Privacy, Data Collection, and Controls

Understand Zerocost data collection options (LLM metadata and session recording), how to enable or disable them, and privacy best practices.

Written By Harshil S

Last updated About 2 months ago

Zerocost supports optional data collection pipelines to power analytics and (when enabled) data products. You control what is collected and how much.

This is not legal advice. Use it as an engineering checklist.

We are trying to follow global data collection privacy standards like GDPR, but we are not legally compliant yet.

WE DO NOT TRACK COOKIES OR ANY PERSONAL DATA. OUR ADS ARE MATCHED BY INDUSTRY DOMAINS - NOT COOKIES.

What can be collected

Depending on configuration and settings (if enabled), collection may include:

  • LLM interaction metadata (prompt/response structure, timestamps, model info)

  • UI interaction metadata (high-level behavioral signals)

  • Lightweight session recordings (UX playback for debugging and analysis)

Recommended baseline posture

  • Default to minimal collection.

  • Use sampling instead of collecting everything.

  • Mask sensitive inputs and selectors.

  • Define retention and access controls.

Common controls and why they matter

  • Enable/disable LLM data: reduces privacy risk and storage cost. You can do this under the Datasets section in your dashboard.

  • ALL DATA COLLECTED IS PII SCRUBBED AND ANONYMIZED; REMOVING ANY PERSONAL INFO.

Operational best practices

  • Document what you collect in your privacy policy and internal docs.

  • Restrict access to raw data to a small set of operators.

  • Periodically audit stored payloads to ensure masking rules work.