A few days after we changed the onboarding flow, we noticed a drop in activation, but the main dashboard didn’t make it clear what had actually gone wrong. Traffic was still growing, signups were still coming in, and the problem wasn’t affecting every user. To dig in, we built a Funnel in Product Analytics from signup_completed to workspace_created and first_project_created, and then segmented the results by plan and device. The loss was concentrated among mobile users coming from a specific campaign. From that same funnel step, we opened the related Session Replays and watched people select an option, try to continue, and then miss the button because it stayed below the visible part of the screen. That direct link between the metric and the real user experience is where PostHog becomes especially useful for us. In other setups, I can see that conversion dropped, but then I have to copy identifiers, hunt for sessions in another platform, and hope I’m still looking at the right cohort. Here, we can go from a segment in Trends, Funnels, Retention, or Paths straight into the relevant recordings without having to rebuild the context manually. Autocapture also gives us a helpful baseline for pageviews, clicks, and form submissions before we decide which events are important enough to instrument properly.
PostHog is one of the most useful tools for understanding what users actually do inside a product. I really like that analytics, funnels, session recordings, feature flags, experiments, surveys, and dashboards are all connected in one place, rather than being split across several different tools. The session replay feature is especially valuable because it lets us observe real user behavior, spot friction points, and better understand where and why users drop off. It also makes product and marketing decisions much easier, since we can tie user actions to clear data instead of relying on guesswork. Overall, PostHog provides a very complete view of the customer journey and helps teams iterate and improve the product faster.