← Docs

Survey question types

Five question types exist for surveys specifically. Each stores a structured answer rather than a line of text, which is what lets the results page chart it and the export keep it readable.

The five types at a glance

TypeUse it toStored asIn the CSVOn the results page
Matrix gridRate several items on one shared scale.One answer per row: { quality: "good", support: "poor" }. With “several answers per row” on, each row holds a list.Quality: Good; Support: PoorA stacked bar per row.
RankingPut options in order of preference.The option ids in rank order: ["price", "speed", "support"].1. Price; 2. Speed; 3. SupportOptions sorted by average rank, with first-place counts.
Net Promoter ScoreThe 0–10 likelihood-to-recommend question.A number from 0 to 10.9The NPS, the promoter / passive / detractor split and a bar per score.
Likert scaleOne statement on a labelled scale.The id of the chosen label: "agree".AgreeA bar per label and the average position on the scale.
Constant sumDistribute a fixed number of points.Points per option: { speed: 70, price: 30 }.Speed: 70; Price: 30Total and average points per option.

All five are in the editor’s question picker (press /), alongside the standard field types.

Matrix grid

A matrix has rows (the items being rated) and columns (the scale). Select the question in the editor to edit both lists. By default each row takes one answer; turn on Allow several answers per row to make each cell a checkbox.

On a narrow screen the table scrolls sideways inside its own frame rather than shrinking, so keep the scale to five columns or fewer and the labels short.

Ranking

Respondents reorder the list by dragging, with the up and down buttons on each item, or with the arrow keys once an item has focus — so it works on a phone and without a mouse.

An untouched ranking is not an answer

If a respondent never moves anything, nothing is stored — the starting order is not silently recorded as their preference. Someone who agrees with the order shown confirms it with Keep this order. A required ranking therefore means the respondent actually ranked.

Results show each option’s average rank (lower is better) and how many people ranked it first. Votense does not run an instant-runoff count; every ballot is in the CSV as a full ordered list if you need to count it another way.

Net Promoter Score

Eleven buttons from 0 to 10 with a label at each end, both of which you can edit. The results page calculates the score as the percentage of promoters (9–10) minus the percentage of detractors (0–6); passives (7–8) count in the total but in neither group.

Likert scale

One statement and a row of labelled answers. It starts with the five-point agreement scale, and you can change the labels or the number of points — a seven-point scale is seven options. The results average the position on the scale, counting the first label as 1.

Constant sum

The respondent distributes a fixed total — 100 by default — across the options, with a running “remaining” figure. The answer is accepted only when the points add up to the total exactly; an untouched question is allowed when it is not required.

Ranking or points?

Ranking gives you the order. Points give you the distance: “A then B” does not say whether B was nearly as wanted as A. Use constant sum when the size of the gap matters, and keep it to about seven options — dividing 100 across fifteen becomes arithmetic.

Common questions

What is the difference between a Likert scale and a matrix?
A Likert scale is one statement with one row of labelled answers. A matrix is several statements sharing the same scale in one table. Use a matrix when three or more statements use identical answers; use separate Likert questions when the scales differ.
How are these answers exported?
Each question is one CSV column and each answer one cell. A matrix is written as “Row: Answer” pairs separated by semicolons, a ranking as a numbered list in rank order, and a constant sum as “Option: points” pairs. Rows and options always appear in the order you defined them.
Does a required matrix need every row answered?
Yes. A required matrix is refused with “Please answer every row” until each row has an answer. Leave it not required if some rows will not apply to everyone.
Can I change the options after responses have arrived?
Relabelling an option is safe: answers are stored under the option’s value, not its label, so existing answers follow the new label. Deleting an option hides the votes it received, and changing an option’s value detaches them. Settle the options before sharing.
Are these types available on the free plan?
Yes, all five, with their charts and their export.

Keep reading