Framework · Research · 3 min read
Usability task scenarios
How to write the tasks for a usability test: a short realistic situation with a goal and the details, no interface words, ordered from exploratory to directed, with a success criterion written before the first session.
RRealistic
A real situation, with a reason and the details the action needs.
"Ana paid €240 for the cabin."AActionable
Ask people to do it, not to say where they'd click.
"Make sure Ana gets your share."CNo clues
No interface words and no steps, unless it's the standard name.
Not "Click Settle up"SSuccess
The end state, written before the first session.
€60 sent to Ana, without helpUse it when
- You prepare a moderated or unmoderated usability test.
- You want to re-run the same tasks release after release and compare.
- A draft task says "Go to Settings and…" or "Test the new filter".
How to use it
- Start from the top goals. List the most important things every user must be able to do, from the research, not from the screens. Output: three to five user goals.
- Write each goal as a scenario. Give a short reason and the details the action needs (the amount, the date, the name). Ask people to do it, not to describe it. Output: one task card per goal.
- Remove the clues. Replace every word that appears in the interface, unless it's the standard name for the thing. Don't describe the steps. Output: tasks in the user's words.
- Order from exploratory to directed. Open tasks first, where people define the end point, then directed ones with a clear end. For a feature people may not find, write a stepped task. Output: tasks in funnel order.
- Define success before the sessions. For each task, write the end state and what counts as done with help. Output: one success criterion per task.
- Check and hand over. Run each task against the ten mistakes, and check dates and prices the day before. In the session, give one task at a time, ask the participant to read it aloud, and keep it in view. Output: a final, checked task set.
Worked example
A bill-splitting app with a new "Settle up" button.
| Task | |
|---|---|
| Instruction (weak) | "Click Settle up and pay Ana." |
| Vague (weak) | "Look around and tell us what you think." |
| Scenario | "Last weekend you went away with three friends. Ana paid for the cabin, €240. Make sure Ana gets your share back." |
| Success | Payment of €60 to Ana sent, without the moderator's help. |
| Stepped follow-up | If they pay outside the app: "Can you find a way to record in the app that you've paid?" Reaching the feature only here is a finding about discoverability. |
Watch out for
- Interface words in the task. The button's label in the task turns the test into a reading exercise.
- Directed tasks first. Early specific tasks teach people what you're looking for, so the open ones are already primed.
- No checkable end. Without a success criterion written before the sessions, success gets decided afterwards, generously.
- An elaborate story. If a task needs a long backstory to make sense, it probably isn't realistic.
Template
One card per task. Keep the success line for your notes, not on the participant's card.
Goal (from research):
Scenario (what the participant reads):
"You ... [situation, reason]. ... [details: amount, date, name].
... [what to achieve, in the user's words]."
Type: Exploratory / Directed / Stepped
Step hint (only if needed):
Success: [end state] without help / with help counts as: ___
Interface words removed: yes / no Dates and prices checked on: ___
Ten-mistake check: where to go · what to do · out of date · too simple ·
elaborate · an ad · emotional · funny · offensive · asking not telling
- Rocket Surgery Made EasySteve KrugBook (Amazon)
- Task Scenarios for Usability TestingNN/g (external site)
- Write Better Qualitative Usability Tasks: Top 10 Mistakes to AvoidNN/g (external site)
- The Funnel Technique in Qualitative User ResearchNN/g (external site)
"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.