Email sunset policy: inactivity thresholds that work

Most email sunset policy threshold guides are wrong on the same point. They tell you to remove subscribers who have not opened in 90 days, or 180, or "six months", and call it done. That advice was written before Apple Mail Privacy Protection, before the 2024 Gmail and Yahoo bulk-sender requirements, and before anyone bothered to measure how engagement decays at different send cadences. We have. Across 2,400 newsletters in the Newsletrix corpus, the half-life of engaged behaviour is not one number. It is four, one per send cadence. This article publishes those four thresholds, the composite signal you should use instead of opens, and the 4-step workflow we run with operators who want a real deliverability lift inside ninety days.

What an email sunset policy threshold is

A sunset policy is the rule that decides when an inactive subscriber gets pulled out of your regular send list. It is not the same as a re-engagement campaign, and it is not the same as deleting subscribers. Our re-engagement playbook covers the campaign that tries to win them back; the sunset policy is what fires when re-engagement fails.

Suppression is the technical state your ESP puts the address in: still on the database for compliance and reporting, but no longer receiving sends. Deletion is rare and usually wrong, because you lose the audit trail you need under GDPR and the new bulk-sender rules.

The 180-day rule that most guides repeat ("sunset subscribers who have not opened in six months") is a folk benchmark, not a rule. It dates to a period when opens were a reasonably honest signal and most lists sent weekly. Both assumptions broke. Apple Mail Privacy Protection inflates opens on iOS, and send cadence varies more across modern newsletters than it ever has. A daily newsletter on a 180-day rule keeps a subscriber who has skipped 180 sends. A monthly newsletter on the same rule suppresses someone who skipped six.

Email sunset policy thresholds, by send cadence

Here is what we measured. Engagement-decay half-life, the point where half of an active cohort stops opening and clicking, clusters by send cadence rather than by industry. The bands below come from the Newsletrix corpus across January to April 2026, 2,412 newsletters tracked, B2B SaaS, creator, DTC, and ecommerce mixed.

Daily senders should sunset after 45 to 60 days of no composite engagement. That is 45 to 60 sends a subscriber chose to ignore. The half-life on these lists is around 28 days. By day 60, the recovery probability is below 4%.

Mid-frequency senders (two or three times per week) should sunset at 90 to 120 days. Half-life around 50 days. Recovery probability falls under 6% by day 120.

Weekly senders should sunset at 150 to 180 days. This is the only band where the conventional 6-month rule lines up with reality. Half-life is 84 days. By day 180, recovery probability is around 5%.

Monthly senders should sunset at 240 to 270 days. Half-life of 140 days. Subscribers on monthly cadences keep some recovery probability longer because they are not pattern-matched to your absence the way daily subscribers are.

The pattern is consistent across niches: B2B SaaS sits 10 to 15% longer on these windows than creator newsletters in the same cadence band. DTC sits roughly even with creator. Ecommerce sits a touch shorter because purchase intent decays faster than reading intent. Send cadence matters more than category. Cross-check with our send frequency benchmarks if you are not sure where your list sits.

Why opens alone are a broken sunset trigger after MPP

If your sunset criterion is "did not open in N days", Apple Mail Privacy Protection has broken your policy. MPP pre-fetches images, including your tracking pixel, on behalf of every Apple Mail user who has it enabled. The open event fires whether or not a human looked at the message. We covered the open-rate inflation in the MPP open-rate piece; on most lists we audit, between 18% and 34% of "opens" come from iOS pre-fetch.

What this means for sunset: a subscriber who has fully disengaged but uses Apple Mail will look engaged forever. Your policy keeps them on. Their inbox never loads your message past the pre-fetch. Your engaged rate dilutes. Your bulk-sender compliance numbers get worse.

Use a composite signal instead. Pixel load is one input. Link click is another. Reply, forward (detected via Resent-From headers on inbound mail), and any positive list action (preference center update, profile edit) round it out. Score each subscriber on a rolling window equal to your sunset threshold. A subscriber with only pixel loads in the entire window is a sunset candidate. A subscriber with a single click in the window is not.

One opinion that is unpopular among Mailchimp users: do not weight Apple-domain pixel loads at full value. Discount them by at least 50% in your composite. This costs you a little reach on iOS-heavy lists, but it stops the policy from being silently broken on the part of your audience most likely to be a heavy Apple user.

The 2024 Gmail and Yahoo bulk-sender angle

The 2024 Gmail and Yahoo bulk-sender requirements changed the cost of holding inactive subscribers. Both providers now enforce a spam complaint ceiling of 0.3% at the sender domain level. Cross it and your deliverability degrades across the entire domain, not just the offending campaign.

Holding inactive subscribers raises your spam complaint risk on two fronts. The first is mechanical: a subscriber who forgot they ever signed up is more likely to mark you as spam than to scroll for the unsubscribe link. The second is denominator effect. Spam complaints are measured as a fraction of delivered messages, so the more inactive recipients you send to, the more your engaged complaints get diluted into a worse-looking number.

Google Postmaster Tools and Yahoo's sender hub both surface an engaged-rate field that, while not officially specced, tracks the proportion of your recipients who positively interact (clicks, replies, moving you out of Promotions). The threshold that gets you flagged for review starts around 30%. Lists with healthy sunset hygiene sit comfortably at 45% to 60%. Lists without it sit at 20 to 30 and wonder why their Friday sends miss the primary inbox.

This is where a sunset policy stops being a list hygiene nice-to-have and becomes deliverability infrastructure. We watched one DTC client go from a 0.42% Gmail complaint rate to 0.18% inside six weeks by applying a cadence-adjusted sunset threshold to 38,000 dormant addresses. Open rate on the surviving list went up nine points. Click rate moved by two. The lift was paid for entirely by the subscribers they stopped sending to. The tradeoff is real and visible on the dashboard: list size dropped 11% the same week.

A 4-step email sunset policy workflow

We run this workflow on every list audit. It works on Mailchimp, Klaviyo, beehiiv, Kit, ActiveCampaign, and Iterable with minor adjustments to where each step lives in the UI. Klaviyo and Mailchimp handle native suppression best; see our Klaviyo comparison for the dashboard layout differences.

Identify candidates using the composite signal, not opens alone. Build a saved segment that captures subscribers with zero clicks, zero replies, zero forwards, and no preference center activity inside the cadence-adjusted window. Discount their pixel loads from Apple domains by 50% before deciding. This is where most policies fail silently, so we do not skip it.

Warn with a final win-back. Send one targeted message with a one-question reply or a single click-to-stay button. Keep the window short, 14 days. Anything longer dilutes the decision and inflates the cohort.

Suppress, do not delete. Move suppressed addresses into a permanent suppression list that your ESP excludes from sends. Keep them queryable for compliance, future cohort analysis, and the rare case where a subscriber reactivates via another channel. Beehiiv and Kit need a tag-based workaround; Klaviyo and Mailchimp do it natively.

Measure the deliverability lift at 30, 60, and 90 days. The visible numbers move on different timelines. Engaged-rate climbs first, usually within two sends. Spam complaint rate falls at 30 to 60 days as the suppressed cohort stops contributing. Inbox placement on Gmail moves last, typically inside the 60 to 90 day window. If the lift does not show by day 90, the criterion was too lenient, not too aggressive.

When not to sunset

Cadence-adjusted thresholds are a starting rule, not a hard one. Two cases need explicit carve-outs, and we get this question often enough that it earns its own section.

Seasonal buyers. Tax software, holiday gift retailers, education and back-to-school senders, and political fundraisers all have audiences that go quiet for predictable months and return on cue. Sunsetting an H&R Block subscriber in March because they did not click in November is a deliverability gain that costs you a paying customer in April. Tag seasonal cohorts at signup and exempt them from sunset between their dormant window and their active window. Klaviyo handles this with a custom property; beehiiv with a tag.

High-LTV B2B accounts. If a single account is worth $50,000 a year, the value of holding a dormant individual subscriber tied to that account is structurally different from a $30 ecommerce customer. We see this on lists where the buying committee has 4 to 6 named individuals. The CFO might never open the newsletter; the marketing director opens every send. Suppressing the CFO costs nothing technically but breaks the narrative when they ask their team "did you see that piece" and the answer is that you stopped sending to them. Pair sunset with CRM enrichment, suppress at the contact level only when the account itself is also inactive.

The carve-outs read like exceptions because they are. Most lists do not need them. If you cannot name the seasonal pattern or the multi-stakeholder account, you do not have one. Stick to the cadence-adjusted threshold and the composite signal, and check your churn benchmarks at 30 and 60 days to confirm the policy is doing what it should.

Frequently asked questions

What is an email sunset policy?

An email sunset policy is the rule that decides when an inactive subscriber gets removed from your active send list. It is not the same as a re-engagement campaign, which tries to recover the subscriber, or as deletion, which removes them entirely. The standard outcome is suppression: the subscriber stays in your database for compliance reasons but stops receiving regular sends.

How long before subscribers are considered inactive?

It depends on your send cadence, not the calendar. Daily senders should treat subscribers as inactive after 45 to 60 days with no engagement. Two or three times weekly: 90 to 120 days. Weekly: 150 to 180 days. Monthly: 240 to 270 days. Engagement-decay half-life clusters by cadence in the Newsletrix corpus across 2,400 newsletters. A flat 6-month rule keeps daily-sender dead weight and prematurely cuts monthly subscribers.

Should I delete inactive subscribers?

No. Suppress them. Suppression removes the subscriber from active sends but keeps the record in your database, which is what you need for GDPR audit trails and for the 2024 Gmail and Yahoo bulk-sender requirements. Deletion is appropriate only when a subscriber requests it under right-to-erasure rules. Klaviyo, Mailchimp, beehiiv, and Kit all support suppression natively.

Does sunsetting hurt list size metrics?

Yes, deliberately. A 50,000-subscriber list at 20% engaged-rate has the same engaged audience as a 12,000-subscriber list at 83%. The smaller list will outperform the larger one on Gmail and Yahoo deliverability, hit a lower complaint rate, and cost less to send. List size is a vanity metric; engaged rate is the operational one. Report both internally so nobody panics when the headcount drops.

How do I sunset subscribers after Apple MPP?

Stop using opens as your primary inactivity signal. Apple Mail Privacy Protection pre-fetches your tracking pixel for every iOS user who has MPP on, which means a fully disengaged Apple user looks active forever. Use a composite of clicks, replies, forwards, and preference center activity instead, and discount Apple-domain pixel loads by 50% in the composite score. The cadence-adjusted threshold then applies to the composite, not to the raw open count.

Related reading