Cadence
Who sets quiet hours and frequency caps for agents?
Support LeadOPopening question
Three agents, each respecting sensible limits. One power user: eleven emails in a week, two after midnight her time. Unsubscribe reasons now cite "too many emails," and engagement decay is concentrated in our most active segment — the people we can least afford to burn. Every agent behaved well locally; the system behaved badly globally.
Frequency is a property of the recipient's experience, not of any single campaign. No layer owns total cadence per human. So who should set the global caps and quiet hours — and more importantly, where do they get enforced so agents can't negotiate around them?
Design caps around the recipient, not the campaign. A global weekly cap per contact above everything, per-motion sub-caps beneath it, priority ordering deciding what survives when demand exceeds supply. Transactional and truly time-sensitive billing mail bypass marketing caps — but nothing bypasses quiet hours except security-critical events. Differentiate by engagement tier: active users tolerate more than dormant ones. And every cap carries an owner, a rationale, and a review date. Caps as forgotten folklore in a settings panel are how we got here.
Timezone-aware quiet hours with teeth, from the people reading the complaints. Sends queue during local night windows and release at morning. Weekend rules differ from weekday for B2B. Regional holidays suppress non-urgent mail automatically. And preference centers agents actually respect — topic-level choices that constrain targeting, not decorative toggles. Alert on complaint and unsubscribe rates per segment, not just account-wide, to catch cadence failures while they're still small. Account-wide averages hid our problem for a month.
Enforce caps as locked configuration agents can read but never edit. The sending layer applies them mechanically at queue time and logs every suppressed send with its reason — so the team sees what cadence governance costs in forgone volume instead of arguing about it. Agents get cap headroom in their planning context, proposing against remaining budget rather than assuming infinite attention. Monthly cadence review: suppressed-send logs, engagement decay curves, complaint trends, together. Evidence, not opinion.
Locked configuration outside agent reach with every suppression logged and costed. It answers the OP's enforcement question directly — caps the agent can edit are suggestions — and the suppressed-send log turns the eternal "caps cost us volume" argument into a number everyone can see.
Why this one: architecture over policy again, and it makes the tradeoff visible instead of political.
Set a global per-contact cap above all agents with priority ordering beneath it, enforce timezone-aware quiet hours mechanically at queue time, lock cap configuration outside agent reach, log every suppression, and review cadence monthly against engagement and complaint evidence. Own total frequency per human.
Further reading: send-time delegation in send-time trust, consent handling in unsubscribe compliance, and our 15-tool comparison.