Flutter session replay lets you watch a real session on a real device — the tap that did nothing, the spinner that never resolved, the form that cleared itself on submit — instead of guessing from a stack trace. On the web that is a familiar purchase. On mobile it arrives with a harder question: a phone screen carries names, addresses, card digits and message previews, and a widget tree gives you few obvious places to stop recording.
This guide is the practical version: what Replay captures in a Flutter build, the privacy decisions to make before the first session, how to attach sessions to accounts you already own, and how to review them without watching hours of video.
What a Flutter session replay records
In a Flutter app you get playback of the rendered session plus the events that explain it: taps and gestures, route changes, rage clicks (repeated taps that hit nothing), dead and error clicks, and heatmaps for where people actually press. A feedback widget, share links, comments and export let a bug report carry the session instead of a description.
Two honest limits. Browser signals — Web Vitals, JavaScript errors — describe the web, not your Flutter build. And native mobile replay covers iOS, Android, React Native and Flutter in one product, so you keep one review workflow instead of one tool per stack.
Masking is decided at capture, not at review
Most replay incidents are not caused by someone watching sessions maliciously. They happen because a field that looked harmless in a test account held a national ID number or a one-time code in production.

Prynt Replay masks every form field at capture on the device and masks it again on the server. There is no unmask switch — not for you, not for anyone on your team. That is a deliberate tradeoff: you give up seeing exactly what someone typed into an address form, and in exchange no reviewer has to be trusted to behave. If you need to know what was submitted, read it from your own database, where access is already logged and scoped.
The same rule covers the wrapper around the screen. A screen name of “Checkout” is useful; “Checkout – jane@…” is a leak with a dashboard on top. Our form-protection guide makes the same case from the web side.
Wiring it into a Flutter app
Mask by default, not per screen
Inherit masking rather than enabling it screen by screen. If a view shows data you would not paste into a ticket, assume it is recorded and name the screen and its metadata accordingly.
Attach identity only where you already have it
Anonymous sessions are useful; signed-in sessions answer “which customers hit this bug”. Once your app knows the user, link the session to identifiers you already own:
identify(userId, { accountId })
Use internal IDs — never an email, phone number or token. The call shape and the mobile setup steps live in our docs.
Keep bots and synthetic traffic out of the sample
Bot sessions do not count against your Replay quota, which keeps automated builds from eating the sessions you wanted to watch. Separate app IDs per environment keeps test devices out too.
Reviewing sessions without watching all of them
Enter Replay with a question. Filter to sessions on the screen where the complaint started, sort by rage clicks to find interactions that failed silently, and attach the share link to the ticket instead of writing a summary.

For triage, Py — the assistant included with Replay — summarizes a session, explains what went wrong and sends a daily brief. It does not replace reading a session when a bug is subtle; it stops you from opening fifty to find the three that matter. Py has a monthly limit per plan.
Then close the loop where the team already works: sessions connect to Sentry, Zendesk and Intercom, so a support reply and the evidence behind it stay in one thread.
Sessions inside your own admin
Support should not need a second dashboard, and your customers should not need a login to another vendor’s tool. Replay lets you list sessions with your secret key and open them in an embedded player, scoped per company or per user — an account page in your own admin can show that customer’s sessions and nothing else. The embedding flow is documented in the replay section of our docs.
Because the identifiers are yours, the boundary is yours: the accountId you passed at identify time is the boundary the player uses. Write that mapping into your data processing agreement before a customer asks — mobile replay reaches security questionnaires quickly.
Retention and plan size
Free includes 1,000 sessions a month with 14-day retention on one site — enough to validate the setup on a staging build and a thin slice of production. Pro is $279 a month for 10,000 sessions, 30-day retention, 3 sites and a 14-day trial; volume plans reach 1M sessions. Retention is what teams get wrong first: 14 days forces triage, 30 days lets a bug reported after a release be traced back to it.

FAQ
Does session replay record what users type?
No. Every form field is masked at capture on the device and masked again on the server, and there is no unmask option in the interface. Text you need for debugging should come from your own logs.
Does replay count bot or automated traffic?
Bot sessions are excluded from your Replay quota, so automated traffic does not consume the sessions you care about. Test your build with replay enabled on a low-end Android device before rollout — that is where instrumentation shows up first.
Is a separate consent flow required for mobile replay?
That depends on your jurisdiction, your user agreement and what you record, and we do not give legal advice. The engineering floor is simple: mask by default, keep personal data out of metadata, set an explicit retention window and document the mapping in your DPA.
Try Prynt Replay
Start on the Free plan with one Flutter app: 1,000 sessions a month, 14-day retention, masked by default. Watch a real session, then decide what your policy should be — pryntid.com.
Try it free
Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.