Data Anonymization

Your users sometimes share personal details in open-text survey answers or AI copilot conversations — things like their name, email, or phone number — even when you never asked for it. Data anonymization automatically detects and removes that information before it is saved anywhere in Screeb.

This feature is designed for teams working in regulated industries (healthcare, finance, insurance, etc.) or any organization that wants a clear guarantee that personal data does not end up in their survey results.

Why it matters

When a user types a free-text answer like "Hi, I'm Sarah Johnson, you can reach me at sarah@example.com", that personal data would normally be stored as-is in your survey responses. With data anonymization turned on, what gets saved instead is "Hi, I'm *, you can reach me at *" — the personal details are gone before they ever touch Screeb's database.

This helps you:

• Meet GDPR, HIPAA, and similar data minimization requirements
• Reduce your exposure in case of a data breach
• Reassure users that their personal details are not being collected

What it covers

Data anonymization applies to the text your users write themselves:

• Survey open-text answers — any question where the user types a free response
• AI copilot conversations — the messages your users send to a copilot, and the responses it sends back

It does not apply to:

• User profile properties you set in your own code (name, email, plan, etc.) — those are values your team provides intentionally and are needed for targeting
• Structured answers like ratings, NPS scores, or multiple-choice selections — there is no free text to anonymize
• Responses collected before the feature was enabled — anonymization is not retroactive

What gets redacted

Personal information is replaced with ****. Screeb detects the following types automatically:

How the protection works

Anonymization happens in two steps, one on top of the other.

Step 1 — Before the answer leaves the browser. The moment a user submits a response, Screeb scans the text for recognizable patterns (email addresses, phone numbers, etc.) and replaces them with ****. This happens entirely on the user's device — the raw text never leaves the browser.

Step 2 — On Screeb's servers, using AI. Once the cleaned answer arrives, an AI model takes a second pass to catch anything more subtle: full names, postal addresses, or any personal information that does not follow a predictable pattern. This step is invisible to the user and does not slow down the survey experience.

Together, these two steps provide strong, defense-in-depth protection.

A note on timing for survey answers

The AI step runs in the background after the answer is saved. There is a short window — usually a few seconds — where the answer is stored as "answer being processed" while the AI is working. Once it finishes, the final anonymized text replaces that placeholder. If the AI is temporarily unavailable, Screeb will retry automatically. In the rare event that all retries fail, the answer is stored as "anonymization failed" rather than keeping the original text.

Things to keep in mind

• "Save to user property" questions are skipped. If you have a survey question configured to save the user's answer directly to their profile, that automatic save is disabled when anonymization is on. Saving anonymized text (full of ****) into a user profile would not be useful.
• It is not retroactive. Responses collected before you turned the feature on are not changed.
• It is a best-effort system. Screeb catches the vast majority of personal data, but no automated system is 100% accurate on all languages and writing styles.

Enabling data anonymization

Turn the feature on from your workspace settings page in the Screeb admin interface. Once enabled, it applies immediately to all new submissions across your surveys and copilot integrations.

Updated on: 29/04/2026