Session Replay
Contents
Session replay records what real users do in your product and plays it back like a DVR, with a synced DevTools panel showing console logs, network requests, and errors at the exact moment they happened. Because PostHog already captures your events, every replay is enriched with the person, their properties, and the feature flags they had, so you can jump from a funnel drop-off or an exception straight to the session behind it.
You can watch replays yourself, ask PostHog AI to find and summarize them, or pull them into your editor over MCP. Those same sessions feed Self-driving, which clusters the problems users actually hit into reports you can act on.
Where you can use it
You watch, filter, and share replays in the PostHog web app. The other surfaces let you investigate sessions and act on them without leaving your existing tools.
PostHog Web
Watch recordings with synced DevTools, build collections, and share them with your team.
Watch replays →PostHog MCP
Search sessions, read summaries, and manage playlists from any MCP client or AI editor.
Search replays →PostHog Desktop Beta
Read the Self-driving reports built from clustered session problems, and decide what to do about them.
Review findings →Where its data comes from
Session replay runs on recordings captured by your PostHog SDKs. What ends up in a replay depends on how you configure capture: which sessions you record, what the browser panels collect, and what you mask before it ever leaves the device.
Recording rules
Sample, gate by URL or feature flag, and record only the sessions worth keeping.
Set recording rules →Console and network
Capture console logs and the network waterfall so the DevTools panel has something to show.
Capture performance data →Privacy and retention
Mask text, inputs, and whole elements before capture, and control how long replays are kept.
Configure privacy →How Session Replay works with Self-driving
Your recordings are a signal source for Self-driving. PostHog analyzes sessions for the moments users get stuck – confusion, abandonment, failures, and blocking exceptions – then clusters those moments across sessions so a pattern, not a single unlucky user, becomes a report. The same scout watches for capture cliffs, where recording volume falls away while your traffic holds.
A replay shows the symptom, not the cause. Rage clicks on one button could be a slow request, a confusing label, or a real bug, and telling those apart takes a person – so this scout stops at the report and routes it to a human reviewer rather than starting a pull request.
Session replay analysis relies on AI data processing, so your organization has to approve it before the source can be turned on. You read the resulting reports in PostHog Desktop or the Self-driving inbox. Other signal sources, like error tracking, do open pull requests.
Want AI to watch recordings and turn what it sees into queryable data instead? That's Replay Vision, which runs on the same recordings.