How would you improve PostHog Surveys for product teams collecting in-app NPS and feedback?

Introduction

PostHog Surveys lets teams launch in-app popups for NPS, CSAT, multiple choice, ratings, emoji reactions and open text, either with no code or through the API for full control. The first 1,500 responses each month are free, and because Surveys sits on the same data as Product Analytics and Session Replay, every answer is tied to a real user's behaviour. That link is Surveys' biggest advantage over standalone tools like Typeform or Hotjar. Yet many teams use it like any other survey tool: they launch an NPS popup, collect scores, and never act on the comments. I'll improve Surveys for product teams collecting in-app NPS and feedback, moving from clarifying questions to segments, pain points, solutions, prioritisation and metrics.

Step 1

Clarifying Questions (5 mins)

• What is the main problem we are solving: low response rates, low-quality answers, or answers that are never acted on?

Why it matters: Each points to a different part of the product. Expected answer: Response rates are acceptable (around 8–12% for in-app NPS), but teams struggle to turn responses into decisions. Impact on approach: Focus on analysis and action, not only on collection.

• Who are the main users of Surveys today?

Why it matters: Defines whose workflow we improve. Expected answer: Founders and product engineers at startups, plus PMs and researchers at growth-stage companies. Impact on approach: Solutions must work without a dedicated researcher.

• Is there an improvement goal tied to PostHog's business?

Why it matters: Aligns the work with company outcomes. Expected answer: Grow paid Surveys usage and increase the share of teams using two or more PostHog products. Impact on approach: Favour improvements that connect Surveys with Replay and Analytics.

• Are there constraints such as privacy or pricing?

Why it matters: Text answers can contain personal data, and pricing is per response. Expected answer: Keep per-response pricing; respect existing masking rules. Impact on approach: Avoid features that inflate low-value responses just to raise revenue.