Sunset Flow or Hard Suppression in Klaviyo

Run a sunset flow only for profiles that have clicked or purchased at some point in their life, and suppress everyone else without sending them anything. The sunset sequence is itself mail to your least engaged cohort, so it buys back a small amount of permission at the cost of more complaint exposure at the exact moment your reputation is thin. If your Gmail spam rate has touched 0.30% on any day in the last two weeks, skip the sunset send entirely, suppress now, and re-earn that audience through acquisition instead of a last-chance email.

The decision is a revenue question, not a hygiene question. Below is how to put a dollar figure on both sides before you choose.

What a sunset flow actually costs you

It costs you complaint exposure, concentrated in the worst possible cohort, delivered over the two or three weeks you most need a clean signal. Mailbox providers score you on complaints against delivered mail, and Google's sender guidelines are explicit that you should keep your reported spam rate below 0.10% and never let it reach 0.30% (Google Email sender guidelines). Yahoo publishes the same 0.30% ceiling in its sender best practices (Yahoo Sender Best Practices). A sunset flow deliberately sends more volume to the people most likely to hit the spam button, which is why a poorly timed one can take an account from Medium to Low domain reputation in a fortnight.

There is a second cost operators underestimate. While the sunset runs, your aggregate metrics get worse, so every campaign sent alongside it looks like it underperformed. You lose the ability to read your own tests for the duration.

Hard suppression has the opposite profile. It costs you the entire residual revenue of that cohort on day one and returns a cleaner signal immediately. In Klaviyo, suppression is distinct from unsubscribing: the profile stops receiving marketing email but stays in the account for reporting and exclusion audiences, which matters for what you do next (see the suppression documentation in the Klaviyo help center).

How much revenue is actually in the unengaged cohort

Build the cohort as a segment first, then pull trailing 90 day email-attributed revenue for members of that segment only, and divide by cohort size. Do not estimate this from a template or from what an agency deck told you. The number varies by an order of magnitude between a consumables brand with a 42 day repurchase interval and a furniture brand with an 18 month one.

Two corrections before you trust the figure. First, Klaviyo's default attribution window credits a purchase to an email opened or clicked days earlier, so some of that revenue would have happened anyway through paid or direct. If you have not reconciled the two systems, the cohort will look more valuable than it is (reconciling Klaviyo revenue with Shopify). Second, strip out flow revenue. Browse abandonment and back in stock will keep firing for these profiles based on site behaviour, and those triggers are evidence of engagement, which means those profiles should not be in the sunset cohort at all.

At ten cents per profile per year, the cohort is funding roughly $12,000 annually against 1.4 million sends that pull your inbox placement down for the other 291,000 profiles. If placement recovery lifts revenue per thousand delivered on the engaged list by even 4%, suppression wins outright. That is the calculation, and in most accounts it is not close.

When hard suppression is the only defensible choice

Suppress without a sunset send when any of these are true, because in these cases the sunset flow adds damage faster than it recovers permission.

• Gmail spam rate has hit 0.30% on any day in the last 14 days in Postmaster Tools, or sits consistently above 0.10%.
• Domain reputation reads Low or Bad, so most of your mail is already landing in spam and a sunset email will mostly go unseen while still generating complaints from what does land (why Postmaster reputation refuses to reconcile).
• The cohort came from a bulk import, a giveaway, a coreg partner or an acquired list, and has never clicked anything. There is no permission to re-establish.
• You are inside 45 days of a peak sending period. Do the suppression now and the sunset never, because there is no safe window to run it (clean, suppress or restart an inherited list).
• Unknown-user bounce rate on the cohort exceeds roughly 2% on a seed send, which indicates the addresses have decayed past the point where re-permission is the problem.

If none of those hold and the cohort has genuine historical engagement, the sunset flow is worth running. That is a narrower set of accounts than the popular advice implies.

How to build the cohort without catching your best customers

Define engagement on clicks, site events and orders, and set the window from your own repurchase interval rather than a round number. A 90 day window on a brand whose median second purchase lands at day 210 will suppress customers who are behaving exactly as expected (sizing the engagement window by repurchase).

Working definition of the sunset cohort:

Profile is in the sunset cohort when ALL of:
  - Clicked email 0 times in the last [2x median repurchase interval, floor 120 days]
  - Placed Order 0 times in the last [3x median repurchase interval, floor 365 days]
  - Active on Site 0 times in the last 180 days
  - Profile created more than 90 days ago
  - Not in any active flow that has fired in the last 30 days
  - Consent status is subscribed (not already suppressed)

Split into:
  A. Ever clicked OR ever purchased  ->  two-email sunset
  B. Never clicked AND never purchased  ->  suppress immediately

Two exclusions people forget. Profiles engaged on SMS or push but silent on email are engaged customers; suppressing their email address is a channel decision, not a hygiene one. And profiles whose only email engagement signal is opens should be treated as never-engaged, because Apple Mail Privacy Protection prefetches images and registers an open the recipient never made. If your account still runs 30 day open-based engagement segments, the split above will look alarming the first time you build it. That is the correction, not an error.

How to run the sunset send so it does not make things worse

Two emails, seven days apart, sent in daily batches on days when no campaign goes out, with a hard kill switch. Five-email sunset sequences exist in a lot of templates and they are simply five chances to be reported as spam by someone who forgot you.

1. Cap the daily batch. Keep the sunset volume under roughly 20% of your normal daily send to the engaged list, so a spike in complaints stays a minority of the day's signal.
2. Isolate the day. Do not send a campaign to the engaged list on a day the sunset batch goes out. If both send, you cannot attribute a reputation movement to either.
3. Make leaving trivial. Confirm the header includes one-click unsubscribe as specified in RFC 8058, which Google and Yahoo both require of bulk senders. A recipient who unsubscribes costs you nothing. A recipient who reports spam costs you placement for the whole domain.
4. Give one action. A single link that confirms they want to keep hearing from you, with a plain layout and minimal imagery. Discount-led sunset emails recover the wrong people: they re-engage the deal-only cohort and teach the rest of your list nothing useful.
5. Watch it daily, per provider. Check Google Postmaster Tools and the Microsoft Smart Network Data Services feed each morning. Gmail and Outlook will not move together, and Outlook problems often show up first as a filtering change rather than a complaint number.
6. Kill switch. If the daily Gmail spam rate exceeds 0.20%, stop the flow and suppress the remaining cohort unsent. Decide this threshold before you start, because you will not want to stop mid-run.

If you are unsure which part of the list is already causing damage, isolate it before the sunset rather than during (finding the segment burning your domain).

What to do with profiles that never respond

Suppress them in Klaviyo and keep the profile record. Deleting profiles feels tidy and creates two problems: the historical order and revenue data attached to those profiles disappears from your cohort analysis, and nothing stops the same addresses re-entering through the next integration sync or CSV import, at which point they are subscribed again and you repeat the damage. Suppression persists across imports; deletion does not.

Suppressed profiles remain useful. They work as exclusion audiences in paid social so you stop paying to retarget people you already own, and as a seed for lookalike audiences. Transactional mail, order confirmations and shipping notifications continue to send, because those are not marketing messages. On most Klaviyo plans suppressed profiles also stop counting toward billable active profiles, so a 121,000 profile suppression usually moves you down a pricing tier. Confirm that against your own contract before you promise it to a CFO.

Keep the cohort documented. Industry sender guidance has been consistent for a decade that permission decays and lists need periodic pruning (M3AAWG Sender Best Common Practices), so this is a recurring quarterly job, not a one-time cleanup. Write the rule down so the next person does not rediscover it from scratch.

How to tell whether it worked

Judge it on revenue per thousand delivered and on Postmaster domain reputation over six weeks, and ignore the open rate. Open rate will jump the moment you remove dead addresses, which tells you nothing about whether you make more money. The metrics that answer the question:

• Total email revenue, weekly. The honest test. If placement was the constraint, total revenue holds flat or rises within four to six weeks despite sending 25% to 40% fewer emails.
• Revenue per thousand delivered. Should rise immediately and keep rising as placement recovers.
• Postmaster domain reputation and spam rate. Reputation moves slowly and lags the volume change by one to two weeks. Do not declare failure on day four.
• Gmail click rate specifically. Split performance by provider. Gmail recovery is the clearest evidence that placement, rather than creative, was the problem.
• Complaint rate as your platform reports it. Klaviyo and Google calculate this on different denominators, so a gap between them is expected and is itself a placement signal (diagnosing the Klaviyo vs Gmail complaint gap).

Trade-offs and what I would do

The case for the sunset flow is that some fraction of the cohort is a dormant customer with a long purchase cycle, and a well-built sunset recovers 2% to 5% of them at a cost of three weeks of noisier metrics. That is a defensible trade when your reputation has headroom and the cohort has genuine history. The case against is that the recovered profiles are usually the ones you would have reached anyway through browse and back in stock triggers, and the complaint cost is paid by your entire sending domain rather than by the cohort that generated it.

The framing of "which loses less revenue" hides the asymmetry. Suppression loses a known, small, bounded amount. A sunset flow run at the wrong time loses an unknown amount across your whole file, and reputation recovery after a Low rating takes four to eight weeks of disciplined sending. Bounded losses beat unbounded ones when the expected values are this close.

My rule: suppress cohort B (never clicked, never purchased) unsent, always, no exceptions. Run a two-email sunset on cohort A only when Gmail spam rate has been under 0.10% for 30 days and you are more than 45 days from a peak. Everything else suppresses.

One exception worth naming. If you have just inherited an account and the cohort is large enough that suppression would take you below roughly 20,000 reachable profiles, a sunset flow on cohort A is worth the risk, because rebuilding acquisition from that base takes longer than reputation recovery does. Below that threshold your constraint is list size, not placement. Above it, placement is what you are actually managing, and the sunset flow is a liability dressed as a best practice.

Sources

• Google, Email sender guidelines
• Yahoo, Sender Best Practices
• RFC 8058, Signaling One-Click Functionality for List Email Headers
• Microsoft, Smart Network Data Services
• M3AAWG, Sender Best Common Practices
• Klaviyo Help Center, suppression and consent documentation