What is a matrix question?
A matrix question lists several items in rows and one answer scale in columns, so people rate each item on the same scale in one grid. It saves space when you want to compare the parts of one experience, such as checkout, delivery and support.
Many items, one scale
A matrix question, also called a grid question, groups related questions that share an answer scale. Each row is an item, such as Checkout, Delivery or Packaging, and each column is a point on the scale, such as Poor to Great or Strongly disagree to Strongly agree. People pick one answer per row.
Because every row uses the same scale, the answers line up and can be compared at a glance: which part of the experience scores lowest, and how each part changes over time. A matrix is also the usual home for a Likert scale, where several statements share the agreement points.
The trade-off is effort. A large grid looks like work, is hard to read on a small screen and tempts people to pick the same column for every row. Kept short, a matrix is one of the most efficient questions a survey can ask.
- Also called
- Grid question, table question
- Layout
- Items in rows, one shared answer scale in columns
- Best for
- Rating several parts of one experience on the same scale
- Read as
- An average per row, or a heat map of the answers
- Keep it to
- About five to seven rows per grid
Matrix question example
A good matrix asks about the parts of one thing, so the rows belong together and the scale fits every row. Keep row labels short, and label every column.
The columns can hold any scale that suits the rows. These are the most common:
- Quality: Poor, Fair, Okay, Good, Great
- Agreement: Strongly disagree to Strongly agree
- Satisfaction: Very dissatisfied to Very satisfied
- Frequency: Never, Rarely, Sometimes, Often, Always
When to use a matrix question
Use a matrix when you want the same judgment on several related items and plan to compare them. When the items need different scales, or one item matters far more than the rest, ask separate questions.
- Rating the parts of a stay, an order or an event
- Asking several agreement statements about one product
- Checking satisfaction with each feature or each team
- Repeating the same grid each quarter to follow every row
How to analyze matrix question results
Read a matrix one row at a time. Give each column a number, from 1 for the lowest point upward, and work out an average per row. Then compare the rows to find the weakest part of the experience, and follow each row over time.
Say 100 customers rate Delivery from 1 (Poor) to 5 (Great). The points add up to 370, so Delivery averages 3.7. If Checkout averages 4.4 on the same grid, delivery is the place to look first.
A heat map, with darker cells where more answers fall, shows the same thing at a glance and makes a split row easy to spot.
- Poor (1)
- 5 answers, 5 points
- Fair (2)
- 10 answers, 20 points
- Okay (3)
- 20 answers, 60 points
- Good (4)
- 40 answers, 160 points
- Great (5)
- 25 answers, 125 points
- Delivery average
- 370 ÷ 100 = 3.7
Matrix question vs separate questions
A grid is faster to answer when the list is short and every row fits the scale. Separate questions take more space but give each item full attention and its own scale.
On a phone, a wide grid doesn't fit, and squeezed columns are easy to mistap. Showing one row at a time with the full scale under it keeps a matrix readable on small screens.
| Matrix | Separate questions | |
|---|---|---|
| Space | One compact grid | One card per item |
| Speed | Fast for short lists | Slower, but each item gets attention |
| Comparison | Rows line up on one scale | Each item can have its own scale |
| On a phone | Best shown one row at a time | Fits as it is |
| Risk | The same answer down every row | The survey feels longer |
Matrix questions in Usertick
Add a Matrix question in the survey editor with up to 50 rows and 20 columns. You can put the rows in random order, allow several answers per row, and add a Not applicable column that stays out of the averages. The matrix shows as a table on wide cards and one row at a time on phones and in popups.
Results draws the answers as a heat map with each row's average, and the over time view follows the grid from period to period. Templates such as Feature satisfaction and the System Usability Scale start with a grid ready to edit, and on every plan, AI can suggest rows or columns the matrix doesn't have yet.
In an email, a matrix shows as a button that opens the survey on its page. Logic rules can check only whether a matrix was answered, so put a choice or rating question first when the next step depends on the answer.
Mistakes to avoid with matrix questions
Small choices that keep the results useful.
- 01
Too many rows in one grid
Past seven or eight rows, people tire and pick the same column down the page. Keep the grid short, or split it in two.
- 02
Rows that need different scales
“Price” and “Delivery time” don't both fit Poor to Great. Group the rows that share a scale and ask the rest on their own.
- 03
No way to say not applicable
Someone who never called support has to guess on a Support row. Add a Not applicable column so guesses stay out of the average.
- 04
Long column labels
Wide labels crowd the grid and wrap on phones. Use one or two words per column.
Frequently asked questions
Short ones work well. Problems start with long grids, where people tire and answer every row the same way, so keep to about five to seven rows on a clear scale.
Straightlining means choosing the same column for every row without reading the rows. It is a sign the grid is too long, and a row or two worded the other way round helps you spot it.
Yes, when each row asks people to tick every column that applies, such as the devices they use each product on. In Usertick, turn on Several answers per row.
Show one row at a time with the full scale under it, as Usertick does on phones and in popups. Keep column labels short so the scale stays readable.
Listen better, act faster, and build what truly matters.
Start free, see what your customers mean, and get each answer to the person who can act on it.
- 150 responses a month on Free
- Every channel and AI analysis
- No card needed