The System Usability Scale (SUS) is a standardized questionnaire used to assess a person’s perception of a system’s usability. Developed by John Brooke in 1986, it consists of ten statements rated on a five-point agreement scale. Responses are combined into a score from 0 to 100, with higher scores indicating more favorable perceived usability. SUS is widely used in human–computer interaction research and product evaluation because it is brief, relatively easy to administer, and applicable to many kinds of interactive systems.
Origins and scope
Brooke developed SUS while working at Digital Equipment Corporation. It was intended as a practical, low-cost way to obtain an overall usability assessment, including comparisons between systems. Its description appeared in the 1996 volume Usability Evaluation in Industry, under the title “SUS: A ‘Quick and Dirty’ Usability Scale.”
Although originally developed in a computing context, SUS does not depend on a particular technology or interface. It has been applied to websites, desktop applications, mobile applications, hardware, and other interactive products.
The questionnaire measures perceived usability, rather than directly measuring how effectively or efficiently people perform tasks. This distinction places it alongside, rather than in place of, behavioral measures collected during usability testing, such as task completion, time on task, and observed errors.
Questionnaire structure
SUS contains ten statements, each answered on a five-point Likert-type response scale ranging from strongly disagree to strongly agree. In the standard coding, these responses receive values from 1 to 5.
The statements alternate between favorable and unfavorable wording. Agreement with an odd-numbered statement indicates a more favorable assessment; agreement with an even-numbered statement indicates a less favorable assessment.
The following table summarizes the item themes rather than reproducing the questionnaire’s exact wording:
| Item | Main theme | Wording direction |
|---|---|---|
| 1 | Willingness to use the system frequently | Favorable |
| 2 | Unnecessary complexity | Unfavorable |
| 3 | Ease of use | Favorable |
| 4 | Need for technical assistance | Unfavorable |
| 5 | Integration of system functions | Favorable |
| 6 | Inconsistency within the system | Unfavorable |
| 7 | Expected speed with which others would learn the system | Favorable |
| 8 | Awkwardness of use | Unfavorable |
| 9 | Confidence while using the system | Favorable |
| 10 | Amount to learn before getting started | Unfavorable |
Despite addressing several aspects of experience, the standard instrument is generally interpreted through its overall score, not as ten independent usability measurements. Individual responses can suggest topics for investigation, but an item alone is not a validated diagnostic scale.
Administration
SUS is normally administered after participants have had an opportunity to use the system, often at the end of a set of representative tasks. In its original administration guidance, it is completed before a broader discussion or debriefing that might influence participants’ answers.
The experience preceding the questionnaire matters. A score obtained after a brief encounter with a prototype does not necessarily represent the experience of an established user completing routine work. Relevant conditions include:
- The tasks attempted and their difficulty.
- Participants’ prior experience with the system and its domain.
- Training, instructions, and assistance provided.
- The system version and testing environment.
- Whether participation involved actual use or only viewing a demonstration.
Standardized wording, item order, and response options help preserve comparability with previous studies. Translated or adapted versions require attention to meaning and measurement properties; grammatical accuracy alone does not establish that a translation behaves like the original instrument.
Scoring
Let denote a participant’s response to item , coded from 1 to 5. Each response is converted into a contribution from 0 to 4:
The participant’s SUS score is:
Equivalently,
The transformation aligns all items so that higher contributions indicate more favorable perceived usability. Multiplication by 2.5 converts the total contribution, which ranges from 0 to 40, into a score from 0 to 100.
For example, responses of 4 to every favorable statement and 2 to every unfavorable statement yield a contribution of 3 per item:
An individual score changes in increments of 2.5, although an average across participants can take other values.
The standard calculation assumes that all ten items have been answered. Missing-response handling is an additional analytical decision, not part of the basic scoring formula, and can affect comparability.
Interpreting scores
A score is not a percentage
A SUS score of 75 does not mean that a system is “75 percent usable,” that 75 percent of tasks were completed, or that 75 percent of users were satisfied. It is a transformed questionnaire score.
Nor is the score itself a percentile rank. A percentile describes a score’s position within a reference distribution; obtaining one requires comparison with an appropriate benchmark dataset.
Benchmarks and descriptive labels
A score around 68 is often cited as an average based on pooled historical SUS datasets. It is a reference point, not a universal passing score. Its relevance depends on the products, users, and circumstances represented in those datasets.
Researchers have also related SUS scores to adjective ratings, acceptability categories, letter grades, and percentile ranks. These interpretation systems are empirical aids rather than intrinsic features of the scoring formula. Different mappings can assign different labels to the same numerical score.
A score near the mathematical midpoint of 50 should therefore not automatically be described as average. The midpoint of the scale and the average of observed scores are different quantities.
Group comparisons and uncertainty
Studies commonly report the mean SUS score across participants. The variability of participant scores and a confidence interval help distinguish the estimated average from the uncertainty around it.
A higher sample mean does not, by itself, establish a reliable or practically important difference between products. Interpretation depends on sample size, participant selection, study design, and the size of the difference. Repeated ratings by the same participants also require analysis that accounts for the dependence between observations.
There is no single minimum sample size suitable for every SUS study. A small sample may provide an initial descriptive result, but a stable benchmark or precise comparison usually requires more observations.
Measurement properties
SUS has been studied extensively using psychometric methods. Research across varied applications has generally supported its use as a compact overall measure of perceived usability.
Its internal consistency concerns how coherently its items function together. High internal consistency does not establish that SUS measures every aspect of usability or that it can identify the cause of a poor experience.
The questionnaire’s dimensional structure has also been debated. Some research has proposed separate usability and learnability components, with items 4 and 10 forming a learnability subscale. Other work has questioned the stability and usefulness of that separation, including the possibility that apparent factors reflect wording effects. The overall ten-item score remains the conventional basis for reporting SUS results.
Alternating positive and negative wording can introduce response errors when participants overlook a reversal or find a statement difficult to interpret. Modified all-positive versions have been studied, but they are adaptations whose equivalence should be evaluated rather than assumed.
Applications and limitations
SUS is used to compare design alternatives, track perceptions across product versions, and provide a common questionnaire measure across studies. Its brevity makes it useful when a lengthy survey would place excessive demands on participants.
Its principal limitations follow from its scope:
- It is global rather than diagnostic. A low score does not reveal which navigation path, control, or workflow caused difficulty.
- It measures perceptions rather than task performance. Favorable ratings can coexist with errors, incomplete tasks, or inefficient behavior.
- It is context-sensitive. Familiarity, training, task selection, and expectations can affect responses.
- It does not measure every aspect of user experience. Visual appeal, enjoyment, trust, and other qualities require additional evidence.
- It is not an accessibility assessment. A favorable average does not establish accessibility or show that people with disabilities can use the system effectively.
- Cross-study comparisons may be confounded. Different participant populations and testing procedures can produce score differences unrelated to interface quality.
Accordingly, SUS results are commonly interpreted alongside observed behavior, task outcomes, and qualitative findings. These complementary measures address questions that a single overall questionnaire score cannot resolve.