What is time on task?

Also called: Task time, Task completion time, Time-on-task

Time on task is a usability metric that records how long a participant takes to complete a task, from the moment they start to the moment they finish or give up. It's used to measure efficiency and to compare designs or versions.

Updated

What it tells you

Time on task measures efficiency: how much effort a design asks of people for a given job. It is most useful as a comparison. A new checkout against the old one, your onboarding against a competitor's, or the same task release after release. A single time on its own says little, because there's rarely an agreed right number of seconds for a task.

How to measure it consistently

  • Define start and end before the test. Start when the participant finishes reading the task, end when they reach a defined success screen or say "I'm done" or "I give up".
  • Decide what happens to failed attempts. Most teams report time for successful completions only, and report the success rate separately, so a fast wrong answer doesn't look like good performance.
  • Use the same task wording and starting screen for every participant and every round.
  • Don't combine it with concurrent think-aloud. Talking slows people down, so think-aloud sessions are for finding problems, not timing them.

How to report it

Task times are skewed: most people finish in a similar time and a few take much longer. The arithmetic mean gets pulled up by those few, so report the median or the geometric mean instead, along with the range or a confidence interval. For example, report "median 48 seconds, range 31 to 190 seconds, 12 of 15 succeeded" rather than "average 72 seconds".

In unmoderated tests, look at outliers before you remove them. A participant who took ten minutes may have been interrupted, or may have been lost. The recording tells you which.

When time on task misleads

  • Faster isn't always better. In a reading or comparison task, more time can mean people are engaged.
  • Small samples move a lot. With five participants, one slow session changes the result, so treat small-sample times as directional.
  • It doesn't explain itself. Pair it with recordings or follow-up questions to find out why a task was slow.

Go deeper

Questions

Should you include failed tasks in time on task?

Usually not in the headline number. Report time for successful completions, then report the success rate next to it. If you want to show how long people struggled before giving up, report that as a separate time-to-failure figure.

How many participants do you need to compare task times?

More than you'd need to find usability problems. Five participants can show you where people get stuck, but comparing two designs on speed needs a larger sample. How large depends on how much task times vary and how small a difference you need to detect, so run a power calculation or a pilot before fixing the number.

Why use the median instead of the average for task time?

Because a few slow sessions pull the average up. The median shows the time a typical participant took, which is closer to what most of your users will experience.

Next step

Run the study, not just the method

Great Question handles recruiting from your own customers, scheduling, incentives, interviews, tests and analysis in one place.

Book a demo

No credit card required.

Related terms

Full glossary
  • Think-aloud protocol

    A think-aloud protocol is a usability research method where participants say what they are thinking, looking at and trying to do while they complete a task, so you hear the reasoning behind each action instead of guessing at it.

  • Task analysis

    Task analysis in UX is the process of breaking down how people accomplish a goal into the steps, decisions, tools and information each step needs, so you can design a product around the work as people actually do it.

  • HEART framework

    The HEART framework is a set of five user experience metric categories from Google: Happiness, Engagement, Adoption, Retention and Task success. Teams use it to choose metrics that show whether a product or feature is working for the people using it.