Open your email platform and look at the number next to your list. Twelve thousand, say. That number has been going up for two years, which felt like progress every time you checked it, and it is the number you quote when someone asks how the email channel is doing.
Now build a segment of everyone who has opened nothing and clicked nothing in the last ninety days. Do not act on it. Just look at the count.
For most stores that have been running a signup popup for a couple of years, somewhere between a third and half of the list lands in that segment. The list is still twelve thousand. The part of it that is listening is six. And you are paying for twelve.
The Bill Is the Cheapest Thing a Dead Subscriber Costs You
There is a version of this problem that is only about money, and it is the easy version. Almost every email platform prices on how many profiles you store. Carry six thousand people who will never open anything again and you are paying a subscription to warehouse them.
That is real, and for a store on a mid-tier plan it is not a trivial line item. But it is the least interesting of the three costs, and if it were the only one, this would be a short article about downgrading a plan.
The second cost is the one that compounds. Inbox providers decide where your mail lands, and one of the strongest inputs they use is whether the people you send to actually engage with it. Send to twelve thousand when six thousand are inert and your open rate is halved by arithmetic alone, before anyone judges your subject line. Google and Yahoo read that halved number as evidence you are not worth prioritizing. So they put more of your mail in Promotions, or in spam. Fewer people see it. Your engagement rate falls again. That lower number becomes the new evidence.
It is a loop, and the direction of travel is one way. The subscribers who cannot hurt you because they never open anything are the precise reason the people who would open it are not seeing it.
The third cost is quieter and it is the one nobody bills you for. Every decision you make about email runs through the list number. What a subscriber is worth. Whether the popup is working. Whether it is worth paying for that giveaway. Whether email revenue per subscriber is going up or down. All of those calculations take the twelve thousand and divide something by it, and all of them are wrong by the size of the dead half.
Three Reasons This Feels Like Setting Money on Fire
The resistance to doing anything about this is not stupidity. Those subscribers cost real money to acquire. You paid for the traffic, you built the popup, you gave away the first-order discount. Deleting them feels like writing off the acquisition spend twice.
Three objections come up every time, and each one deserves a straight answer rather than a lecture.
"I am cutting off sales opportunities." The people in this segment have received ten or fifteen emails from you and opened none of them. They are not opportunities being neglected. They are non-opportunities being carried. And because carrying them pushes your mail out of the primary inbox, the cost of keeping them is paid by the customers who do want to hear from you and are now finding you in Promotions.
"They might buy from me later." They might, and nothing about this stops them. Suppressing someone from your email list does not remove them from your business. They can come back through a search, an ad, a friend, or the same social post that got them the first time. What you are concluding is narrower than it feels: email is not the channel this person responds to. That is a fact about the channel, not a verdict on the customer.
"Some of them will re-engage." A few will. The practitioners who do this at volume all say the same thing, that the number is small, and the question is whether a small recovery rate justifies degrading inbox placement for everyone else. It does not. But this objection is also the reason you do not simply delete the segment on a Tuesday afternoon, and we will come back to it, because the correct answer is to give them a real chance to raise their hand first.
Nobody Who Teaches This Agrees on the Number
Here is the part that should change how you read every article on this subject, including the ones with confident screenshots.
We went through six practitioners who do email for ecommerce brands professionally and teach this exact process. Every one of them gives a specific threshold for what counts as unengaged. No two of them match.
On the window: three say ninety days. One offers a hundred and twenty. One says six months. One says three hundred and sixty five. That is a four-fold spread, and each one is presented as the standard.
On how many emails someone must have received before you judge them: one says five to fifteen depending, two say ten, one says thirteen, one says fifteen.
On the goodbye sequence itself: one teaches a careful three-email flow with different copy for customers and non-customers. Another says never send more than one, on the grounds that you are writing to people who do not open your email, so writing three times is a way of feeling thorough rather than being effective.
They disagree about whether to strip out Apple's automatic opens, which inflate open data on Apple Mail. One adds a filter for it. One deliberately does not, and says plainly that the data is too murky to act on either way. They disagree about whether to put a discount in the re-engagement email. One calls it counterintuitive and mostly wrong. One finds it softens the blow and cuts unsubscribes.
None of these people are careless. They run this for brands considerably larger than yours and they get results. The disagreement is not a sign that one of them is wrong.
It is a sign that the number is not a property of email. It is a property of your sending cadence and your purchase cycle. Ninety days of silence from someone who gets four emails a week is a completely different signal from ninety days of silence from someone who gets one a month. A store selling coffee should read a three-month gap as alarming. A store selling dining tables should not read it as anything at all.
Which means copying anyone's number, including a number from this article, produces a segment that describes their business rather than yours.
Find Your Own Threshold in About Ten Minutes
There is exactly one technique in all of that material that computes the number instead of asserting it, and it comes from Derek Stroh, who has worked on retention for a long list of ecommerce brands. It is simple enough to do this afternoon.
Build the unengaged segment with your best guess at the threshold. Then clone it, raise the "has received at least N emails" condition by one, and write down the segment size. Clone it again, raise it again, write it down. Keep going.
What you are looking for is not a number you like. You are looking for the cliff.
In his worked example the counts run thirteen hundred, then eight hundred and seventeen, then eight hundred and three, then seven hundred and thirty five, and then two hundred and five. Four gentle steps and then the floor drops out. That last drop is the boundary between people who have had a genuine, repeated chance to engage with you and people who have not. Thirteen is the answer, and it is the answer for that store only.
Run it on your own account and you will get a different number, and that number will be defensible in a way that ninety days copied off a video never is. It takes about ten minutes and it is the difference between cleaning your list and guessing at it.
Give Them a Door Before You Close It
Nobody should go from your active list to suppressed without being asked. This runs in two stages, and the staging is what makes it defensible rather than brutal.
The first stage is a re-engagement check-in, sent while someone is drifting rather than gone. The mechanics that matter here are consistent across every source, which is worth noting given how much else they disagree about. Send plain text, not your designed template, because a plain message has a better chance of reaching the primary inbox and it does not look like the marketing they have been ignoring. Send it from a person, with a name, not from the brand. Ask a real question and invite a reply.
The reply is the point. A reply is the strongest positive signal a recipient can send about you, stronger than an open and stronger than a click, and it improves where your mail lands for that person afterward.
There is one trap in this, and it is easy to walk into precisely because the advice above is good. If you write a warm personal note asking someone to reply, and you include no link at all, your click rate on that email is zero by construction. The segment is defined on clicks. Nobody leaves it. You will conclude the campaign failed when what actually happened is that you built a message your own measurement could not see. Include a link, even a plain one to your homepage.
The second stage is the sunset flow, for people who ignored the check-in. It says, kindly and without drama, that you will stop sending unless they say otherwise, and gives them one obvious way to stay. Then it does what it said it would do.
Two details save you grief here. Build the segment on business activity, not just email activity, so that no order and no site visit are part of the definition alongside no opens and no clicks. Someone who ignores your email but browsed the store last week is not disengaged, and they will be furious to receive a goodbye note. And turn off any setting that suppresses sends to people who recently got a campaign, for this flow only. Otherwise the people you most need to process get skipped, never get marked, and sit in the unengaged segment permanently while you wonder why it never shrinks.
Suppress. Do Not Delete.
These are not the same operation and the difference matters more than it sounds.
Deleting a profile removes the person and their history. The orders, the value, the acquisition source, gone from your reporting. Suppressing keeps every bit of that and simply stops the sending. You retain the ability to analyze what happened, the customer stays in your history where they belong, and if they come back on their own the relationship can resume.
On the money question, Klaviyo's own documentation is unambiguous: "suppressed profiles do not contribute towards your billing plan's profile count." So the saving is real and it is the platform's own stated position, not a practitioner's estimate. If you are on a different platform, check its equivalent page before you count on it, because this is the one claim that decides whether the exercise pays for itself directly.
The Bill Has Two Meters. Watch the One You Are Not Watching.
This is the detail most likely to cost you money in the week after you do all of the above, and it comes from Kieran Dwyer, who builds these systems for ecommerce brands.
Email platforms typically meter two things: how many profiles you store, and how many emails you send per month. Cleaning your list moves the first meter. It does not automatically move your invoice.
Worth knowing exactly how the asymmetry runs, because it is documented in one direction only. Klaviyo states that if your profile count outgrows your plan, "you may be automatically upgraded the next billing cycle." It publishes no equivalent promise about moving you back down when the count falls. So after you suppress, go into billing and look at your tier yourself rather than assuming the saving arrived. Plenty of stores do the work and never check, which is how the exercise ends up feeling pointless.
The send meter deserves more care than the profile meter, and this is the part worth getting right. Exceeding your plan's sending limits does not politely hold your mail back. Klaviyo's documentation on throttling describes emails that "cannot be sent in a given time period due to volume restrictions" when auto-upgrade and flexible sending are switched off and a send runs past the plan's ceiling. What gets caught in that is not only your newsletter. It catches flows: the abandoned checkout sequence, the post-purchase series, the back-in-stock alert. Those are the highest-converting messages in the account, because they fire on intent rather than on a schedule, and they are the last things you would choose to switch off. We have written about the five automations a store should be running, and every one of them sits behind that same meter.
So when you set the plan, work the sends out deliberately. Campaigns per month multiplied by the size of the segment you actually send to, plus headroom for the flows. Leave real slack. The saving from suppressing six thousand dead profiles is not worth capping the automations that make the money.
The same logic applies with more force to text messages, where every send has a hard cost rather than a metered one. Carrying an unengaged SMS list is considerably more expensive than carrying an unengaged email list, and the cleaning discipline should be tighter there, not looser. It is the same category of accumulating cost as the software bill that grows every time you sell: nothing about it is dramatic in any single month, which is exactly why nobody looks at it.
What the Real Number Is For
Do this and you will end up with a smaller list and a worse-looking dashboard, and both of those things are improvements.
The engaged number is the one worth building on. It is the denominator for revenue per subscriber, the input for what a new signup is genuinely worth, and the honest answer when someone asks how big your list is. Forecasting off a number you know is half fiction is how stores talk themselves into paying for subscribers who were never going to buy anything.
There is a useful habit in this beyond the cleanup itself. Once a quarter, look at what share of your list has engaged in the last ninety days, and watch that percentage rather than the total. A list growing in size while shrinking in engaged share is a store with an acquisition problem wearing a growth costume. That ratio will tell you something about the health of your email channel that the subscriber count structurally cannot.
None of this is difficult. It is an afternoon, it is reversible, and the main thing standing in the way is that a smaller number feels like a step backward when it is the first accurate one you have had.
If you would rather have someone work out your real threshold, build the two stages properly, and set the plan so the flows never get capped, that is the kind of unglamorous work we do for ecommerce stores every week. Or do it yourself this afternoon. Start by building the segment and just looking at the count, which costs nothing and tells you immediately whether the rest of this applies to you.
