The mistake to avoid: a 90-day open-based sunset rule applied to a brand whose customers buy every six months. You will suppress people who fully intended to purchase again, and you will never see the revenue you lost, because suppressed profiles do not show up in any report as a missed opportunity.
Sunset your list against your repeat purchase cycle, not a number you read in a blog post. And after Apple's Mail Privacy Protection, do not define engagement by opens at all.
Why sunsetting matters anyway
The case for it is real. People who have mentally checked out but never formally unsubscribed are the group most likely to hit the spam button, and complaints are the most damaging signal you can generate. Klaviyo's guidance is to keep spam complaint rates well below 0.1%, and since 2024 the stakes are enforced: Gmail and Yahoo begin rejecting bulk senders outright at a 0.3% complaint rate, at the SMTP level, before the email ever reaches a spam folder.
Every send to someone who will never open teaches mailbox providers that your mail is unwanted, and that judgment is applied to your whole list, including the customers who love you.
So the question is not whether to sunset. It is where the line sits and what counts as engaged.
Stop using opens to define dormancy
Apple Mail Privacy Protection pre-fetches tracking pixels the moment an email is delivered to an Apple Mail client, which registers a false open in Klaviyo whether or not the recipient ever saw the message. Open rates for lists with a meaningful Apple audience are inflated by roughly 15 to 35%.
For sunsetting, this cuts both ways and both are bad:
- False keeps. A truly dormant Apple Mail user shows phantom opens forever, so an open-based rule never sunsets them. Your dead weight stays.
- False cuts. A Gmail user who reads without images loading, or reads on preview, shows no opens and gets suppressed while still interested.
Build dormancy on clicks, site visits, and orders. Klaviyo can exclude MPP opens from segment logic, and any sunset segment built before 2022 that still keys on opens is due for a rebuild.
Set the threshold from your own repeat purchase window
Find your median time between first and second purchase. Then set your dormancy threshold at roughly twice that, with a floor of 120 days.
A brand selling a consumable on a 30-day cycle can reasonably treat 90 days of silence as dormant. A furniture brand, a jewelry brand, or anything with a considered purchase and a multi-year cycle cannot. Applying the consumable rule to the considered purchase is how brands quietly suppress their best future customers.
Worked examples:
- Daily supplement, 30-day supply: median reorder around 35 days. Dormancy line at 120 days (the floor). Anyone silent that long has missed three cycles.
- Fishing tackle: purchases cluster in season. Dormancy is one full season with no engagement, not 90 days that happen to span the winter.
- Leather goods: repeat purchases run years apart. Dormancy starts at 18 to 24 months, and the re-engagement surface is gift season, not elapsed days.
If you sell subscriptions, exclude active subscribers from the calculation entirely. A subscriber who never opens an email is still paying you every month, and treating email engagement as the measure will suppress your most valuable cohort.
The two-part structure
A working sunset path has a genuine re-engagement attempt followed by a decisive cull. Most brands do the first part and skip the second, which achieves nothing.
Part one: re-engagement, two or three emails
Ask a direct question rather than sending a discount. Discount-led re-engagement buys a purchase from someone who was going to buy anyway and tells you nothing about intent. A plain message asking whether they still want to hear from you produces a cleaner signal, and the people who click are genuinely worth keeping.
Structure that works: email one is a simple preference check. Email two, a week later, shows what they have missed and asks again. Email three states plainly that you will stop emailing, which is reliably the highest-click message in the sequence because honesty is rare in an inbox.
Part two: the cull, and it has to be automatic
If they do not engage, suppress them. Not a segment you mail less often. Suppressed.
This is the step brands avoid because watching list size drop feels like losing something. You are not. You are removing people who were already gone and were costing you placement with everyone else. Klaviyo's list cleaning guidance covers the mechanics.
Roll it out in stages rather than one pass: suppress the oldest cohort first, watch the metrics for two weeks, then proceed. Staging is what makes the decision reversible if your threshold was wrong.
What to measure afterward
Do not measure success by list size, which will obviously fall. Watch these instead:
- Spam complaint rate, which should drop within two to three weeks
- Click rate on campaigns, the engagement metric MPP cannot inflate
- Revenue per recipient, which usually rises because you are mailing a better audience
- Total campaign revenue, which should hold roughly flat or grow despite a smaller list
If total revenue falls meaningfully and stays down, your threshold was too aggressive. That is recoverable information, and staged rollout is what makes it cheap to learn.
Frequently asked questions
How long should a sunset flow wait before suppressing someone?
Roughly twice your median repeat purchase interval, with a floor of about 120 days. For most consumable brands that lands between 120 and 180 days. For considered purchases it is measured in years.
Should we use opens to define engagement?
No. Apple MPP inflates opens by 15 to 35% and makes them unreliable in both directions. Use clicks, site activity, and orders.
Should subscribers ever enter a sunset flow?
Not while their subscription is active. Their engagement is the recurring charge, not the email open.
Does a smaller list mean less revenue?
Usually not. Removing people who were not going to buy tends to raise revenue per recipient enough to offset the lost volume, because placement improves for everyone who remains.
Should we discount in a re-engagement email?
Prefer not to. A discount tests price sensitivity, not interest, and it trains your remaining list to wait for offers.
Sources
- Klaviyo Help Center. How to clean your email list to maintain good deliverability.
- Klaviyo Help Center. How to reduce spam complaint rates.
- Klaviyo Help Center. Identifying Apple Mail Privacy Protection opens.
- Mailgun. Google and Yahoo bulk sender requirements.