What Mentoor collects
Only the signals it needs to draft and improve guides on the visible UI:
- Route + page titleso guides know which screen they belong to.
- Visible labels, roles, and stable anchorsso a guide can point at the right button after a redesign.
- Guide eventsstarted, completed, dismissed — to know which guides work and which need a rewrite.
- Visitor id and traits you passonly the values your code provides via identify() / setTraits().
What Mentoor never collects
These are explicit exclusions in the SDK — they cannot be turned on:
- Screenshots or pixel datawe never capture the rendered page.
- Typed field valuescharacters in inputs are not recorded during product learning.
- Passwords, tokens, payment infosensitive form fields are never read.
- Database contentMentoor has no path to your backend.
Where the data lives
Guide events, anchor inventory, and visitor identifiers are stored in Mentoor’s database on managed infrastructure inside the EU, encrypted at rest (AES-256) and in transit (TLS 1.3).
The browser SDK also keeps small caches in localStorage so returning visitors don’t see the same guide twice. You can disable that by calling setConsent({ storage: false }).
Retention
Product-learning signals are kept for 90 days, then aggregated and dropped.
Published guides and their event history are kept for the lifetime of your account and deleted within 30 days of contract termination.
Visitor rights
You are the data controller for your end users. Mentoor exposes APIs so you can fulfil access / deletion / export requests on behalf of any visitor identified through identify().
For requests directed at Mentoor itself, email [email protected]. We respond within 30 days.
Locking surfaces down
Mark any region with data-mentoor-ignore="true" and Mentoor will skip it entirely — no observation, no guide steps, no agent actions.
For full compliance environments (healthcare, finance), set privacy: "essential" in init — Mentoor runs reviewed guides only and turns passive learning off.