5 point vs 7 point Likert scale

A 5 point Likert scale is the better default for most surveys: every point can carry a plain word label, it fits a phone screen, and people answer it quickly. A 7 point scale gives slightly finer distinctions and suits audiences who will think about each answer, such as an academic study. In the 280 finished forms on this site, five points and four points are used almost equally, and seven points appears once.

What was counted

The set is the 280 form examples on this site, read on 21 October 2026. Rating grids store their column labels in each form’s source definition, so a script read all 59 grids and set aside 13 that are not rating scales at all (shift times, dietary needs, yes or no consent per item), leaving 46 grids in 31 forms. Single choice questions were read from the entries, and the 24 that are rating scales were picked out by hand: satisfaction, quality, agreement, frequency and likelihood to recommend. An opt out column such as Not seen was counted separately and not as a scale point.

The forms were written for fictional businesses, so the counts show what careful form writers chose, not which scale produces better data. The research below covers that.

The count

Five point scales are the most common, 26 of 70. They are mostly agreement grids, such as the employee engagement survey’s five grids from Strongly disagree to Strongly agree, frequency grids from Never to Always, and five level satisfaction questions.

Four point scales are close behind at 25. They are quality ratings with no middle, Poor, Fair, Good, Excellent, and frequency grids such as the 360 feedback form’s Rarely, Sometimes, Often, Almost always with a Not seen column. Leaving out the middle is deliberate on quality ratings: a session was either good or it was not.

Three point scales, 12 of them, are mostly checklists that rate a condition (Good, Fair, Needs attention on a property inspection) and bipolar questions such as Too slow, About right, Too fast.

Six questions use 0 to 10: five likelihood to recommend questions and one pain rating on a physiotherapy intake.

And one question uses seven points, the customer effort score survey, which asks how far the customer agrees that the company made it easy to handle their issue, from 1 strongly disagree to 7 strongly agree. That is the customer effort score’s standard form, and it is the only reason seven points appears at all.

What the research says

• More points help, up to a point. Preston and Colman’s 2000 study in Acta Psychologica compared scales from two to eleven points and found the shortest scales performed worst on reliability, validity and discriminating power, with improvements tailing off around seven to ten points. Respondents rated the longer scales as better for expressing their feelings, but not as quicker to use.
• Labels matter. The Nielsen Norman Group describes Likert items as typically five points, sometimes seven or nine, and notes that word labels are easier to answer than numbers but hard to find for every point beyond five.
• Say what a neutral answer means. The same Nielsen Norman Group article suggests a separate Not applicable option so that people with no view are not mixed with people the question does not fit, and Pew Research Center’s questionnaire design guidance stresses that the way a question is worded and ordered changes the answers it gets.

The practical reading: five points fully labelled is a good default; seven points is worth it when you need to detect small shifts over time in a large sample and your respondents will read each label.

Choosing your scale

1. Agreement or satisfaction, general audience, phone first: 5 points, every point labelled.
2. Quality of a specific thing: 4 points with no middle, and a Not applicable column if some people did not use it.
3. Customer effort score: 7 points, because that is the standard version and comparisons rely on it.
4. Likelihood to recommend: 0 to 10; the NPS survey best practices post covers it.
5. One scale per survey for questions you intend to compare.

The Likert scale page defines the format, the neutral option page covers the midpoint, and the matrix question page covers grids.

In the product

A rating grid in the form builder is a matrix with your own rows and columns, one answer per row or several, available in classic and conversational forms; a required grid counts as answered once any one row has an answer. A single scale question is a single choice field with the scale points as options. Responses export to CSV with each grid in one column. The rating matrix page shows the grid, and the guide to creating a conversational form with AI shows how to ask for one in a prompt.