An unsubscribe is the one negative signal you asked for. You put the link in the message. Someone used it. That makes it the rare engagement metric where a rising number can be the better outcome.
Why is a higher number sometimes good?
Because the alternative to an easy opt-out is the spam button.
RFC 8058, which specifies one-click unsubscribe, gives the behavior it was written to prevent:
Hence, the mailers want to make it as easy as possible for recipients to unsubscribe; if an unsubscription process is too difficult, the recipient's alternative is to report mail from the sender as junk until the mail no longer appears in the recipient's inbox.
A person who wants your mail to stop will make it stop. The only question is which button they use. One of the two damages your sender reputation. The other does not.
So an unsubscribe rate that rises while complaints fall is a program getting healthier. Suppressing the unsubscribe to protect the number converts cheap departures into expensive ones.
What is a normal rate?
No standards body or mailbox provider publishes one.
Mailbox providers publish complaint-rate thresholds because complaints feed their filtering. None of them publishes an unsubscribe benchmark, because an unsubscribe is a transaction between you and your recipient that the provider mostly cannot see. The figures in circulation come from vendor aggregates over other companies' lists, averaged across audiences, frequencies and acquisition methods that have nothing to do with yours.
What is available to you is your own history. A single list, mailed at a steady cadence, produces a stable rate. Departures from it tell you more than a cross-industry average can.
How do I read a change in it?
By its shape, which points at different causes.
| What you see | What it usually means |
|---|---|
| One campaign well above your own baseline | Something about that message: the subject, the offer, or the audience it went to |
| A slow rise across every campaign | Frequency, or a change in how people are getting onto the list |
| A spike right after a list import | The imported addresses did not expect you |
| Near zero on a large list | A broken opt-out path, more often than a beloved newsletter |
A broken opt-out is both a compliance failure and a complaint generator, so treat a rate of zero as something to check. Send yourself a campaign and click the link.
What is the relationship to complaints?
They are the same decision made through two different doors. Only one door has a published ceiling.
Complaint rate has thresholds from Google and Yahoo and direct consequences for placement. Unsubscribe rate has neither. Read together, they say something neither says alone:
- Unsubscribes up, complaints down. The opt-out is doing its job.
- Both up. The mail itself is the problem rather than the mechanics. Look at frequency and at what changed.
- Complaints up, unsubscribes flat. People cannot find the way out, or do not trust it. Check that the one-click unsubscribe headers are present and that the endpoint works.
Can I reduce it without hiding the link?
Yes. Change the frequency, or offer a choice short of leaving.
Frequency is a lever on the whole list at once. Choice means letting someone leave one stream instead of all of them, which is what a preference center is for. It converts a full departure into a partial one.
Making the exit harder is not an option the rules leave open. CAN-SPAM gives you ten business days to honor an opt-out. CASL gives the same ten. Google and Yahoo expect a one-click unsubscribe request honored within two days.
In short
An unsubscribe is the cheap version of a complaint.
The alternative to an easy opt-out is the spam button, which costs far more. A rise in one beside a fall in the other is an improvement.
Nobody who would know publishes a benchmark.
No standards body or mailbox provider publishes a normal figure. The numbers in circulation are vendor aggregates over other people's lists.
Read it against complaint rate, never alone.
Complaint rate has published ceilings and real consequences. Unsubscribe rate is only interpretable beside it.
A spike is a message problem; a slow rise is a frequency problem.
One campaign out of line points at what that campaign did. A steady climb points at cadence or at how the list was built.