Frequency capping across channels

A customer gets an email at nine, a push at eleven, a retargeting ad through lunch, and an SMS in the afternoon, all from the same company, none of which knew about the others. Each channel was inside its own frequency cap. The customer, who experiences one brand rather than four channels, is being hit far past the point of fatigue, and the result shows up as rising unsubscribes, falling open rates, and the slow deliverability decay that follows when recipients start marking mail as spam rather than opening it.

The reason this happens is structural. Frequency caps are almost always set per channel, by the team that owns that channel, in the tool that runs it. The email cap lives in the ESP, the push cap in the mobile platform, the ad cap in the ad account. There is no unit in the stack that represents total contact to a person, so nothing can cap against it. Each channel is behaving responsibly on its own terms and the sum is exhausting.

This recipe moves the cap to where the customer actually experiences contact: the person, across every channel at once.

Contact capped per person across channels, with fatigue and its downstream costs falling as a result. Realistic effect: a reduction in unsubscribe and opt-out rates and a stabilizing of deliverability, with the size depending entirely on how over-contacted the audience was to begin with. A business running tight per-channel caps that rarely overlap sees little change; one where email, push, and SMS routinely stack on the same customer sees a real drop in fatigue signals. The honest framing is that this recipe removes a self-inflicted wound rather than adding a new lift, and its value is proportional to how deep the wound was.

The second outcome is decision clarity: a single contact budget per person that every channel respects, so adding a new campaign becomes a question of what it displaces rather than a free addition on top of everything else.

A unified contact store that the activation layer can query at send time, typically maintained in the CDP or warehouse with the channel tools writing send events to it. Composable stacks suit this because the contact log is a shared resource all channels read, which is awkward to bolt onto a packaged suite where each channel's frequency logic is internal and not designed to consult an external budget. The capability that decides whether the recipe works is the timing of the read: the cap has to be checked at send time, immediately before the message leaves, because a cap checked at campaign-build time is already stale by the time the send fires hours later.

Work in this order. The cap counts per person, across channels, on what was received.

1. Design the cross-channel contact log schema.
2. Resolve every send to the person. A cap counted per email address and per phone number gives the same customer a full budget on each, and the cross-channel cap never exists.
3. Categorize purpose, transactional or marketing, and audit it. A loose definition lets marketing sends get relabeled as transactional, and within a quarter everything important is exempt.
4. Set the per-person budget policy, with value tiers.
5. Read the budget at send time and allow or hold each message on what the person has already received.
6. Count received, not sent. Bounced and filtered messages consume no budget, which needs deliverability feedback flowing into the contact log.
7. Suppress once the budget is spent, for the window, with the same mechanism as any other exclusion.
8. Tune the cap from fatigue correlation.

Frequency and fatigue management across channels covers steps 1 to 4, 6 and 8, channel eligibility check step 5, and suppression and exclusion logic step 7.