System Usability Scale (SUS): the 10-question system usability survey and how to score it

The System Usability Scale (SUS) is a 10-question system usability survey that people fill in right after using a product, and it turns their answers into one score from 0 to 100. The average score is 68, so above 68 means people found your product easier to use than most products measured, and below 68 means harder (MeasuringU).

At ten items it is quick to give and score, and it will not tell you what to fix. Use it as a thermometer, and run a usability test to find the cause of the fever.

1. Use the SUS because it was built for tired people at the end of a session

John Brooke created the scale in 1986 while setting up a usability engineering programme at Digital Equipment Corporation, and published it in a 1996 book chapter called "SUS: A quick and dirty usability scale" (Brooke). His test sessions ran 20 minutes to an hour, users were often frustrated by the end, and he expected a questionnaire of more than 25 questions to go unfinished.

He chose the 10 items from a pool of 50. Twenty colleagues rated one very easy system and one that was nearly impossible to use, and he kept the statements that drew the most extreme answers. Half the items are positive and half negative, so people have to read each one instead of ticking the same column all the way down.

That age is a feature. Because the questionnaire has barely changed in decades, there is a large pile of scores to compare yours against (MeasuringU's average of 68 comes from 500 studies).

2. Ask the 10 items exactly as written

Each item is answered on a five-point scale from 1 (strongly disagree) to 5 (strongly agree). Here is the wording from Brooke's chapter, which MeasuringU also lists:

1. I think that I would like to use this system frequently.
2. I found the system unnecessarily complex.
3. I thought the system was easy to use.
4. I think that I would need the support of a technical person to be able to use this system.
5. I found the various functions in this system were well integrated.
6. I thought there was too much inconsistency in this system.
7. I would imagine that most people would learn to use this system very quickly.
8. I found the system very cumbersome to use.
9. I felt very confident using the system.
10. I needed to learn a lot of things before I could get going with this system.

Odd items are the positive ones and even items the negative ones. Keep all ten and keep the order. The 68 benchmark was built on the standard questionnaire, so a homemade variant no longer compares cleanly.

Brooke's instructions for running the SUS test are short. Give it after the person has used the system and before any debriefing or discussion, ask for their immediate reaction to each item, and have them mark the middle point (3) on any item they cannot answer (Brooke). The SUS is a Likert scale, so the usual cautions about agreement scales apply; Likert scale questions covers them.

3. Score each person in four steps

Brooke gives the method:

1. For each odd item (1, 3, 5, 7, 9), subtract 1 from the answer.
2. For each even item (2, 4, 6, 8, 10), subtract the answer from 5.
3. Add the ten results. Each one is between 0 and 4, so the total is between 0 and 40.
4. Multiply the total by 2.5 to get a score between 0 and 100.

Here is a made-up example. Priya manages the front desk at a small dental clinic. She spends 30 minutes booking, moving and cancelling appointments in a new scheduling app (including one patient who has rescheduled three times this week), then fills in the SUS.

The odd items add up to 16 and the even items to 15, so the total is 31. Multiplied by 2.5, Priya's SUS score is 77.5.

Two mistakes are common. Averaging the raw answers skips the reversal of the negative items, so a person who strongly disagrees that the app is cumbersome drags the number down. Reporting 77.5 as "77.5% usable" is the other one: SUS scores are not percentages, and a 68 sits at the 50th percentile.

Brooke also warns that scores for individual items are not meaningful on their own. The scale is the ten items together.

4. Average everyone's score, then compare it with 68

The study's SUS score is the average of the individual scores. Continuing the made-up example, five front desk staff try the app and score 77.5, 70, 85, 62.5 and 72.5. Those add up to 367.5, and 367.5 divided by 5 is 73.5.

Jeff Sauro and Jim Lewis built a curved grading scale that turns scores into letter grades and percentiles. A few rows:

The clinic app's 73.5 is a B minus, better than roughly two thirds of the products in the benchmark. Grades are easier to explain to a manager than a raw number, which is the main reason to use them.

5. Report five-person scores as rough, and save comparisons for bigger samples

The SUS can be used with as few as two people, but small samples give imprecise estimates, so MeasuringU recommends reporting a confidence interval. With five users, the average lands within six points of the large-sample score only half the time (MeasuringU). The clinic app's 73.5 could easily belong to a product whose real score is 68, and it could just as easily belong to one at 79.

Short version: with five people, say "about average to good" and move on to the findings. Nielsen Norman Group says self-reported scores like the SUS need at least 20 to 30 users to work as numbers.

The SUS earns its keep in comparisons: this version against the last one, or your product against a competitor's, with the same tasks each time. Brooke built it to compare one software version with the next. For a competitive test, NN/g gives it after the participant finishes all the tasks for one product.

6. Pair the score with a usability test, because the SUS says how usable and never why

MeasuringU says plainly that the SUS was not intended to diagnose usability problems: a low score tells you a problem exists without telling you what it is. NN/g makes the same point and recommends following the numbers with "why" questions.

So run it at the end of a usability test. The person does the tasks while you watch, fills in the SUS before you chat, and then you debrief. If Priya marked 3 on "I needed to learn a lot of things," ask what she had to figure out. The answer is a lead to follow, while the SUS score stays one number for the whole product.

The SUS also sits next to the Net Promoter Score, which asks whether people would recommend you. They are related: in MeasuringU's data, SUS scores explain about 40% of why people recommend software and websites, with promoters averaging 82 and detractors 67 (MeasuringU). NPS measures loyalty, which has causes other than ease of use. Use the SUS when the question is specifically about ease.

7. Skip the SUS when the person has not used the whole product

The ten items ask about the system as a whole, so the person needs a real session with it first. For a quick check after one task, NN/g points to the Single Ease Question, a single 7-point rating of how hard the task was. For pop-up surveys inside a live product, short questions tied to the moment work better; product survey questions has examples.

Your next step

Write three realistic tasks for your product, book five people who do that work every day, and give them the SUS before the debrief. Instant Expert can find people matching a description such as "front desk managers at small dental clinics who schedule patients daily," send them paid invitations for a short call, and you pay only if someone books. For help designing the study itself, see the directory pages for user research experts and UX consultants.