Images in this article were generated by AI and may contain imperfections. Real application screenshots will be added soon.
This article may describe features that are still in development. If you need any of them prioritized, please contact us via the contact form (applies to Business subscription or higher).
How JSON/RRWeb Export Works
RRWeb is an open-source standard for recording and replaying web sessions. All our recordings are internally stored in this format. JSON export gives you direct access to the same data we use for playback.
Each session exports as an array of events: DOM snapshots, mutations, mouse movements, clicks, scrolls, and form events. The format is fully documented and supported by the rrweb-player library for playback.
What you can do with the data: replay sessions in your own tools, programmatically analyze behavior patterns, feed ML/AI models, integrate with your own analytics systems, long-term archive outside the platform.
LLM Analysis: The JSON format is ideal for processing by large language models. You can send session events to Claude, GPT-4, or other LLMs and ask for user behavior analysis, UX problem identification, or insight generation.
![[img: horizontal illustration showing JSON code fragment with RRWeb event structure, visible event types and timestamps, technical documentation style]](/images/features/json1.webp)
Real-World Use Case
An e-commerce company wants to understand why specific users abandon their carts. They export sessions with abandonments as JSON and send them to Claude with the prompt: "Analyze these sessions and identify common behavior patterns before abandonment."
The LLM analyzes event sequences and finds a pattern: users who spend >30 seconds on the delivery selection page and then leave. Analysis reveals that delivery options are unclear for customers from smaller cities. Solution: better option explanations. Abandonments drop by 18%.
![[img: horizontal illustration showing data flow: JSON export → LLM processing → generated insight report, AI integration diagram]](/images/features/json2.webp)
Frequently Asked Questions
What is the RRWeb standard?
RRWeb (Record and Replay Web) is an open-source project defining the serialization format for web interactions. It's widely used in the session replay industry and supported by an active community. This means exported data can be played by any rrweb-compatible tool.
How large are JSON export files?
Size depends on session length and page complexity. A typical 5-minute session is 1-5MB JSON (gzip compressed: 100-500KB). Long sessions on complex pages may be larger. We offer an option to export without DOM snapshots for smaller files.
Can I replay exported JSON locally?
Yes. Use the official rrweb-player library (available via npm) to replay exported sessions in any web application. You can also build your own player or use other rrweb-compatible tools.
Does the export contain sensitive data?
Export respects masking rules configured in your project. Masked fields are exported as masked (e.g., "****"). You can additionally filter events during export or process the JSON before sharing.
How do I send a session for LLM analysis?
Export the session as JSON, then upload the file to the model API (Claude, GPT-4) or paste events into the chat interface. For large sessions, you can filter only selected event types (e.g., clicks and navigations). LLMs understand the event structure and can analyze patterns.
