October 17, 2026
What Is a Likert Scale? Points, Examples and Common Mistakes
A Likert scale measures how strongly someone agrees with a statement. Learn the 5 and 7 point formats, midpoint debates, labeling rules and how to analyze the data.
TL;DR
A Likert scale gives people a range of ordered answers, usually from strongly disagree to strongly agree, so you can measure intensity instead of a yes or no. Five points is the default, seven adds nuance, and every point should be labeled. Report the distribution, not just the average, and never sum a single Likert item into a score without checking that its wording is neutral.
Author: Inputly
What is a Likert scale?
A Likert scale is an ordered set of answer options that measures how strongly a person agrees or disagrees with a statement. Instead of a binary yes or no, the respondent picks a point on a range, typically from "strongly disagree" to "strongly agree". The result is a number that carries intensity: a 5 means something different from a 3, and that difference is the whole point. The format is named after psychologist Rensis Likert, who published the technique in 1932, and it remains the default for surveys because it is fast to answer and simple to summarize. It measures attitude, not behavior, so pair it with a behavioral question or a product metric before you make a decision on it.
What is the difference between a Likert scale and a Likert item?
A Likert item is one statement with its response options. A Likert scale is the set of several items that you combine to describe an attitude, such as five statements about onboarding clarity that together produce one score. People mix the two terms constantly, which causes a common error: treating a single item as a reliable scale. One item can flip on wording alone. Three to five items on the same theme, checked for internal consistency, give a steadier picture. If you only have room for one question in a product, call it a rating question and read it as a single signal.
How many points should the scale have?
| Format | Best use | Trade-off | | --- | --- | --- | | 5 points | Product surveys, in-app prompts, eNPS style questions | Fast, mobile friendly, less nuance | | 7 points | Research panels, benchmarks, long surveys | More discrimination, slower to answer | | 4 points | Forced choice where no opinion is not useful | No midpoint, can feel pushy | | 10 or 11 points | Recommend-style questions such as NPS | Familiar, but people cluster at the ends |
Five points is the practical default. It fits on a phone screen without scrolling, and most respondents answer in under three seconds. Seven points helps when you compare segments and need the extra discrimination. Whichever you choose, keep it fixed, because a midpoint change or a point added halfway through a quarter makes every earlier number incomparable.
Should you label every point?
Label every point, or at minimum the two ends and the midpoint. Numbers alone force people to invent their own meaning for 2 and 4, and different respondents invent different meanings, which is measurement noise you cannot remove later. Full labels such as "strongly disagree, disagree, neither, agree, strongly agree" remove that ambiguity at the cost of screen space. A common compromise labels all five points on desktop and shows endpoints only on mobile, but the underlying values stay the same. Whatever you choose, keep the labels identical across every survey that feeds the same metric.
Should you keep a neutral midpoint?
Keep it for product feedback. A midpoint gives people who truly have no view a place to stand, and removing it forces a side, which inflates both ends and hides indifference. The exception is a decision you must resolve: if you are choosing between two onboarding flows, a forced choice produces cleaner preference data, and you should say so in the question. Do not disguise a forced choice as an attitude question. And if you worry that people hide behind the midpoint, ask a follow-up open question instead of deleting the option.
What are the most common Likert mistakes?
- Double-barreled statements: "The setup was fast and easy" fails when the setup was fast but confusing. Split it.
- Leading wording: "How much did you love the new dashboard?" pre-loads the answer. Ask about satisfaction, not affection.
- Reversing items without warning: a negatively worded statement in the middle of positive ones catches careless clicking, but it also catches honest readers who misread it.
- Mixing directions in one grid: keep the positive end on the same side for every row, or people will answer the wrong way.
- Reporting only the average: the mean of a polarized 2.1 and a lukewarm 3.0 can look identical. Show the distribution.
These overlap with the broader traps in survey bias, where wording and sampling effects compound each other.
How do you analyze Likert data?
Start with the distribution: the share of people who picked each point, then the top-two share, which is the percentage choosing the two most positive options. The top-two box is the most useful number for a product team because it separates "delighted" from "mildly fine". Show the mean only alongside it, and state the sample size. For several items on the same theme, check that they move together before combining them into an average, and if they do not, report them separately. Segment by plan, tenure or role, because an average across all customers usually hides the group that is actually unhappy. Below roughly 100 responses, show raw counts instead of percentages.
How should product teams use Likert scales?
Use them where you need intensity and consistency, not where you need a verdict. Agreement questions work well for measuring whether a feature is understood, not whether it is liked. Follow each scale with one open question so people can explain the number, and route low scores to a follow-up survey such as a CSAT question tied to the specific interaction. If you track the relationship over time, keep those items separate from product-level feedback and read how NPS differs. Finally, treat the score as a trend, never as a target: teams that chase a single number start writing questions that flatter the number.
Ready to put these questions in front of real users? Start free with Inputly and collect structured feedback with voting and a public roadmap.
Frequently Asked Questions
Is a Likert scale the same as a rating scale? A Likert scale is one kind of rating scale, specific to agreement with a statement. A rating scale is broader and can ask about frequency, quality or likelihood, such as a 1 to 5 star rating for a support ticket.
Should a Likert scale have 5 or 7 points? Five points is the practical default for product surveys because it is fast on mobile and easy to compare over time. Seven points gives more nuance for research panels or long-running benchmarks. Pick one and keep it, because switching midstream breaks the trend.
Should you include a neutral midpoint? Yes, for most product feedback. A midpoint lets people who genuinely have no opinion say so, and removing it pushes them to a side they do not hold. If you need a forced choice for a specific decision, offer a separate no opinion option instead.
Can you average Likert scale responses? You can compute a mean for a single five or seven point item, but treat it carefully because the distance between points is an assumption. Reporting the share of agree responses plus the full distribution is more honest than a lone average.
How many responses do you need for a Likert question? Around 100 responses give a usable read on a single item and several hundred make subgroup comparisons safe. Below that, report the raw counts and read the open comments alongside the numbers.