Designing Surveys
Help the user design effective surveys using frameworks from 9 product leaders who have built rigorous research and feedback systems.
How to Help
When the user asks for help with surveys:
- Clarify the goal - Determine if they're measuring satisfaction, identifying problems, or prioritizing features
- Choose the right metric - Help them select between NPS, CSAT, PMF survey, or custom approaches
- Design clean questions - Ensure each question measures one thing precisely
- Target the right respondents - Help them reach users with fresh, relevant experience
Core Principles
NPS is scientifically flawed
Judd Antin: "NPS is the best example of the marketing industry marketing itself. The consensus in the survey science community is that NPS makes all the mistakes. Customer satisfaction, a simple CSAT metric, is better. It has better data properties, it is more precise, it is more correlated to business outcomes." Use CSAT with 5-7 item scales instead.
Force prioritization with constraints
Nicole Forsgren: "Let them pick three, just three. Of those three, how often does this affect you? Is this hourly? Is this daily? Is this weekly?" Limit respondents to their top barriers to keep data clean, then measure frequency to weight impact.
Survey your best customers at the right time
Gia Laudi: "Very importantly, they signed up for your product recently enough that they remember what life was like before. Generally, we say that's in the three to six-month range." Target customers who have been using the product 3-6 months so their memory of the 'before' state is fresh.
Onboarding surveys improve conversion
Laura Schaffer: "We just asked for forgiveness and put these questions into the signup flow. An improved conversion by like 5%, just improved signups." Adding 'good friction' in the form of targeted questions can increase conversion by reassuring users they're in the right place.
Avoid double-barreled questions
Nicole Forsgren: "You're asking four different questions there. If someone answers yes, was it the build? Was it the test? Was it slow or was it flaky?" Ensure each survey question only asks about one specific variable.
Use MaxDiff for feature prioritization
Madhavan Ramanujam: "Identify the most important for you, and the least important. If you do this a few times, you will be able to prioritize the entire feature set in a relative fashion." MaxDiff (Most/Least) surveys are superior to simple ranking for identifying value drivers.
Questions to Help Users
- "What specific decision will this survey inform?"
- "Are you asking about one thing per question, or multiple things?"
- "Who are your 'best' customers and when did they sign up?"
- "Are all scale options visible on mobile without scrolling?"
- "How will you force respondents to prioritize rather than rate everything high?"
Common Mistakes to Flag
- Double-barreled questions - Asking about speed AND complexity in one question
- Too many options - Allowing respondents to select unlimited items instead of forcing prioritization
- Wrong timing - Surveying customers who are too new (no experience) or too old (forgot the 'before')
- NPS worship - Relying on a metric with known scientific flaws over simpler, better alternatives
- Hidden scale options - Mobile surveys where users can't see all options create response bias
Deep Dive
For all 10 insights from 9 guests, see references/guest-insights.md
Related Skills
- Writing North Star Metrics
- Defining Product Vision
- Prioritizing Roadmap
- Setting OKRs & Goals
