Your list has forty thousand subscribers and sixteen thousand of them want your email. Gmail sees the whole send, not the sixteen thousand.
List size is the vanity metric of email marketing, and it survives because it is the number that appears at the top of every dashboard. It correlates with almost nothing that matters, and past a certain point it correlates negatively with the thing that matters most.
Mailbox providers do not grade you on subscriber count. They grade you on how the people you send to behave. A send where a large share of recipients ignore the message tells Gmail that your mail is unwanted, and Gmail applies that judgment to everyone on the send including the people who would have bought.
Which produces the uncomfortable conclusion in the title. For a lot of brands, removing subscribers is the highest-return action available in the email program.
What Providers Actually Grade
The mental model worth adopting is engagement recency: what share of your sends go to addresses that opened or clicked recently.
Brands with high engagement recency get inboxed. Brands with low engagement recency get filtered toward promotions or spam. The mechanism is straightforward once stated, and it explains why a program can decline steadily without any single thing having gone wrong.
Send to a list where forty percent engage and sixty percent are dormant, and the filter sees a brand whose recipients do not want its email, even though sixteen thousand people genuinely do.
This sits downstream of authentication. Correct SPF, DKIM and DMARC records get your mail judged fairly rather than rejected outright; hygiene determines what that judgment concludes. Both are necessary and they solve different problems, and a brand with perfect authentication and a bloated list still ends up in promotions.
One diagnostic worth running before anything else: what percentage of your list opened or clicked in the last ninety days? Reporting suggests that if that figure sits below thirty percent, the list is bloated and reputation is paying for it. That single number tells you whether this article describes a problem you have.
Subscribe to Evolve Media on YouTube for new videos every week.
Why Unengaged Subscribers Cost You
Four separate mechanisms, and they compound rather than merely adding up.
- They drag engagement rates. Providers monitor open and click rates per campaign. A large dormant segment pulls the average down on every send, signalling that your content is not relevant.
- They raise complaint rates. Someone who mentally unsubscribed months ago but never clicked the link is materially more likely to hit the spam button than to unsubscribe politely.
- They hide spam traps. Abandoned addresses get recycled by providers into traps. A long-dormant address is exactly the profile of an address that has become one.
- They cost money directly. Platforms commonly bill on active profile count, so you are paying monthly to damage your own deliverability.
The trap point deserves elaboration because it is the one with the sharpest consequence. A recycled spam trap is an address that was a real person, was abandoned, and was reactivated by the provider specifically to catch senders who never clean their lists. Hitting one is a strong negative signal, and the only reliable protection is not sending to addresses that have been silent for a long time.
Notice that every mechanism points the same direction, and that continuing to send to a dormant subscriber has no upside path that offsets them. The subscriber is not going to buy; the question is only how much damage they do while remaining on the list.
Defining Engaged
Before you can sunset anyone you need a definition, and the definition should reflect how your customers actually buy.
The common starting point is a core engaged segment built from subscribers who opened, clicked, or placed an order within the last ninety days, used as the default audience for campaigns. Simple, and it works for most consumer brands.
Ninety days is a default rather than a rule. A consumable with a monthly replenishment cycle can use a tighter window. A considered purchase bought once every two years cannot, because a subscriber quiet for six months may be entirely normal for your category. Applying a ninety-day sunset to a furniture brand suppresses people who were always going to take longer.
A note on opens specifically. Open tracking has become less reliable as mail clients pre-fetch images, which inflates opens for some audiences and suppresses them for others. Clicks and orders are harder signals and should carry more weight in your definition than opens alone.
That argues for a definition using several signals rather than opens by themselves: opened, clicked, ordered, visited the site, or engaged on another channel. The more behavioral evidence you accept, the less likely you are to suppress someone who is genuinely interested but whose opens are not being recorded.
Three Thresholds, Not One
Search for the spam complaint threshold and you will find 0.01%, 0.1%, and 0.3% each presented as the answer. They are three different kinds of number and conflating them causes brands to manage to the wrong one.
| Figure | What It Actually Is | How To Use It |
|---|---|---|
| 0.10% | Google's stated guidance: keep spam rates below this | The number to manage to. Google's guidance indicates rates above it already hurt placement for bulk senders. |
| 0.30% | Google's stated ceiling: never reach this | Not a target. A line that means the program is already in trouble. |
| 0.01% | A sending platform's internal account-health bar | Deliberately stricter than any provider requires. Useful as an early warning. |
One source makes the point directly: merging an ESP's internal bar with a provider's enforcement line is how merchants end up managing the wrong number. A brand told its platform wants 0.01% and reading elsewhere that Gmail tolerates 0.3% will conclude it has thirty times more headroom than it does.
Google's sender guidelines are the primary reference and worth reading directly rather than through summaries. Klaviyo's deliverability reference is the equivalent for its own bars, and the two are answering different questions.
Reporting also indicates Google began ramping enforcement on non-compliant traffic from November 2025, including temporary and permanent rejections, and defines a bulk sender as one sending close to 5,000 messages or more to personal Gmail accounts in twenty-four hours.
The Metric That Improves When You Fail
This is the most important thing in this article and almost nobody writes about it.
Google's Postmaster Tools computes your spam rate against messages that actually reached an engaged recipient's inbox, and counts only DKIM-signed mail. Read that carefully, because the denominator is doing something unhelpful.
If your mail is being heavily auto-filtered to spam, fewer messages reach an inbox. Fewer messages in the denominator means the reported spam rate falls. A domain in genuine deliverability trouble can watch its spam complaint rate improve, and conclude things are getting better.
Never read spam rate alone. Read it alongside volume delivered to the inbox and alongside engagement. A falling spam rate accompanied by falling opens, falling clicks, and falling revenue is not an improvement; it is the signature of mail being filtered before anyone could complain about it. A falling spam rate with stable delivery and stable engagement is a genuine improvement.
The same reasoning applies to open rate during a deliverability decline. Opens can hold up among your most engaged recipients, whose mail still inboxes, while the broader list stops receiving anything. The average looks tolerable while the program hollows out underneath it.
Which is the underlying lesson for this whole topic. Email metrics are ratios, and ratios can improve because the denominator collapsed. Always look at absolute delivered volume and absolute revenue alongside any rate, because those cannot be flattered by failure.
Building The Sunset Flow
The structure is a genuine attempt at re-engagement followed by a decisive suppression, and both halves matter.
A segment of subscribers with no open, click or order in your defined window. Automated entry so new lapsers flow in continuously rather than in batches.
A short series asking directly whether they still want to hear from you, with a low-friction way to say yes. Two or three emails, not eight.
Anyone who opens, clicks or orders leaves the flow and returns to normal sending. This is the win-back and it justifies the whole exercise.
Set a profile property marking them suppressed and exclude that property from every campaign and flow audience. Decisive rather than gradual.
Stage four's implementation detail matters more than it sounds. Setting a property such as an engagement status, then excluding that property from all audiences, keeps the record intact while stopping the sending. That is different from unsubscribing them on their behalf and very different from deleting them.
On the re-engagement sequence itself, restraint outperforms desperation. A subscriber ignoring your mail for six months is not going to be recovered by a heavier discount and an urgent subject line; they are going to be recovered, if at all, by a plain question about whether they want to stay. Escalating aggression at exactly the moment someone is least interested is how you convert an ignore into a complaint.
Email is also not the only route to a dormant subscriber. If you have another channel where they are active, a check-in there before suppression costs nothing and occasionally works.
Suppress, Do Not Delete
A distinction that seems administrative and is genuinely consequential.
Suppression stops the sending and keeps the record. Deletion removes the profile entirely, taking with it the historical purchase data tied to that address: lifetime value, ordering patterns, and the predictive metrics that feed your future segmentation.
That data has uses well beyond email. It informs which products get repurchased, what a cohort is worth over time, and which customers to build lookalike thinking around. Deleting a past customer to save a monthly fee trades a permanent analytical asset for a small recurring saving.
Only consider it for profiles with zero purchase history, and only where platform billing genuinely justifies it. Anyone who has bought from you should stay in the database permanently regardless of whether you ever email them again. Their behavior is an input to decisions you have not made yet.
Suppression is also reversible in a way deletion is not. A suppressed subscriber can be reactivated if they later engage through another route, and a deleted one cannot be, because the history that made them interesting is gone.
Our guide to customer lifetime value covers why that historical data is worth protecting, and the short version is that most of its value is realized in analyzes you have not thought of yet.
The Revenue Math
Why the title is literal rather than a rhetorical flourish.
Suppression pays twice, which is unusual for any marketing decision.
The second return is the larger one and the harder to see, because it accrues to subscribers who are not the ones you suppressed. Better placement for your engaged segment shows up as improved campaign revenue that nothing in your reporting attributes to the suppression that caused it.
That attribution gap is why suppression gets postponed. It looks like removing subscribers who might theoretically buy, and the benefit arrives disguised as general improvement. Running it as a deliberate change with before-and-after measurement on your engaged segment is the way to see it.
The framing that makes this easy to defend internally is that list hygiene has become a profit and loss line rather than a marketing hygiene task. It reduces a cost and increases a revenue line simultaneously, and our contribution margin playbook covers where both belong.
The Ecom Profit Box
Our collection of ecommerce growth resources, including the retention frameworks behind this work.
Get It FreeBefore You Send Hardest
Suppression works before peak season, not during it. If you have not run one this year, there is still time.
Book A CallWhat The Benchmark Data Says
Two datasets circulate widely. Both are useful and both need reading with their sourcing attached.
Flows against campaigns. Klaviyo's benchmark dataset, reported as covering more than 183,000 brands, indicates automated flows account for roughly 41 percent of email revenue while making up about 5.3 percent of sends, with flow click rates around 5.58 percent against 1.69 percent for campaigns and placed-order rates of 2.11 percent against 0.16 percent. Large sample, and it is a platform's own data about its own accounts, so treat it as strong directional evidence rather than a universal law.
Inbox placement. Validity data reported global inbox placement at 83.5 percent for 2024, with 6.7 percent in spam and 9.8 percent missing entirely, dipping to 82.3 percent in the fourth quarter while retail sat higher at 87.7 percent. The source citing these flagged them as seed-list panel data with no published sample size from a company that sells deliverability monitoring, and advised using the direction rather than the decimals. That is the right instruction and it is repeated here.
The genuinely useful takeaway from the first dataset is not the specific multiples. It is that triggered flows massively outperform broadcast campaigns per send, which means a brand worried about revenue loss from suppression is usually worried about the wrong channel. Suppression reduces campaign reach. It barely touches flows, because flows are triggered by behavior and dormant subscribers were not behaving.
That is the reassurance most brands need before pressing the button. The revenue that matters most in email is already coming from people doing things, and those people are by definition not the ones you are about to suppress.
Bounce Handling And Traps
Adjacent hygiene that should be automatic and frequently is not.
- Hard bounces suppressed immediately. The address does not exist. There is no second attempt worth making, and repeated sending to invalid addresses is a strong negative signal.
- Soft bounces suppressed after several consecutive failures. Reporting suggests three to five. A single soft bounce is a full mailbox; five consecutive is an abandoned one.
- Watch bounce rate as a leading indicator. A rising bounce rate on a stable list means addresses are decaying faster than you are removing them.
- Treat long dormancy as trap risk. You cannot detect a recycled spam trap directly. Not sending to long-silent addresses is the protection.
Most platforms handle hard bounce suppression automatically. Soft bounce handling is more often configurable and left at a permissive default, which is worth checking rather than assuming.
One caution from the previous section applies here too. If your bounce rate climbs suddenly and the bounces cluster at a single receiving domain with an identical error message repeating, that is unlikely to be list decay and more likely to be an authentication or policy failure at that provider. Read the bounce strings before concluding the list is the problem.
Do This Before Q4, Not During
Timing advice that is specific rather than general, and it is relevant the week this publishes.
The rule is to suppress before the quarter you plan to send hardest, not during it, and to keep excluding those segments afterward. The reasoning is mechanical: reputation responds to recent sending behavior, so you want your improved engagement profile established before your highest-volume period rather than in the middle of it.
The failure pattern is the opposite and it is common. A brand notices declining placement during the holiday period, when volume and stakes are highest, and suppresses reactively. That helps eventually and it does not help this week, because the reputation damage was accumulating through exactly the sends that mattered most.
Reactivating a dormant segment for a holiday campaign, on the theory that lapsed subscribers might buy at Christmas. Sending to a large stale segment immediately before your most important campaigns is a reliable way to damage placement at the worst possible moment. If they were not worth emailing in October, the season does not change what they are.
The sequence for the remaining weeks is straightforward. Define your engaged segment, run the sunset flow now so re-engagement has time to work, suppress the non-responders, then send your peak campaigns to a cleaner list. That order gives your reputation time to reflect the improvement before volume rises.
If you have never run a sunset flow, doing it once now and then leaving it running continuously is far better than doing it perfectly in January.
When Not To Suppress
The honest limits, because aggressive suppression applied without thought destroys real value.
Long purchase cycles. If your product is bought every two or three years, ninety days of silence means nothing. Your window should approximate your repurchase interval rather than a default from an article about supplements.
Seasonal buyers. A customer who buys only in one season is dormant for most of the year by design. Segment them separately and time your contact rather than sunsetting them.
High-value customers. A subscriber who has spent significantly and gone quiet deserves a genuine attempt through other channels before suppression. The economics of one large customer differ from those of a list average.
Recent subscribers. Someone who joined six weeks ago and has not opened yet may have a signup problem rather than an interest problem. Check your welcome flow before concluding they are dormant.
When the real problem is authentication. If mail is being rejected rather than ignored, suppression addresses nothing. Confirm delivery is actually happening before concluding that recipients are uninterested.
The Version That Fits On A Card
Define engaged as opened, clicked, or purchased within a window matched to your purchase cycle. Send campaigns to that segment by default. Run a short re-engagement sequence for people who fall out of it, suppress the ones who ignore it, keep their records, and let the whole thing run automatically forever. That is the entire practice.
We do this work for clients as part of our email marketing services, so weigh that accordingly. It is also entirely buildable in an afternoon inside any competent platform, and a brand that builds it themselves gets the same deliverability benefit and the same reduced bill.
What To Remember
- Providers grade engagement recency, not list size. A send where most recipients ignore the message damages placement for the recipients who would have bought.
- Google's spam rate is computed against messages that reached an inbox, so heavy filtering makes the reported rate fall. A domain in real trouble can watch the metric improve.
- The 0.01%, 0.1% and 0.3% figures are three different things. Provider guidance, a provider ceiling, and an ESP's stricter internal bar. Managing to the wrong one creates false headroom.
- Suppress rather than delete. Deletion destroys purchase history, lifetime value, and predictive metrics that feed future segmentation, in exchange for a small recurring saving.
- Suppression pays twice, reducing platform cost billed on active profiles while improving placement for everyone who remains, which is why sending less genuinely earns more.
- Suppression barely touches flow revenue. Triggered flows are driven by behavior, and dormant subscribers were not behaving, so the revenue at risk is smaller than it feels.
- Suppress before your heaviest sending quarter, not during it, and never reactivate a stale segment for a holiday campaign.
Where This Came From
- Google, email sender guidelines, for spam rate guidance and the bulk sender definition. Google Postmaster Tools requires a sign-in and is cited by name rather than linked.
- Klaviyo, email deliverability best practices reference, for its own account-health bars and sunset flow guidance.
- Agency and platform reporting on sunset flow structure, engaged segment definitions of opens, clicks or orders within ninety days, sunset windows of 90 to 180 days calibrated to purchase cycle, and the implementation pattern of setting a profile property and excluding it from audiences rather than deleting profiles.
- Klaviyo benchmark data reported as covering more than 183,000 brands, indicating automated flows account for approximately 41 percent of email revenue from 5.3 percent of sends, with flow click rates of 5.58 percent against 1.69 percent for campaigns and placed-order rates of 2.11 percent against 0.16 percent. This is a platform's own dataset describing accounts on that platform.
- Validity inbox placement figures for 2024 of 83.5 percent globally, 6.7 percent in spam and 9.8 percent missing, with 82.3 percent in the fourth quarter and 87.7 percent for retail. The source citing these described them as seed-list panel data with no published sample size from a company selling deliverability monitoring, and recommended using the direction rather than the decimals.
- Agency reporting on the computation of Google's spam rate against messages reaching an engaged recipient's inbox and counting only DKIM-signed mail, with the consequence that heavy auto-filtering can cause the reported rate to fall; and on Google ramping enforcement on non-compliant traffic from November 2025 including temporary and permanent rejections.