Marcus Andrey Study
Ler em português Library

Framework · Measurement · 2 min read

Task success and time on task

The core performance metrics for internal tools: the percentage of users who complete a task, how long it takes (median), and how many errors they make.

STask success

% of users who reach the defined end state of the task.

Typically: 14 of 20 = 70%
TTime on task

How long the task takes. Report the median, not the mean.

Typically: median 3:10
EErrors

Mistakes and corrections per task, key for data quality.

Typically: errors per attempt
Task success, time on task and errors together show whether people can do their work in the tool.

Use it when

  • You need the simplest, most convincing evidence that a tool works or doesn't.
  • You are setting up a UX benchmarking baseline for the top tasks of an app.
  • You want to show the effect of a redesign in a before/after report.

How to use it

  1. Define each task and its success criteria. Write the task as a realistic scenario and state the exact end state that counts as success, such as "booking saved with the correct customs code". Decide upfront how you treat partial success and keep that rule every round. Output: task list with written success criteria.
  2. Measure success. For each participant and task, record success or failure against the criteria. Task success rate = successes ÷ attempts. Output: a success rate per task.
  3. Measure time on task. Record time from the start of the task to success. Say whether you include failed attempts. Report the median. Output: median time per task.
  4. Track errors alongside. Count mistakes and corrections per task, such as wrong field values or undone steps. For operational tools these drive data quality. Output: error rate per task.
  5. Compare over time. Use the same tasks, criteria and method each round, and report before/after. Output: a before/after table.

Worked example

A baseline test of one task on an internal operations platform, 8 participants (fictional data). Success criteria: "booking saved with the correct customs code, without help".

Participant Success Time (min:s) Errors
P1 Yes 2:40 0
P2 No — 3
P3 Yes 3:05 1
P4 Yes 2:55 0
P5 No — 2
P6 Yes 3:30 1
P7 Yes 9:50 2
P8 Yes 3:15 0

Success = 6 / 8 = 75%. Median time of successful attempts = 3:10, while the mean is about 4:13 because of P7. Errors: 9 across 8 attempts.

Watch out for

  • Success criteria set afterwards. It invites generous scoring. Write them down before the first session.
  • Using the mean for time. Time data is skewed by a few slow attempts. Use the median.
  • Success without errors. A task "completed" with wrong data is not a success for an operational tool. Track errors too.
  • Small samples read as precise. With 5–8 users, 75% is a rough estimate. Report the counts ("6 of 8"), not just the percentage.

Template

Copy this into your test plan and fill one block per task.

Task:                      Scenario given to user:
Success criteria (end state):
Partial success counts as:  fail / half / success

| Participant | Success (Y/N) | Time | Errors | Notes |
|-------------|---------------|------|--------|-------|
|             |               |      |        |       |

Success rate:   __ / __ = __%
Median time:    (successful attempts only / all attempts)
Errors:         __ per attempt

"See on Amazon" links are affiliate links. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.

Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close