Customers Do Not Unsubscribe, They Press Spam

Retention teams have a number for this. Unsubscribe rate sits on the campaign report next to opens and clicks, and when it stays under a fraction of a percent the cadence gets called safe. Klaviyo, Attentive and every ESP dashboard reinforce the same reading: if people are not unsubscribing, they are still with you.

This is a voice-of-customer study of more than 3,200 real comments from people describing marketing emails and texts they no longer want, and what they did about them. I went in expecting an argument about frequency thresholds, some number of sends per week above which people leave. Frequency does turn out to be the trigger. But the more useful finding is about the exit itself. When people decide they are done, most of them never touch the link you are measuring.

About this study

Three notes on scope before the findings. First, this corpus captures people who chose to talk about the messages, which skews toward irritation. Quietly tolerant subscribers are underrepresented here by design, so read the rankings as a map of where the damage happens rather than as a measure of how many customers are unhappy. Second, some of the communities in this corpus are marketing practitioner spaces, and I kept them because practitioners describe their own behaviour as recipients as often as they describe their sending. Where a comment is clearly a marketer talking about strategy rather than a person talking about their inbox, it is flagged in the text. Third, the SMS conversation is much smaller than the email conversation here, roughly 49 coded mentions, so section 5 is reported as a direction and not as a measurement.

The research question

When do frequency, repetition and irrelevant promotions push an otherwise interested customer to opt out, and what cadence still feels useful? The short version: frequency is the trigger by a wide margin, but the more consequential finding is the exit route. The unsubscribe link has lost enough credibility that most people bypass it, and the substitute they choose most often is the spam button, which converts one person's fatigue into a deliverability cost paid by your whole list.

1. What pushes an interested customer out

Start with the trigger, because it settles a long-running argument in lifecycle marketing. The standard line is that frequency is a proxy problem and relevance is the real one, that you can send daily if the content earns it. That may be true as strategy. It is not how customers narrate the decision. They almost never say a message was irrelevant. They say a number.

The frequency complaints are specific in a way that is useful, because people name the brand, the cadence and the duration together. That combination is what makes the memory stick.

Second place is worth separating from first, because it is a different failure. Repetition is not high frequency, it is zero new information at any frequency. In several cases the sender did not even intend it, which does not help: from the recipient's side a broken loop and an aggressive cadence look identical.

Third place is small in volume and large in consequence. About one coded trigger in nine is a message that arrived on the wrong day, and these are the comments where the tone changes completely. Nobody in this group is complaining about cadence. They are describing an automated calendar campaign walking into a private situation.

2. What people actually press

This is the finding that reframes the metric. When people describe doing something about messaging they no longer want, the unsubscribe link is not the default. It is third, behind two actions that are invisible on your campaign report and expensive in different ways.

The spam-button comments are not confused about what they are doing. Several are explicit that they have chosen it over unsubscribing on purpose, and that the point is collective damage to the sender rather than personal relief.

The fourth and fifth options are the quietest and the hardest to detect, because they leave your metrics completely intact. A filtered subscriber still counts as a subscriber. An aliased address still accepts mail. The list looks healthy while the audience behind it has already gone.

3. Why the unsubscribe link stopped being trusted

The obvious question about section 2 is why. One-click unsubscribe is a legal requirement, it is one tap in Gmail, and it is faster than building a filter. People avoid it anyway, and the corpus is unusually clear about the reasoning.

The belief in third place is the load-bearing one, and it gets repeated as established fact, with a mechanism attached. It comes from real advice about genuine spam and phishing. The problem for legitimate senders is that customers do not apply it selectively.

Read those three together and the shape of the problem is clear. The customer is not making a judgment about your brand. They are applying a general rule about marketing mail, formed from experience with the worst senders, to every sender at once. Your compliant footer link inherits the reputation of the industry it sits in.

4. When the unsubscribe itself is the injury

The belief in section 3 would fade if the mechanism worked reliably. It does not, and the corpus contains a long record of why. This is the section where the brand stops being merely annoying and starts being untrustworthy, because a failed opt-out is a promise broken in a place the customer can verify.

The first category is overwhelming, and the detail that recurs is the count. People do not say the unsubscribe failed. They say how many times they tried, which is how a service failure becomes a grudge.

The second category is where design choices get read as intent. Worth noting that some of these are not deliberate at all. The clipped-message complaint below is a message-size problem, not a dark pattern, but it produces exactly the same conclusion in the reader's head.

5. Email and SMS are not the same permission

A caveat first, because this is the thinnest section in the study. The SMS conversation in this corpus is roughly 49 coded mentions against several hundred for email, so what follows is a direction rather than a measurement. That imbalance is itself informative: people write long complaint threads about their inbox and short, final ones about their phone.

The acceptance condition is stated almost identically across the corpus, and it is not about consent paperwork. It is about whether the recipient can work out why this arrived before they have finished reading it.

The penalty side has no equivalent in the email comments. Nobody in this corpus says a marketing email put a brand on a permanent do-not-buy list. Several say it about a text.

6. What still earns a place in the inbox

The corpus is dominated by complaint, which is what a corpus gathered this way will always be. But a real minority of comments describe senders people keep, and they are consistent enough to be worth ranking. The answer is less about content quality than the industry assumes.

The gap between first and last place is the uncomfortable part of this study. Affection for the brand barely registers as a reason to keep receiving messages from it. What people describe instead is a rhythm they can predict, and specifically one they had some say in.

That last comment deserves its own reading, because it is the whole study compressed into one customer. This person wants the operational messages, does not want the promotional ones, cannot separate them, and so turns everything off. The brand loses the delivery notification it was never being punished for, because it bundled it with the marketing it was.

There is a dissenting position too, and it comes from the practitioner side of the corpus. It is worth including because it is the strongest available argument against a frequency cap, and because the disagreement is real rather than manufactured.

The question this corpus could not answer

Whether any of this shows up in revenue. The comments are rich on what people feel and what they press, and silent on what they spent afterwards. A person who marks a brand as spam may well keep buying from that brand, and several in this corpus say as much about retailers they use constantly. Channel churn and customer churn are different things, and nothing here measures the distance between them. What the corpus does establish is a cost that is not about revenue at all: the most common exit route damages your ability to reach the customers who never complained.

A customer glossary

If you work on lifecycle, retention or CRM for a consumer brand, these are the terms already in circulation among the people you are sending to.

Frequently asked questions

What do customers actually do when they want marketing emails to stop?

Mostly not what the brand expects. Of the coded actions people describe taking against messaging they no longer want, about 27% mark it as spam or junk, about 24% block the sender or delete the app, about 22% click the unsubscribe link, about 17% build a filter that hides it in a folder, and about 11% redirect it to a burner or aliased address. That means roughly four in five coded responses route around the unsubscribe link entirely, and the single most popular alternative is the one action that damages sender reputation for every other recipient on the list.

Is frequency or relevance the bigger reason people unsubscribe?

In the language people actually use, frequency dominates. About 60% of coded exit triggers are the sheer rate of sending, described as daily or several times a day. Another 17% are the same message arriving again with nothing new in it, and about 11% are a message that lands badly on the day it arrives. Explicit irrelevance is named in only about 6% of coded triggers, and permanent sale noise in about 4%. That does not mean relevance is unimportant. It means customers rarely reach for the word relevance. They reach for a number, and the number is how many messages arrived this week.

Why do people avoid clicking the unsubscribe link?

Because a widely repeated belief says clicking it confirms a human is reading. In the coded conversation about the unsubscribe mechanism, about 47% argue the spam button is the tool that actually works, about 24% state a flat rule of never clicking anything inside the message, about 22% specifically say clicking confirms the address is live and valuable, and about 7% report that unsubscribing appeared to make the volume worse. The belief originates in genuine spam and phishing advice, but customers apply it to legitimate brands too, which is why a compliant one-click unsubscribe still goes unused.

Does unsubscribing actually work?

Often not, according to the people describing it. Of the coded complaints about the unsubscribe process itself, about 63% describe unsubscribing and continuing to receive the messages anyway, sometimes across several attempts and several months. About 29% describe a link that is hidden, tiny, disguised or buried inside a clipped message, about 6% describe being made to log in or complete a form to leave, and about 2% describe losing something else, such as account access or rewards, by opting out. Every one of these failures teaches the next customer that the link cannot be trusted.

How is SMS tolerance different from email tolerance?

Text is a narrower permission with a faster penalty. In the smaller SMS conversation in this corpus, about 51% of coded mentions accept a text when the reason is immediately obvious, such as an order, delivery, appointment or restock. About 22% describe an unsolicited marketing text ending the commercial relationship on the spot, up to and including a personal do-not-buy list, and about 20% describe a text as more intrusive than an email because it interrupts. The SMS sample here is small, about 49 coded mentions, so treat the split as a direction rather than a measurement.

What messaging cadence do customers say still feels useful?

A rhythm they picked, carrying something they could not have got by visiting the site. Of the coded reasons people give for keeping a sender, about 42% describe a predictable cadence they chose or could choose, most often weekly or monthly. About 29% describe content that is genuinely useful rather than promotional, and about 27% describe messages that only arrive when there is a real reason, such as a restock or an actual new thing. Only about 2% credit affection for the brand itself, which is the uncomfortable part: liking you is not what keeps you in the inbox.

This report analyzes consumer language and perceptions about marketing email and SMS. Statements about the practices, policies or compliance of any named company are reported as claims made by commenters at the time of writing and were not independently verified. Beliefs described here, including the widely repeated claim that clicking an unsubscribe link confirms an address is active, are reported as what customers believe and repeat, not as verified fact. Nothing here is legal advice on CAN-SPAM, TCPA, GDPR or any other messaging regulation. Brand and product names appear only as they were mentioned by commenters. Percentages are directional estimates from this corpus, not a census of subscriber behaviour.