What it evaluates
| Check | Question answered |
|---|---|
| Consent | Is there valid consent or another recorded basis for this person, channel, purpose and jurisdiction? |
| Suppression | Is this contact on a suppression list, globally or for this tenant or channel? |
| Contact window | Is it inside permitted hours for the recipient’s time zone and local rules? |
| Frequency and volume caps | Has this person or number been contacted too often? Is the campaign within its limits? |
| Kill switches | Is the campaign, agent, provider, tenant or all outbound activity switched off? |
| Channel and provider limits | Are provider rate limits and channel permissions satisfied? |
Why execution time matters
Eligibility changes between scheduling and sending. The engine therefore runs immediately before each action, regardless of what was approved earlier. The decision, with its inputs, is stored with the attempt, giving you evidence of why something was sent or blocked.
Fail closed
If the policy store cannot be reached, outbound calls and messages are blocked rather than allowed. Inbound handling continues on last-known policy with an alert. Emergency stops propagate to enforcement points quickly, and releasing a switch requires a post-check.
Consent records
Consent is recorded per channel and purpose with evidence, timestamp, source and expiry. Withdrawal creates a suppression record immediately. History is append-only so you can show what was known at any point in time.
Policy packs are configuration, not legal advice. Country and channel rules differ. Each pack has an owner and review date, and regulated use cases should be reviewed by your counsel before launch.