Step 8 of 11 — Surveys
When (and when not) to run a survey
Surveys are for confirming patterns at scale, not discovering them. Run interviews first, find a pattern, then survey 50 people to check if it holds. Skip this order and your survey will ask the wrong questions.
Why this matters
Interviews tell you WHY. Surveys tell you HOW MANY. Both are useful, but surveys are useless as the first move because you don't yet know which questions matter. Once interviews have surfaced 2–3 patterns ("most say price is fine but they don't trust strangers online"), a survey to 50 people confirms whether that's 80% of your audience or 20%. That difference changes your strategy.
What you'll do in this step
- Pick one of 3 survey types: Customer & Problem Discovery, Solution & Idea Validation, or Pricing & Commitment Signals — the same taxonomy used in Step 7 interviews.
- AI generates 5–7 questions based on your unresolved assumptions — a mix of multiple choice, 1-10 scales, yes/no, and willingness-to-pay ranges.
- Edit the questions (this is your survey, not ours), add or remove questions, reorder by dragging.
- Save as draft, share the public link via WhatsApp, email, college groups. Watch responses come in.
A real example
Ananya ran 8 interviews and noticed: students like the idea of senior mentors but worry about "what if the senior is a stranger?" Now she runs a survey to 50 students to confirm.
Question 1 — Problem check (yes/no)
"Have you applied to an internship in the last 6 months and not received a response?" — Confirms the baseline problem at scale.
Question 2 — Solution preference (multiple choice)
"When applying, which would help you most? (A) Professional resume review. (B) Talking to a senior from a similar college who got an internship. (C) Sample cover letters. (D) Guaranteed interviews." — Now you learn 70% pick B, 20% pick A. That's strategy.
Question 3 — Trust (1–10 scale)
"If we connected you with a senior you'd never met, on a scale of 1–10 how comfortable would you be paying for that?" — Reveals the trust-gap you heard in interviews.
Question 4 — Pricing (willingness-to-pay ranges)
"What's the most you'd pay for a 30-min call with a verified senior? Nothing / Under ₹100 / ₹100-300 / ₹300-500 / Over ₹500." — Now you have a distribution, not a guess.
Common mistakes to avoid
- Running a survey first, interviews never — you'll ask shallow questions and get shallow answers.
- Writing questions based on the answer you want — "would you pay a reasonable price for this amazing feature?" is leading.
- Ignoring the open-text answers — the juiciest insights are usually in the 2 people who wrote a paragraph, not the 48 who ticked a box.
How Margawise helps
- 3 survey types match your 3 most critical assumption buckets.
- AI generates type-specific questions from YOUR unresolved assumptions, not templates.
- Force-regenerate bypasses the cache when you want a fresh set of questions.
- Public share link works anywhere — WhatsApp, email, social, college groups.
- Coverage indicator shows which survey types you've run so you can spot gaps.
Ready to try this in your own project?
Open Surveys stepFree to start · No credit card required
Explore other steps
Have feedback on this article? hello@margawise.com