Engagement gating: how to set it up and why it belongs at send time

An engagement gate is one rule: before a marketing message goes out, intersect whoever it targets with "has done something in the last N months". It is not a targeting refinement. It is the thing standing between you and a complaint rate that gets your domain filtered.

Nearly every tool supports it. Most shops apply it in the wrong place, which is when they build the audience rather than when they send to it.

Why the timing matters more than the rule

A segment built on Monday and sent on Friday is four days stale. In those four days people unsubscribed, bounced, complained, bought, or came back after eight months of silence. A gate evaluated at build time knows none of it.

1. Unsubscribes between build and send are the worst case. Mailing somebody who opted out on Tuesday is a breach, not an inconvenience.
2. Hard bounces from a previous send in the same week should not be attempted again.
3. Re-engagement works the other way: somebody who clicked yesterday should be back in, and a stale list keeps them out.

Evaluate at send time, against the current state. If your tool can only filter at build time, rebuild immediately before sending and treat that as the workaround it is.

Choosing the window

• Six to twelve months is the usual band. Twelve is generous; six is what to use if your complaint rate is anywhere near 0.1 %.
• Count clicks and orders, not opens. Opens have been unreliable since Apple began pre-fetching images, and an "engaged" segment built on opens is partly fictional.
• Include on-site behaviour if you have it — a browsing session is engagement even without an email interaction.
• Shorten it before a large or unusual send. A campaign to twice your normal volume should go to a tighter audience, not a looser one.

What the gate must not block

Transactional messages are a different legal and technical category and must never be gated. An order confirmation to somebody who has not clicked an email in two years still has to arrive.

• Order, shipping, refund and account messages: always exempt.
• Consent confirmations and password resets: always exempt.
• The re-engagement campaign itself: obviously exempt, or it can never reach the people it exists for.

This is one of several reasons transactional and marketing mail belong on separate subdomains — the separation makes the exemption structural rather than a checkbox somebody can forget.

The gate is not a sunset policy

Gating excludes somebody from a send. Sunsetting removes them from marketing altogether. You need both, and they are different.

1. 6 months: gated out of ordinary campaigns, and entered into a re-engagement sequence.
2. 9 months: one short "still want these?" with a single obvious button.
3. 12 months: marketing stops. Keep the record and the consent state — nobody is deleted.
4. Any click, order or form submission reinstates immediately.

What it costs and what it returns

A gate typically removes 30–60 % of a neglected list, and that feels like deleting revenue. It is not. Those addresses were producing almost nothing and were actively taxing the deliverability of the people who do buy.

The measurable outcome is not "more revenue from fewer emails" on the first send. It is a complaint rate that stops drifting upward and an inbox placement rate that recovers over the following weeks. Report it as revenue per recipient before and after, and expect the total to dip briefly and the rate to rise.