GloboPoll International

GloboPoll blog · News

Matrix questions: several rows rated on the same scale

Questionnaires gain the matrix: a table with one row per item and one column per score, answered at a glance and read by comparing.

· Cibernatural · 3 min read · Updated on 2026-10-11

A week ago we opened the questionnaire: several questions of different types in one poll. We wrote down what was missing, and the first item on the list was the matrix. It is here today.

What it is

A matrix is the classic satisfaction-survey question: "Rate each item from 1 to 5", followed by a table with one row per item (price, service, cleanliness, speed) and one column per score. Voters tick one box per row and are done in seconds; whoever reads the results sees at a glance which item scores best and which worst.

In GloboPoll it is not a new question type, and that is deliberate. Each row of the matrix is a rating question like the ones that already existed, with its average, its median and its histogram. What the matrix adds is a shared title and one rule: the rows that share it are shown together, as a table, and must use the same scale. So voting, counting, the final result fixed at close, exports and the API stay exactly as they were.

How to create one

The questionnaire editor now has an "Add matrix" button next to "Add question". It creates a card with the matrix title, the scale (numeric, stars or NPS, with minimum, maximum and the labels of both ends) and two rows to start with; "Add row" adds as many as you need. Everything shared, the title, the scale and whether it is required, is changed once and applies to every row.

A standalone rating also becomes a matrix: just give it a title in the "Matrix title" field that appears under its scale, and from then on it accepts rows.

Each row counts as one question of the questionnaire for plan purposes (Pro allows ten per questionnaire). The scale locks as soon as there are answers, as with any rating.

How it is voted and read

On the poll page the matrix is a table with the scores in the header and one radio button per cell. Each row says on its own whether it is optional, and unanswered required rows block the next page, like every other question. On a phone the table scrolls sideways if it does not fit.

In the results, both on the public page and in the dashboard, the matrix shows one row per item with its average as a bar and as a figure, plus how many people answered it. In an NPS matrix the figure is each row's NPS, with its sign. That comparative reading is what a matrix is for: "price pleases less than service". Each row's histogram is still in the export, where every row appears with its matrix in front ("Rate each item · Price") so it never loses its context.

What is left

From the advanced questionnaire, long-text questions grouped by topic and conditional logic are still pending. They come in that order.

If you are not sure which poll type fits, Nine poll types: which one to use, and when covers each one with an example, how it is counted and which plan includes it.

Other languages