UX Research · Step 4 · 7 min read
Tasks and scenarios
In a usability test, the task is the question, and a badly worded task gets a misleading answer. How to turn user goals into realistic scenarios, avoid giving away the path, order tasks from exploratory to directed, define success before the session, and check your tasks against the ten most common mistakes.
If the task tells people where to click, the test can only tell you whether they can read. A usability test is only as good as its tasks. Write "click Settings and change your password" and everyone succeeds, and you learn nothing about whether anyone would ever find Settings. Write a real situation with a real goal, and you see what people actually do. The words on the task card are the most important design decision in the whole test.
What you'll be able to do after reading:
- Turn the users' goals into task scenarios
- Write tasks that are realistic, actionable and free of clues
- Order tasks from exploratory to directed, and step them when needed
- Define success for each task, and check tasks against common mistakes
1. From goals to scenarios
Start with the goals, not the screens. Marieke McCloskey's question at NN/g is the right one: what are the most important things every user must be able to accomplish? Then turn each goal into a task scenario: a short situation that gives a reason and the details needed for the action, instead of simply ordering people to "do X" (NN/g, McCloskey, Task Scenarios for Usability Testing, 2014).
Usability testing with five people gave you the basic rule: a realistic goal, no hints. McCloskey's three tips make it precise:
- Make it realistic. Ask people to do what they'd really do. If you test a hotel site, recruit people who book the travel at home. Or let them bring their own task: someone really shopping for a car continues their research during the session.
- Make it actionable. Ask them to do it, not to describe it. "Tell me where you'd click" gets words, and people are poor at predicting what they'd do. If a participant takes their hand off the mouse and starts explaining, the task wasn't actionable enough.
- Avoid clues. Don't describe the steps, and don't use the interface's own words. If the button says "Subscribe", the task says "get information on upcoming events sent to your email regularly". But avoiding clues doesn't mean being vague: where something has a standard, well-known name, use it.
Click "Split unevenly" and set Ana's share to 50% so the internet bill is fair.
- Uses the label on the button: you test reading, not finding
- Gives the steps, so you never see their path
- Gives the answer (50%) instead of a situation
The internet bill was €36, shared by you, Ana and Rui. Ana was away for the first half of the month. Make the split fair, the way you'd do it at home.
- A real situation with a goal, not an instruction
- No interface words: they have to find "split unevenly"
- Clear end point you can check: the amounts on screen
- "Fair" is theirs to define, which shows their mental model
2. Ten mistakes to check for
Amy Schade's list at NN/g is the checklist to run over every draft (NN/g, Schade, Write Better Qualitative Usability Tasks, 2017):
| Mistake | Instead |
|---|---|
| Telling users where to go (words from the interface) | Describe the goal in the user's words |
| Telling users what to do (warning them they'll need to register) | Let them meet the steps as they would in real life |
| Out-of-date tasks (a flight that's already left) | Check dates and content the day before testing |
| Too simple (just find the page) | Make them use the information: compare, decide |
| An elaborate scenario | Keep it short; a long story to justify a task means it isn't realistic |
| Writing an ad ("our exciting new feature") | Plain, user-centred language, no internal jargon |
| Risking an emotional reaction ("your mother") | Neutral relationships: a friend, a colleague |
| Trying to be funny (jokes, famous names) | Plain and polite |
| Offending (politics, health, religion, age, money) | Avoid sensitive details unless the product needs them |
| Asking rather than telling ("How would you…?") | Ask them to do it |
3. Order: from exploratory to directed
The funnel from The interview guide works for tasks too. Start with broad, exploratory tasks, where people define the end point and do it their own way, then move to directed tasks with a clear end and success criterion, and only then, if you need to, to tasks that point at a specific feature. Do it the other way round and the early tasks teach people what you're looking for, so you never see what they'd have done on their own (NN/g, Rosala and Moran, The Funnel Technique, 2022).
When you need to test a feature people might not find on their own, use a stepped task: start natural, and add a more specific instruction only if they don't get there.
- "Choose a pair of headphones you'd like to buy and add them to your cart." (Will they find the comparison feature by themselves?)
- "Your friend can't decide between these two models. Decide which one you'd recommend, and why." (A situation that invites comparing.)
- "Can you find a way to compare them without switching back and forth?" (A strong hint.)
If you only get the participant to the feature at step 3, that's a finding about discoverability, and you still get to see whether the feature itself works (Rosala and Moran, 2022).
4. Define success, and hand the task over
For every task, write down before the sessions what success means: the end state, and what counts as done with help. "Bill added with the right amounts, without the moderator's help" can be scored the same way for every participant. Without it, success gets decided afterwards, generously.
Then hand over one task at a time, on a card or a screen. NN/g often asks participants to read the task aloud, which makes sure they read all of it and helps the note-taker, who always knows which task is under way (NN/g, Moran, Usability Testing 101, 2019). Keep the card in view so they can check the details (the amount, the date) without asking you.
See Usability task scenarios for the whole method on one page, with a task card template.
5. Common mistakes
- Tasks written from the screens ("test the filter") instead of the users' goals.
- The button's label inside the task.
- Directed tasks first, so the exploratory ones are already primed.
- Tasks with no checkable end, so success is a matter of opinion.
- Dates, prices or stock in the task that are no longer true on test day.
- Too many tasks for the time, so the last ones are rushed.
6. Vocabulary
| Term | Meaning |
|---|---|
| Task | What you ask the participant to do in a usability test |
| Task scenario | A task set in a short, realistic situation that gives a reason and the needed details |
| Exploratory task | An open task where the participant defines the end point |
| Directed task | A task with a defined end point and success criterion |
| Stepped task | A task that starts natural and gets more specific only if needed |
| Success criterion | What has to be true for the task to count as completed |
| Priming | When earlier words or tasks influence what the participant does next |
7. Try it (no tools needed)
- Pick an app you use and one goal people have in it.
- Write the task three ways: as an instruction ("Click…"), as a vague request ("Look around…"), and as a scenario with a goal and the details.
- Run the ten-mistake table over the scenario. Fix what it catches, then write its success criterion in one line.
Reflection questions: Which interface words were hard to avoid? Would a real person ever be in the situation your scenario describes?
8. If you already practise
- Read your tasks to someone outside the project before the pilot. Every question they ask is a gap in the task.
- Keep a bank of tasks for your product's top goals, so you can re-run them release after release and compare (see Measuring usability).
- Let some participants bring their own task, in the same session as your scripted ones. Their goals sometimes aren't on your list.
- In unmoderated tests, the task card is all there is: pilot it twice.
Checklist
Learn more
- NN/g: Task Scenarios for Usability Testing (McCloskey, 2014) · Write Better Qualitative Usability Tasks: Top 10 Mistakes to Avoid (Schade, 2017) · The Funnel Technique in Qualitative User Research (Rosala and Moran, 2022) · Usability (User) Testing 101 (Moran, 2019)
Books, courses and videos to go deeper
- Rocket Surgery Made EasySteve KrugBook · New Riders, 2010 (Amazon)
- Write Better Qualitative Usability TasksNN/gfree, article (external site)
Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.
Related
Usability testing with five people · Planning a usability test · Moderating and thinking aloud · User needs and jobs to be done · Users, tasks and context · Test with three people · Usability task scenarios