Unmoderated test builder
What it does
An unmoderated test puts a task in front of real people with nobody watching live. They work through it on their own time, their screen and voice are captured, and you review the recordings afterwards. It goes by several names: unmoderated usability testing, async testing, self-guided testing, and remote unmoderated research. The method lives or dies on the task wording. A task that hints at the answer produces a room full of people succeeding at something you never learned anything about.
This skill builds the study for you inside Great Question. You give it the thing being tested, whether that is a rough concept or a working prototype, along with what you are trying to find out, and it assembles the blocks: a prototype test pointed at a Figma prototype URL, a website test pointed at the live URL under test, card sorts and tree tests where the question is about structure, and the question blocks that sit around them. You open it, run through it as a participant would, and fix anything that reads badly before it goes out. It works from Claude, Cursor or ChatGPT with the Great Question MCP connected.
What you get back
- An unmoderated study created in your Great Question account, unsent.
- Task wording written from the goal you described, phrased to avoid leading.
- Question blocks in the format each question needs: linear scales with your own minimum, maximum and per-value labels, single and multi select with an optional free-text other and a none of the above, yes/no, and open text.
- A card sort set up open, closed or hybrid, with card and category order randomised and every card required to be sorted if that is what you want.
- A tree test built as a nested tree of any depth, with the correct answers marked and sibling order randomised.
- Screener criteria for the audience you named.
How to use it
Example Prompts:
"Build an unmoderated test of this checkout prototype, focused on where people hesitate at the payment step."
"Turn this task list into an unmoderated study for mobile customers, five tasks, open-ended follow-up after each."
"Make an unmoderated version of the navigation study we ran in March."
What it won't do
It won't host your prototype or fix a broken link inside it. Welcome, thank you, permissions and AI-moderated interview blocks come with the study itself and cannot be added afterwards. A prototype block also pulls its screens from the design tool in the background, so they are not there the instant the block appears. Read the task wording as a participant would before it goes out.
Questions
Can it test a live product as well as a prototype?
Both. A website test block takes the live URL under test, and a prototype test block takes a Figma prototype URL. The rest of the study, tasks and follow-up questions, gets put together the same way in either case.
Does it write the tasks or do I?
It drafts them from what you tell it you want to learn, then you edit. Task wording is where unmoderated studies go wrong, so the draft is a starting point rather than a finished script.
Do I need to build the study first and then recruit?
The study gets built first, yes. Screening your own panel against it is handled by the recruiter skill, which reads the criteria on the study you just built.
