bar chart visually equates groups of very different sizes. In the worked example above, Team B’s bar looks exactly as substantial as Team A’s despite representing a quarter of the responses. Before drawing conclusions, look for n-values, group totals, or a companion chart of counts; if none are shown, treat cross-group comparisons as provisional. As a chart maker, always disclose base sizes—an “n=” label per bar costs nothing and prevents the most serious misreading this chart type invites.
Use labels when small differences matter
Human perception estimates segment lengths only roughly, and a few percentage points of difference can disappear visually. When the decision depends on whether a segment is 12% or 16%, add percent labels to the segments or pair the chart with a small table of exact values. Visual charts are for pattern recognition; labels and tables are for precision—use both when the stakes justify it.
Common mistakes and how to avoid them
Most segmented bar chart failures trace back to a handful of avoidable mistakes. This section covers the three that cause the most damage in reports and dashboards.
Mistaking percentages for counts
The most consequential error is reading segment size as volume. A segment covering 40% of a small group’s bar can represent fewer people than a segment covering 15% of a large group’s bar—exactly what happened in the worked example. The fix is twofold: chart makers should label group totals on or near each bar, and chart readers should ask “percent of what?” before comparing. If absolute volume is the real question, switch to a regular stacked or clustered bar chart of counts.
Using too many segments
Beyond roughly five or six segment categories, slices become thin ribbons that readers cannot distinguish or track across bars. The chart stops communicating and starts decorating. Practical remedies include grouping minor categories into an “Other” segment, filtering to the categories that matter for the question, splitting the chart into small multiples, or falling back to a table. If a rare category is substantively important despite being visually tiny—rare but severe incidents, for instance—call it out with a label or a dedicated companion chart rather than hoping readers notice a sliver.
Changing colors or segment order
When the same categories appear in multiple charts across a report or dashboard, each category should keep the same color everywhere; reusing a color with a different meaning breaks cross-chart comparison and can invert the perceived trend. Segment order is equally deliberate: order ordinal categories along their natural scale (Very dissatisfied through Very satisfied), lifecycle categories by stage, and unordered categories by overall magnitude. Alphabetical or data-entry order is a default, not a decision—and defaults rarely match the analytical question.
Examples of segmented bar chart use cases
Abstract rules stick better with concrete scenarios. Here are three contexts where segmented bar charts routinely earn their place.
Survey responses
Survey analysis is the classic use case: response distributions compared across departments, cohorts, regions, or time periods. Segmented bars make it easy to see, for example, that dissatisfaction concentrates in one team even when team sizes differ—provided n-values are disclosed. For ordinal scales like satisfaction, keep segments in scale order so the chart reads as a gradient from negative to positive rather than a shuffled mosaic.
Product mix or budget allocation
Business reporting leans on segmented bars whenever relative share matters more than raw volume: product mix by region, budget share by cost category, or risk mix by portfolio. A portfolio manager watching whether one product line’s share of revenue is growing cares about the mix shift itself, and a segmented bar chart per quarter shows that shift cleanly. When leadership also needs the total revenue trend, pair the segmented chart with a simple line or bar of totals rather than forcing one chart to do both jobs.
Published datasets and exploratory pages
Segmented-style composition views also appear when datasets are published for open exploration rather than a fixed report. TablePage, for example, lets you drag and drop CSV, TSV, XLSX, or XLS files and instantly generates a public dataset page where anyone can explore charts, insights, and a filterable table—no signup needed. Its homepage demo of global fuel pricing includes a “Pricing regime by region” composition chart labeled “% · n=150,” a useful model of the base-size disclosure this article recommends. Published example pages such as the Electoral Democracy Index (1789–2025) show how a shared interactive page lets readers verify percentages against the underlying filterable table—the same count-versus-percent check a static chart cannot offer.
Frequently asked questions about segmented bar charts
The questions below come up repeatedly once readers move from definition to practice; each answer summarizes guidance covered in more depth above.
Is a segmented bar chart the same as a 100% stacked bar chart?
Conceptually, yes, in most contexts: both describe bars normalized to a full value and divided into proportional segments, and Statistics How To defines the segmented bar chart as a kind of stacked bar chart where each bar shows 100%. The difference is mainly vocabulary. Statistics education favors “segmented bar chart” or “segmented bar graph,” while spreadsheet and BI software usually labels the chart type “100% stacked.” Naming is not perfectly universal, so when in doubt, check whether the bars are normalized to a common total—that property, not the label, is what defines the chart.
Should a segmented bar chart show counts or percentages?
Percentages are the default, because the chart’s purpose is composition comparison and normalization is what makes composition comparable across differently sized groups. Show counts instead—via a regular stacked or clustered bar chart—when absolute volume is part of the message, when group totals differ dramatically, or when the audience may not realize the bars are normalized. A strong middle path is percentages in the chart with group totals labeled per bar, which preserves composition reading while disclosing scale.
How many segments are too many?
There is no universal threshold, but readability degrades quickly beyond roughly five or six segments, because thin slices become impossible to compare across bars. The practical test: can a reader identify each segment, track it across every bar, and estimate its size without squinting? If not, group minor categories into “Other,” filter to the relevant few, split into small multiples, or present a table. The right number of segments is however many your audience can actually read—not however many your data contains.