Real people and what they say
How do I practise my questions?
Why it works this way
Practise your questions on simulated people first, and keep what they say apart from what real people tell you.
Practising on simulated people can catch a leading question before it wastes a real person. But a simulated answer is not testimony, and the screens keep that clear. So practice and evidence live in different places.
What we gave up: real quotes sitting beside simulated answers on the test page. A real quote and a simulated answer sit in separate lists.
What it is
A weak question costs you a real conversation. So before you speak to anyone, you can run your questions past simulated people, who are AI, and see which questions will not serve you. The same screens keep what a simulated person says clearly apart from what a real person says.
How to use it
- Open a test that has questions, such as an interview. Press Guide, then Instructions. Your questions are under For this test.
- Press Simulate interview. A few simulated people answer each question.
- Read each question's verdict. A weak question is marked with its reason, for example a question that asks people to predict what they would do, or one that hands them the answer. A good one says "Fine as it is".
- Press Use the rewritten questions to accept the rewrites in one click, or edit them yourself (Edit, then Done, and Add a question). Your guide updates.
- Press Simulate again after you change them.
- Later, a point a simulated person made can be put on To check with real people (press Check with real people under their answer). It sits there marked "From [name], simulated · not heard yet from a real person" until a real conversation says it too. Then it shows the real person's own words: "Heard from [who]".
- What real people wrote is a separate list, marked Real, with a link to each source. These are quotes your research found. They are real people, but not your conversations, so they do not count toward your test.
What it will never do
- Judge your customers. It checks your questions. What a simulated person says about customers is not a finding.
- Tick a point off because a simulated person said it again. Only a real person's words tick it.
- Change your questions without you. Accepting a rewrite is your click.
If something looks off
- "There are no questions to simulate yet. Add one first." Add a question, then run it.
- "Could not run the simulated interview just now." Try again in a moment. Your questions are untouched.
- "Every question here has changed since the simulation, so none of these verdicts still applies. Run it again." Press Simulate again.
- "Nothing to fix…" The check found nothing to change in your questions.
- A point stays waiting. No real person has said it yet. That is the point of the list.