Trust & safety
Verification you can inspect
Resuvouch's value rests on recruiters believing what they read. That only works if we show our checks — including their limits — instead of asking you to take our word for it.
How identity is checked
Every reference has an identity level, shown to recruiters and used in the weighting. There are three:
Verified company email
StrongestThe reference proved they control an email address at the company's domain: we send a 6-digit code to that address and they enter it within 10 minutes. It shows someone at that domain really answered — without us calling anyone.
Google sign-in
MiddleThe reference signed in with a Google account. Google has verified the address; we haven't tied it to a company.
Email link only
WeakestThe reference proved only that they can read a personal inbox. Their answers still count, at a lower weight — and recruiters see this level plainly.
Trust flags: what we look for
We flag four patterns automatically:
Reciprocal pairs
Two people vouched for each other. Sometimes legitimate — colleagues do overlap — but worth knowing.
Sign-up clusters
Several reference accounts were created close together in a pattern that suggests coordination.
Single-purpose accounts
An account that exists only to answer for one person and has never done anything else.
No tenure overlap
The dates the two people say they worked together don't actually overlap.
The policy
- Flags are shown neutrally, in plain language, to recruiters — and to the candidate. Nobody is flagged behind their back.
- Our staff review flagged patterns.
- Flags never auto-ban anyone. A flag is information for a human reader, not a verdict.
How data is protected
Plain facts about what the system does — nothing here is a marketing adjective:
Tokens are never stored raw
Every invite and share link token is stored as a SHA-256 hash only. If our database leaked, the links in people's inboxes would not.
Sensitive text is encrypted
Reference comments and answers not yet attached to an account are encrypted with AES-256-GCM before they touch the database.
Uploads are scanned
Files are checked and scanned before they are ever served to anyone.
Row-level security
The database itself enforces who can read which rows — access control isn't only application code.
Strict content security policy
No inline scripts run on any page, which closes off the most common class of script injection.
Append-only audit log
Sensitive actions are recorded in a log that can be added to but not rewritten.
Rate limits
Invites, sign-in attempts, and reports are all rate-limited to blunt abuse and scraping.
What we collect and why lives in the privacy policy; what references specifically agree to is in the reference consent notice. Found a vulnerability? Report it here.
Report abuse
- Reading a share link? Every share link has a built-in “report a concern” page — no account needed.
- Got an invite from someone you don’t know? The invite email has a “don’t know this person?” link. More in the reference guide.
- Anything else: use the support form.