The best in-app survey question is short, single-choice, and asked at the exact moment the user has the answer in their head. Timing beats wording every time: "what almost stopped you?" gets honest answers seconds after someone finishes setup, and useless ones in an email three days later. Below is a working library of questions organized by the moment that earns them.
The rules before the questions
- One question per moment. A second question halves the response quality; a form kills it.
- Single-tap answers. Predefined options plus an optional free-text "other". Typing is a favor users rarely owe you.
- Ask what you will act on. Every question should have a decision waiting on its answer.
- Skip must be effortless. A survey that blocks the product trades a data point for resentment.
- Close the loop when you can. Users who see their feedback change something answer again.
During onboarding: segment and personalize
Onboarding questions are surveys with a second job: their answers should visibly shape the experience, as covered in our onboarding best practices.
- "What brings you here?" The goal question. Drives personalization and becomes your segmentation backbone.
- "How did you hear about us?" Self-reported attribution. Imperfect, but it catches the channels attribution tools cannot see, like podcasts and word of mouth.
- "Have you used an app like this before?" Experience level. Routes beginners to guidance and experts past it.
- "What have you tried for this already?" Names the alternative you are being compared against.
After activation: catch friction while it is fresh
- "How easy was that to set up?" A 1-to-5 tap right after the first core action. Trend it by cohort against every onboarding change.
- "Was anything confusing or almost made you give up?" The single highest-yield question in the product. Ask it once, within the first session or two.
- "What were you hoping to do that you have not figured out yet?" Surfaces discoverability gaps your funnel charts cannot.
Around conversion: learn why people pay
- "What convinced you to upgrade?" Asked right after purchase, while the reason is articulable. This is your paywall copy, written by customers.
- "What almost stopped you from subscribing?" The objections that survived. Feed them back into paywall framing.
- For users who dismissed the paywall: "not for you, or not yet?" One tap distinguishes wrong-audience from wrong-moment, which demand completely different fixes.
At churn risk and cancellation: get the reason code
- "Why are you cancelling?" Predefined reasons: too expensive, not using it, missing a feature, found an alternative, technical problems. The distribution across these five drives completely different roadmaps.
- "What would have made you stay?" Free text, optional. Low response rate, but the answers that come are often the most specific feedback you will ever get.
- For dormant users on return: "what brought you back?" Tells you which win-back lever actually worked.
Periodically: measure fit, sparingly
The classic product-market-fit question ("how would you feel if you could no longer use this?") and satisfaction scores belong on a slow cadence, quarterly at most, targeted at users past their first month. They measure the product, not a moment, and lose meaning when over-asked.
Targeting mechanics: who sees what, when
A question library is only half the system; the other half is the routing. The practical setup, in order of importance:
- Event triggers over timers. "After third completed session" beats "on day 3". The event proves the user has the experience your question asks about.
- Audience filters. Segment by plan, cohort age, and behavior so churn questions never hit new users and onboarding questions never hit veterans.
- Frequency caps and cool-downs. A global "at most one prompt per session, two weeks between prompts" rule, enforced by the system rather than by every team remembering.
- Sampling. At scale, ask 10% of eligible users, not all of them. You need hundreds of answers, not tens of thousands, and the unasked users stay fresh for the next question.
- Retirement. Every survey gets an end condition: enough responses, or the decision it fed got made. Zombie surveys are how apps end up interviewing users about a feature that shipped two quarters ago.
Pipe the answers somewhere with the rest of your product data, not into a silo. "Users who answered too expensive at cancellation" is a segment you will want to target later, with different pricing or a win-back offer, and that only works when survey responses live next to behavioral events.
Ship surveys like product, not email
Everything above depends on asking in-app, at the trigger moment, in a surface that feels native to your product. Setgreet ships native in-app surveys you can target by event and user attributes, publish without an app release, and read alongside your flow analytics, so the loop from answer to product change stays short.
Frequently asked questions
What response rate should an in-app survey get?
Well-timed single-question in-app surveys routinely reach response rates an order of magnitude above email surveys. If a prompt is converting under roughly one in ten viewers, suspect the moment or the length before the copy.
How many surveys are too many?
Cap it per user, not per feature: one prompt per session at most, with cool-down windows of weeks between asks. Users forgive being asked; they do not forgive being interviewed.
Should surveys offer rewards?
For one-tap product surveys, no; incentives distort answers and attract noise. Save rewards for genuinely costly asks like user interviews or long research surveys, and keep those out of the app.