What is a think-aloud protocol?
Also called: Thinking aloud, Think-aloud method
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.
Updated
Where it comes from
The method comes from cognitive psychology. K. Anders Ericsson and Herbert Simon set out the rules for collecting verbal reports as data in Protocol Analysis (1984, revised 1993). Jakob Nielsen brought it into usability work in the early 1990s and called it the single most valuable usability engineering method.
Concurrent vs retrospective think-aloud
- Concurrent: the participant talks while they work. You get reasoning in the moment, including hesitation and wrong guesses. Talking can slow people down, so task times are less reliable.
- Retrospective: the participant completes the task in silence, then watches the recording and explains what they were doing. Task times stay natural, but people tidy up their reasoning after the fact and forget small moments of confusion.
- Most product teams use concurrent think-aloud for moderated and unmoderated usability tests, and keep retrospective for tasks where timing matters.
How to run it without leading the participant
- Show a short practice task first, so people hear what thinking aloud sounds like before the real tasks start.
- When they go quiet, use neutral prompts such as "What are you looking at now?" or "Keep talking." Avoid "Is that confusing?", which hands them an answer.
- Don't explain the interface or confirm whether they're right. The point is to hear what they understand without help.
- In unmoderated tests, put the instruction to think aloud in the task text and repeat it before each task, because nobody is there to prompt.
What it's good and bad at
Think-aloud is strong at finding why people get stuck: mislabelled buttons, wrong mental models, information they expected and couldn't find. It is weak at measuring things. Talking changes behaviour, so don't use think-aloud sessions to report task time or to compare designs on speed.
Go deeper
- Usability testing guide
How to plan the tasks a think-aloud session runs on.
- Unmoderated testing guide
Running think-aloud tasks when no moderator is in the room.
- Prototype testing guide
Think-aloud on prototypes before anything is built.
- Unmoderated prototype testing in Great Question
Run think-aloud prototype tests with your own customers.
Questions
How many participants do you need for a think-aloud study?
For finding usability problems in one design, five to eight participants per distinct user group usually surface the most common issues. Nielsen's well-known five-user figure assumes one homogeneous group and a fairly simple interface, so add participants for each extra segment.
Does thinking aloud change how people behave?
Yes, a little. Concurrent think-aloud slows people down and can make them more careful. That's why it suits finding problems, not measuring task time or success rates against a benchmark.
Can you use think-aloud in unmoderated tests?
Yes. Ask participants to talk through what they're doing in the task instructions, and record screen and audio. Expect more silent stretches than in moderated sessions, so repeat the instruction before each task.
Run the study, not just the method
Great Question handles recruiting from your own customers, scheduling, incentives, interviews, tests and analysis in one place.
No credit card required.
Related terms
Full glossary- Data saturation
Data saturation is the point in qualitative research where new interviews or observations stop producing new themes or insights, which tells you that adding more sessions would mostly repeat what you already know.
- Triangulation
Triangulation is checking a research finding against more than one source, method or researcher, so you only act on it when different kinds of evidence point the same way.