design-survey

Design Survey

Walk the user through designing a new survey, end to end. Output is a finalized question set + output schema, ready to hand off to /turn-into-app.

Usage

/design-survey

Optionally with a use-case hint:

/design-survey for a startup founder doing customer discovery
/design-survey for an event we're running next month
/design-survey for our gig-worker retention diagnosis

Process

Always invoke the rethink-surveys skill first. Then walk the user through these steps in order, asking ONE question per turn. Don't dump the whole flow at once.

Step 1 — Research goal (mandatory)

Ask: "In one sentence, what's the research goal? Pretend you're explaining to a busy stakeholder why this survey is worth doing."

If the user gives a vague answer ("to understand users"), push back: "Specifically — what decision will the survey help you make? If 100 responses came in tomorrow, what would change?"

Don't proceed without a sharp goal.

Step 2 — Audience

Ask: "Who's the respondent, and how will they encounter the survey? (e.g., QR code at venue, email to warm list, push from in-app)"

The answer determines the modality default:

• QR / cold → form, ≤60s
• Email / warm → form + voice, ≤90s
• In-app / engaged → AI-interviewer mode possible

Step 3 — Length budget

Ask: "What's an honest length budget? 60 / 90 / 180 seconds, or longer?"

Reference Jarrett's rule: never lie about length on the landing page. If they say 60s, they need a 60s instrument.

Step 4 — Hypotheses

Ask: "What 2–4 things are you trying to prove or disprove? Each becomes a section of the survey."

If they don't have hypotheses → suggest reading references/use-cases.md for templates with pre-formed hypotheses they can fork.

Step 5 — Match to template (if applicable)

If the use case matches event organizers / startup founders / gig-economy workers → load the template from references/use-cases.md and propose it as the starting point. Make clear they can fork/adapt.