Server-Free Considerations

Using rex-local-download without a server is a valid approach for many studies, but there are trade-offs worth understanding before you commit to it.

Data collection risks

Data loss. If a participant uninstalls the extension, switches browsers, or resets their browser profile before uploading, their data is gone. There is no recovery path.

No real-time visibility. You have no way to know whether participants are actively collecting data until they upload. You cannot identify drop-off until the study ends.

Participant burden. You are asking participants to take an active step (download and upload the file) rather than having data flow automatically. This adds friction and is a point of failure.

Upload compliance. Some participants may forget, delay, or submit incomplete exports. Follow-up is harder than with server-based collection where you can see who has submitted data.

Data security risks

Data in transit. When participants upload their JSON to Box, Google Drive, or another service, the data passes through that service’s infrastructure. Make sure the service meets your IRB and institutional data security requirements.

Participant error. Participants may upload to the wrong folder, share a link publicly, or send the file via insecure channels if not given clear instructions.

No encryption at rest in the extension. Data stored in the extension is accessible to anyone with access to the participant’s browser profile.

When server-free is fine

• Pilot studies where data loss is acceptable
• Small studies with known, engaged participants you can follow up with directly
• Studies where the data is low-sensitivity and the upload step is manageable
• Situations where setting up a server is not feasible and the trade-offs are acceptable

When to use a server instead

If any of the following apply, consider rex-passive-data-kit with a backend:

• Large participant pools where individual follow-up is not practical
• Studies requiring real-time compliance monitoring
• Sensitive data requiring controlled handling throughout the pipeline
• Long study durations where participant drop-off is a real risk