A Candidate Asked to Stop Contact. Your Other Tools Did Not Listen.

The second email that should never have been sent

A candidate replies to a sourcing email: “Please stop contacting me.” The recruiter marks the conversation as closed in the email tool. Three weeks later, an agency partner uploads the same person under a different address, and an automated nurture campaign sends another invitation. The candidate sees one organization that ignored a clear request; the recruiting team sees two records in separate systems. ^↗

The root problem is treating an opt-out as a free-text note rather than a structured event that downstream systems must honor. The precise legal basis and obligations vary by jurisdiction and communication type, but the operational requirement is universal: a process cannot respect a preference it does not propagate. ^{↗ ↗}

Map every route by which a person can be contacted

Inventory ATS messages, recruiting CRM campaigns, calendar reminders, sourcing extensions, employee referral workflows, agency exports and event platforms. For each system record the person identifier, opt-out field, ingestion path, sync frequency, campaign scheduler and person responsible for fixing failures. Do not assume a CRM flag automatically blocks ATS or agency outreach. ^{↗ ↗}

Separate a general recruiting-marketing opt-out from essential transactional communications about an active application. The distinction should be reviewed with privacy counsel and explained in plain language. A candidate who asks to stop promotional outreach may still need a requested interview confirmation; a candidate who requests no further contact may require a different response. ^↗

Build a minimum-necessary suppression record

A suppression list does not need to retain an entire resume, interview history or private message thread. A minimal record might contain a normalized contact token, preference scope, timestamp, source system, applicable jurisdiction and audit identifier. Protect it with restricted access and a defined retention policy. Hashing alone is not a complete privacy solution if tokens remain linkable. ^{↗ ↗}

For duplicate identities, do not automatically merge people based on a shared name. Use verified contact channels or controlled identity resolution and provide a way to correct mistakes. A false suppression can also harm candidates by blocking wanted application updates. ^{↗ ↗}

Treat synchronization failure as a measurable incident

When an opt-out arrives, publish a suppression event to connected systems and require an acknowledgement. If an integration is offline, hold outbound campaigns rather than assuming delivery succeeded. Keep a retry queue and alert on unresolved failures. Campaign sends should check suppression at send time, not just when a list is assembled. ^↗

Run a weekly reconciliation across tools using a privacy-preserving test set. Count records with an opt-out in one system and no corresponding state elsewhere. Track time to propagation, unacknowledged events and messages sent after an opt-out. A dashboard showing “99% synced” is not enough if the remaining 1% contains repeated unwanted contact. ^{↗ ↗}

The agency and vendor handoff

Contracts and operational runbooks should define how preferences are communicated to agencies, subprocessors and campaign providers. Do not send full candidate profiles merely to update a suppression status. Specify a secure channel, expected acknowledgement time, escalation owner and deletion/correction workflow. ^↗

If a vendor cannot accept a suppression update, treat that limitation as a product risk. The right response may be to disable the affected campaign or remove the vendor from that workflow until the gap is closed, not to ask recruiters to remember every exception. ^↗

A ten-record control test

Use ten synthetic contacts with distinct preferences and aliases. Trigger an opt-out through each major entry point. Verify that all outbound tools stop the relevant messages, transactional communication rules still work, and an opt-in reversal requires appropriate evidence. Document observed behavior, not only vendor configuration screens. ^↗

The worksheet below estimates propagation exposure for an operational queue. It is not a legal compliance determination; it helps prioritize system remediation and incident review. ^↗

Run the operational check