Likert scale
A response scale with ordered points, from strongly disagree to strongly agree, that converts opinion into a comparable number.
Also known as: likert scale · likert scale 1 to 5 · likert agreement scale · likert
Try CRPRO free for 7 daysA Likert scale is the response scale on which a person marks their level of agreement with a statement, on ordered points from one extreme to the other — from "strongly disagree" to "strongly agree." It was proposed by psychologist Rensis Likert in 1932 and is the structure behind most of the satisfaction surveys you have ever answered.
A Likert item is not a question, it is a statement
This is the detail almost everyone gets wrong. Instead of "was support good?", you write "support solved my problem" and the customer picks a point on the scale. The statement needs to cover just one thing: "the agent was fast and polite" is impossible to answer for someone helped by someone fast and rude — and whatever answer comes back means nothing.
5 or 7 points, and what to do with the middle
| Points | When to use | Side effect |
|---|---|---|
| 3 | a quick survey, a busy audience | does not distinguish "good" from "great" |
| 5 | market standard, support | balance between nuance and clarity |
| 7 | academic or product research | the labels start to look alike and the choice becomes almost random |
| 4 or 6 | when you want to force a side | pushes the undecided toward a position they do not hold |
An odd number creates a neutral point in the middle. It welcomes people who genuinely have no opinion, but it also becomes a refuge for those who do not want to commit. An even number removes that exit, at the cost of inventing an opinion the respondent did not have.
Ready-made examples for support
- "It was easy to reach someone at the company."
- "The agent understood what I needed."
- "My problem was solved on the first contact."
- "The stated deadline was met."
Each of these statements becomes one line of the survey, always with the same scale and always in the same order of points. Changing the order midway through the questionnaire destroys the comparison.
Where the Likert scale is deceiving
- It is not an indicator. Likert is the ruler; CSAT and NPS are indices. Saying "we use Likert" does not say which number you track.
- The data is ordinal. You know "strongly agree" is more than "agree," but not that the distance between the two equals the distance between "neutral" and "agree." That is why the most defensible thing to report is the percentage of responses in the two positive points, called the top-2-box, and to use the average only to track a trend.
- Acquiescence bias. There is a natural tendency to agree with statements. Mixing positive and negative statements reduces the problem, but confuses some respondents — if you do this, say so in the instructions.
- A vague label measures nothing. "Support was satisfactory" produces generalized agreement; "support solved my problem today" produces information.
Likert by message
On WhatsApp, five points is the practical ceiling: each point becomes a clickable option, numbered 1 to 5, with the label written next to it. Asking someone to type out "somewhat agree" is a guarantee of an answer impossible to tabulate, because everyone writes it differently. If the option list does not fit, prefer a simple numeric scale over squeezing seven labels onto a phone screen.
Frequently asked questions
Does a Likert scale have to have 5 points?
Five is the practical standard, but seven shows up in academic research that needs more nuance. Below five, the scale loses its ability to distinguish opinions, and above seven the labels start looking so alike that the respondent picks almost at random. In message-based support, five points is the limit of what fits on a screen without confusing anyone.
Should I include the neutral point in the middle of the scale?
It depends on what you intend to do with it. With an odd number of points, there is a middle, which welcomes people who genuinely have no opinion but also becomes a refuge for those who do not want to commit. With an even number, you force a choice of side, at the cost of pushing the undecided toward a position they do not feel.
Can you take the average of a Likert scale?
Technically the scale is ordinal, meaning the order of the points is known but the distance between them is not necessarily equal. That is why the most defensible number is the percentage of responses in the two positive points, called the top-2-box. The average circulates a lot in the market and works to track a trend, as long as it is not treated as an exact measure.
What is the difference between a Likert scale and NPS?
The Likert scale is the response ruler; NPS is a specific indicator, with a fixed recommendation question and a 0-to-10 scale. NPS is not Likert because it does not measure agreement with a statement and because it groups scores into promoters, neutrals and detractors. A questionnaire can have several Likert items, while NPS is always just one question.
How do I use a Likert scale on WhatsApp?
Reduce it to five points and turn each point into a clickable option, numbered 1 to 5, with the meaning written next to it. Avoid asking for a typed-out answer, because variation in wording breaks the tabulation. Keep the same scale across every question so the answers can be compared to each other.
Related pages
Related glossary terms
Satisfaction survey
A short questionnaire sent after a contact to turn the customer's perception of that episode into a comparable score.
CSAT
An indicator that expresses, as a percentage or an average, the score a customer gives to a specific interaction right after it ends.
NPS (Net Promoter Score)
A loyalty metric calculated from the 0-to-10 question about the likelihood of recommending, subtracting the percentage of detractors from promoters.
Customer satisfaction
The result of comparing the expectation a customer had against the experience they perceive they received.