A crash in production that doesn't reproduce on test devices is a common pain for mobile teams. Session Replay provides an accurate recording of user actions seconds before the error, reducing root cause search time by 2–3x. Across 15+ projects (from fintech to e-commerce), we've developed practices to record sessions with less than 1% overhead and full masking of personal data, compliant with GDPR and PCI DSS.
What is Session Replay and why do you need it?
Session Replay is a technology that captures all user actions: taps, scrolls, text input. The resulting recording allows you to reproduce the session exactly, which is critical for debugging errors and UX analysis. Unlike logging, it provides a complete visual picture. The recording works in the background without user intervention.
Screenshot or Wire-frame: what to choose?
Screenshot-based (UXCam, Smartlook) takes screenshots at 1–5 fps, masks sensitive areas, and sends them to the server. It accurately captures custom Views and WebViews but requires 200–500 KB of traffic per minute and loads the CPU by 3–8%.
Wire-frame based (Sentry, Datadog) serializes the View hierarchy — positions, colors, text — and reproduces the UI on the server using templates. Data volume is 50–150 KB/min, CPU load < 1%. WebView and complex graphics are not transmitted accurately. The choice depends on priority: accuracy or performance.
| Parameter | Screenshot-based | Wire-frame based |
|---|---|---|
| CPU (background) | 3–8% | < 1% |
| Traffic (per min) | 200–500 KB | 50–150 KB |
| Accuracy | High (all elements) | Medium (WebView missing) |
How does Session Replay help debug crashes?
Sentry SR for mobile is available from SDK 8.x, using the wire-frame approach. The key parameter onErrorSampleRate = 1.0 — record replays for all sessions with errors. The SDK buffers the last N seconds in memory and sends them with the error report. More details in Sentry Session Replay documentation.
// iOS import Sentry SentrySDK.start { options in options.dsn = "https://[email protected]/project" options.experimental.sessionReplay = SentryReplayOptions( sessionSampleRate: 0.1, onErrorSampleRate: 1.0 ) } // Android SentryAndroid.init(this) { options -> options.dsn = "https://[email protected]/project" options.experimental.sessionReplay.apply { sessionSampleRate = 0.1 onErrorSampleRate = 1.0 } } How to configure masking of sensitive data?
By default, Sentry masks UITextField and fields with isSecureTextEntry = true. That's not enough — you need to hide card numbers, personal data, OTP fields. Example for iOS and Android:
// iOS — mark a View for masking class PaymentCardView: UIView { override func didMoveToWindow() { super.didMoveToWindow() SentrySDK.replay.maskView(self) } } // Android — masking via tag val cardNumberField = findViewById<EditText>(R.id.cardNumber) cardNumberField.setTag(io.sentry.android.replay.Recorder.MASK_TAG, true) For SwiftUI and Jetpack Compose, use modifiers:
// SwiftUI Text(userEmail).sentryReplayMask() // Kotlin Compose Text(text = cardNumber, modifier = Modifier.sentryReplayMask()) We guarantee that masking passes a privacy audit compliant with GDPR and PCI DSS. Verification is done via Privacy Audit in the Sentry/Datadog UI.
Integration with crash reports
In Sentry, the replay is automatically attached to the error report. In Datadog Session Replay, it's linked to the RUM View — you can open the screen and view replay with metrics (latency, FPS) on the timeline.
Implementation checklist
- Choose the tool (Sentry / Datadog) based on stack and budget.
- Integrate SDK and configure sample rates (onError = 1.0, session = 0.1).
- Mark all sensitive screens and fields (masking).
- Conduct a Privacy Audit.
- Integrate with existing crash reports.
- Document configuration and hand over to the support team.
Tool comparison: Sentry vs Datadog
| Criterion | Sentry Session Replay | Datadog Session Replay |
|---|---|---|
| Recording mode | Wire-frame (default) | Wire-frame + screenshot |
| Masking | Automatic + manual | Automatic (maskUserInput) |
| Error integration | Automatic (error report) | Via RUM |
| CPU load | < 1% | 1–3% (wire-frame) |
Sentry wire-frame outperforms UXCam screenshot mode in CPU by 2–3x. If you need accurate WebView recording, Datadog in screenshot mode (CPU 3–8%) may be suitable.
What's included in our work on Session Replay implementation
When ordering our service, you get:
- Audit of the existing app: identification of sensitive screens and fields.
- Selection of the optimal tool and approach (screenshot/wire-frame).
- SDK integration with sample rate and masking configuration.
- Custom masking development for specific UI elements.
- Privacy Audit with a report.
- Integration with your existing error monitoring system (Sentry/Datadog).
- Documentation on configuration and recommendations for your team.
- Support during testing and deployment.
Session Replay implementation process
Implementation goes through 5 stages:
- Analysis — identify screens and sensitive fields, choose the tool (Sentry or Datadog).
-
SDK integration — configure
onErrorSampleRate = 1.0,sessionSampleRate, DSN. - Masking setup — mark all input fields, card data, OTP.
- Testing — verify masking, run Privacy Audit.
- Deployment — release to stores, monitor metrics.
Example: for a fintech project, we configured masking of 15 screens in 2 days. After the audit, we confirmed that no sensitive character was captured in the recordings.
Timelines and cost
Basic integration with masking: 2–3 days. Full integration with privacy audit: 4–5 days. Cost is calculated individually. Time savings on crash debugging after implementation can reach 60% — for a team of 5 developers, that can mean significant monthly savings.
Contact us for a consultation on integrating Session Replay. Get an estimate for your project and tool recommendations.