Cross-tabulation (a "crosstab") is a way of analyzing survey data that shows how answers to one question break down across the categories of another question. Instead of looking at a single question in isolation, you compare two variables at once in a grid, which reveals relationships that an overall total would hide.
What cross-tabulation is
A crosstab puts the categories of one variable in the rows and the categories of another in the columns, then fills each cell with a count or percentage. One question is the banner (the groups you compare, often a demographic like age or region) and the other is the stub (the question whose answers you are breaking down). Reading across rows or down columns lets you spot where groups differ.
Crosstabs are usually shown as percentages within each group rather than raw counts, because groups are rarely the same size. A column percentage answers "of the people in this group, what share chose this answer?".
Example cross-tab table
Suppose you asked which plan respondents prefer, and you cross-tabulate by age group. The cells show the percentage within each age group:
| Preferred plan | 18-34 | 35-54 | 55+ | |---|---|---|---| | Basic | 30% | 45% | 60% | | Pro | 50% | 40% | 30% | | Enterprise | 20% | 15% | 10% | | Total | 100% | 100% | 100% |
The overall average might say "Pro is most popular," but the crosstab reveals that preference flips with age: younger respondents lean Pro, older respondents lean Basic. That is the insight a single total would bury.
Why it matters
Cross-tabulation turns a flat result into a comparison. It helps you:
- find which segments drive an overall number,
- detect relationships between two questions (for example satisfaction by region),
- tailor decisions to specific groups instead of the average customer.
The main caution is sample size: when you slice a survey into many cells, each cell can get small fast, so differences in thinly populated cells may be noise. Confirm that any standout cell has enough responses behind it before acting. See statistical significance in surveys for how to judge that.
How to do this in SegmentOS
SegmentOS has a built-in Segment & Cross-Tabulate tool on the results page. You pick a question to Segment by (your banner, for example age or gender) and a question to Compare (your stub), and SegmentOS draws a grouped bar chart where each bar is the percentage of respondents within that segment who chose each answer. It also shows each segment's size so you can judge reliability at a glance. To go deeper, use segmentation to focus on one group, and let notable differences auto-surface where a segment diverges most from everyone else.
Frequently asked questions
What is the difference between a crosstab and segmentation?
A crosstab compares two questions side by side in a single grid, showing how one breaks down across the categories of the other. Segmentation typically means filtering your whole results view to a single group. They are related, and SegmentOS combines both in its Segment & Cross-Tabulate tool.
Should crosstab cells show counts or percentages?
Usually percentages, calculated within each group (column), because groups are rarely equal in size. Percentages let you compare groups fairly; counts are still useful for checking whether a cell has enough responses to trust.
Why do crosstab columns add up to 100%?
When percentages are calculated within each group (column), every respondent in that group falls into exactly one answer category, so the column sums to 100%. This is called column percentaging and is the most common way to read a crosstab.
How many responses do I need per crosstab cell?
There is no hard rule, but small cells are unreliable. Treat cells with fewer than about 30 responses as directional only, and be skeptical of dramatic differences that rest on just a handful of people.