Write an unmoderated usability test script that catches silent confusion, not just task completion

Write an unmoderated usability test script for the flow below. The specific risk with unmoderated testing is that there's no facilitator present to notice a participant hesitating, backtracking, or looking confused in the moment — a participant who eventually completes a task can still have been confused the whole way through, and a script that only checks "did they finish" will miss that entirely.

FLOW BEING TESTED
A redesigned settings page for managing notification preferences across email, push, and SMS.

SPECIFIC CONCERN PROMPTING THIS TEST
We're worried users won't realize SMS notifications need a separate opt-in from push notifications.

PARTICIPANT SCREENING CRITERIA
Existing users who have never changed their default notification settings.

TEST LENGTH
15 minutes, unmoderated, recorded via screen-and-audio capture.

STRUCTURE
1. Write a pre-task question that establishes the participant's expectation before they start — what do they think is about to happen — so a mismatch between expectation and reality is detectable afterward rather than only inferred from completion time.
2. Write the task prompts themselves as realistic scenarios, not literal instructions naming the UI elements to click — a prompt that says "tap the blue button" tests nothing, since it tells the participant exactly what to do instead of observing what they'd naturally try; describe the goal in the participant's own terms and let the interface do the rest.
3. After each task, write a small set of follow-up questions specifically aimed at surfacing silent confusion: what did they expect to happen when they took the key action, was there any point they weren't sure what to do next even if they figured it out, and would they have done anything differently if this weren't a test. Do not ask a generic satisfaction question here ("how would you rate this task") — a rating doesn't reveal where the hesitation happened.
4. Include an explicit instruction for the participant to think aloud, stated plainly at the start of the script, since unmoderated tests depend entirely on the participant self-reporting what a moderator would otherwise observe.
5. Tie the follow-up questions for the task most related to We're worried users won't realize SMS notifications need a separate opt-in from push notifications. directly back to that concern, phrased so they could surface disconfirming evidence, not just confirm the concern is real.

WHAT NOT TO DO
Do not write task instructions that name specific UI elements or steps. Do not write more tasks than fit comfortably in the stated test length, and do not front-load easy tasks only — include the task most related to the specific concern even if it's the hardest one.

OUTPUT FORMAT
The full script in the order above: think-aloud instruction, pre-task expectation question, then each task with its scenario prompt and follow-up questions, ending with a short debrief question set.