PuckSurveys

The Likert scale alternative: let "somewhat" be a real answer.

Keep the anchors everyone understands. Replace the five boxes with a puck that slides from 0 to 100, and see the gradient your boxes were collapsing.

The five boxes, briefly

The Likert scale has been the default since the 1930s: a statement, then five or seven labeled steps from strongly disagree to strongly agree. It is quick, familiar, and easy to tabulate. It also has a known cost. Everyone between "agree" and "strongly agree" becomes the same person. The respondent who hesitated for five seconds and the one who slammed the same box in half a second look identical in your data. And when every question offers the same five boxes, straight-lining through them is effortless and invisible.

Same question, continuous answer

Keep your anchors.

Strongly disagree on the left, strongly agree on the right. Your question wording does not change.

Gain the gradient.

Respondents slide a puck to exactly where they stand, 0 to 100. A 58 and an 84 stop being the same "agree."

See the conviction.

The puck records how each answer happened: hesitation, direction changes, and how sure the final position looked. Boxes cannot see any of that.

When boxes are fine

If you only need a yes-or-no pulse, or you must match a legacy instrument box for box, a Likert scale is fine. And 0–100 data can always be reported in Likert-style buckets afterward, so the safest default is to collect the detail and choose your reporting later. The reverse is not available: once an answer lands in a box, the detail is gone.

Try it on your own questions

Paste in your Likert items, keep the anchors, and send one link. Most respondents finish a ten-question survey in about 30 seconds.

Make your own (free) →

Free to use. Questions: pucks@fastmail.com

Common questions

What can I use instead of a Likert scale?

A continuous 0–100 scale with the same anchors. Respondents slide a puck between your two labels instead of picking one of five boxes, so you keep the familiar wording and gain the detail between the steps.

Do respondents take longer on a 0–100 scale?

Not meaningfully. Answering by sliding a puck takes about 3 seconds per question, and most ten-question surveys finish in about 30 seconds.

Can I still report results in five groups?

Yes. Collect on 0–100 and bucket the answers into any grouping afterward. Grouping detailed data is easy; recovering detail from grouped data is impossible.

Can I use agree and disagree anchors in PuckSurveys?

Yes. Every question runs between any two anchors you choose, so classic Likert wording carries straight over.

Is PuckSurveys free?

PuckSurveys is currently free: unlimited surveys, unlimited responses, full signals. No card.

All rating scales →