UX metrics

Why it matters

Findings like "four of five people missed the button" are powerful, but teams also need numbers: to compare two designs, to show that a redesign helped, or to track a product over time. UX metrics give you those numbers. The most common ones are simple to collect in the usability tests you already run.

The core idea: measure behaviour and attitude

UX metrics fall into two families:

• Behavioural (what people do): task success, time on task, errors.
• Attitudinal (what people say): ease ratings and questionnaires such as the System Usability Scale.

Use both. People sometimes rate a task as easy even after failing it, and sometimes complete a task while hating every second.

Task success rate

The most important usability metric. For each task, record whether each participant succeeded, using the success criteria you set before testing.

success rate = successful attempts / total attempts x 100

Decide up front how to count assisted completions (when the facilitator helped). A strict approach counts only unassisted successes. If 5 participants did a task with results success, fail, success, assisted, success, the strict rate is 3 / 5 = 60%.

Time on task and errors

Time on task is how long a participant takes from the start of a task to success. It is most useful for comparing designs or tracking frequent tasks. Times are skewed by a few slow participants, so report the median (or the geometric mean) rather than the plain average, and only include successful attempts.

Errors are wrong actions, such as tapping the wrong option or entering an invalid value. Count them per task to show where people stumble.

Single Ease Question (SEQ)

After each task, ask: "Overall, how easy or difficult was this task?" on a 1 (very difficult) to 7 (very easy) scale. It takes seconds and adds an attitude measure per task.

The System Usability Scale (SUS)

SUS, created by John Brooke in 1986, is a ten-item questionnaire given at the end of a test. Participants rate each statement from 1 (strongly disagree) to 5 (strongly agree):

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-numbered items are positive; even-numbered items are negative. That is why scoring needs care:

• For odd items (1, 3, 5, 7, 9): contribution = response - 1.
• For even items (2, 4, 6, 8, 10): contribution = 5 - response.
• Add the ten contributions (0 to 40) and multiply by 2.5. The result is 0 to 100.

Worked example. Responses: 4, 2, 4, 2, 4, 2, 4, 2, 4, 2.

• Odd items: 4 - 1 = 3, five times = 15.
• Even items: 5 - 2 = 3, five times = 15.
• Sum = 30; 30 x 2.5 = 75.

A SUS score is not a percentage. Across many published studies the average is about 68, so 75 is above average. Scores around 80 or more are generally considered good. With five participants, treat SUS as a rough signal, not a precise measurement; it becomes more reliable with more participants and is most useful for comparing versions of the same product.

Reporting metrics honestly

• Report the number of participants next to every number: "60% (3 of 5)".
• With small samples, avoid decimals that suggest precision you do not have.
• Pair numbers with findings: the metric tells you how much, the observations tell you why.
• Compare like with like: same tasks, same type of participants.

Common mistakes

• Averaging raw SUS responses instead of applying the odd/even scoring.
• Calling SUS a percentage. 68 is average, not "68% usable".
• Mean time on task dominated by one very slow participant.
• Counting assisted completions as success without saying so.
• Numbers without context: a success rate with no participant count.

Quick recap

• Measure behaviour (task success, time, errors) and attitude (SEQ, SUS).
• Task success = successes / attempts; decide how to treat assisted completions in advance.
• SUS: odd items response - 1, even items 5 - response, sum x 2.5; average is about 68.
• Report sample sizes and pair metrics with qualitative findings.

Try it

On the Do tab you will write small JavaScript functions that calculate task success rate and SUS scores, then use them on your own test data.