Unsubscribe Rate

Email performance used to be read entirely through how many people opened and how many clicked. High on both counts, and the send got reported as a success without further questions.

What those two numbers leave out is what was lost. A send that came back at 42 percent opens can also lose 900 people from the list the same day. Someone who opts out isn't in the population for next month's send, and there is essentially no way back. Unlike traffic you can buy again, this person explicitly asked not to receive anything, so they cannot be added back. The loss is permanent in a way almost nothing else in marketing is.

So sends came to be measured on both sides at once: how many people the message moved, and how many people you can no longer reach as the price of it. The unsubscribe rate measures the second one, which is the half that compounds.

Divide the number who opted out by the number the send reached. Four things to settle.

1. Choose the denominator. It can be messages sent or messages that actually landed in an inbox. The second is more accurate, because a message that never arrived had no chance to produce an opt-out. The share that lands is deliverability, and that figure gets checked before the denominator is settled on.
2. Choose the window. Opt-outs attributable to one send mostly arrive within 72 hours. Without a fixed window it stops being clear which send is responsible, and on a list that receives two sends a week the window has to be short enough that the two don't overlap and blur together.
3. Divide. Reach 12,000 with 24 opt-outs is 0.2 percent. The number is small enough to wave away, and on a list mailed twice a week the same value is a loss of over a hundred people a quarter, none of whom get replaced by the next campaign.
4. Count spam complaints separately. Both mean stop sending, but they carry different weight. An opt-out removes someone from your list; a spam complaint teaches a mail provider not to trust your sending address. Enough complaints and messages to people who never opted out start landing in spam folders. Merged into one column, the more dangerous of the two gets hidden behind the harmless one.

On the same report, this sits next to open rate, the share who opened, and click-to-open rate, the share of openers who clicked. Read as a row, the three say whether this send earned attention, produced action, and cost you people. Any one of them alone can be made to look fine.

The conventional line is under 0.5 percent. Around 0.2 percent is stable, and above 1 percent something about that particular send is wrong rather than something about the list. Thresholds shift with industry and with the character of the list, so comparing against another company's figure is of little use — a list full of people who signed up two days ago and a list full of two-year subscribers respond differently to the same message.

The way to read it is change rather than level. Five sends at around 0.1 percent on the same list followed by one at 0.6 percent points at this send, not at the list. The subject line didn't match the contents, the message was a different kind from what people signed up for, or the frequency just went up without notice. Send a discount announcement to people who subscribed for product news and this figure doubles that day, before anyone has had time to discuss whether the campaign worked.

It also has to be read against frequency. A single send at 0.1 percent still costs 4 percent of a list per quarter at three sends a week. Notification fatigue, the state where alerts come often enough that people ignore or disable them, is usually what a quietly rising figure is describing.

Lowering the number by making opting out difficult is available and costs more than it saves. Someone who can't find the exit marks the message as spam instead, and complaints erode deliverability. So opting out stays a single click, and the alternatives sit next to it rather than behind it: dropping from weekly to monthly, keeping product notices while ending promotions. All-or-nothing loses everyone who merely found the volume slightly excessive, which is most of the people who leave.

"Opens are at 42 percent and unsubscribes are at 0.8."

Reading the gain and the loss from the same send, in the same sentence.

"Make unsubscribing one click. Require a login and it turns into spam complaints."

Obstruction produces the more expensive outcome, and it arrives with a delay.

"Check whether the list grew on net this month."

Arrivals and departures have to be counted together before anyone calls the month good.

At Pagedeck, this number reacted first in the quarter Ravi raised the sending cadence to twice a week. Open rate stayed around 40 percent, while the unsubscribe rate went from 0.15 percent to 0.55.

Rather than reverting the cadence and losing the reach it bought, Ravi added an option. The opt-out screen gained a "reduce to once a week" choice, and 38 percent of the people heading for the exit took it. The next quarter's rate came back to around 0.2 percent and the list returned to net growth. People who had wanted less had simply had nowhere to say so, and leaving was the only control available to them.

The most common mistake is reading it as a number that should be as low as possible. It measures the people you lost, so zero looks like the ideal. But a figure near zero also appears when nobody opens the mail at all, since messages that go unread give nobody a reason to leave. The number only means something beside the open rate. Both low means you're outside people's attention entirely; a high open rate with a high opt-out rate means they read it and were disappointed by what they found.

Reading a shrinking list as pure loss is the other one. Fewer recipients means less reach, so the arithmetic is natural. But the remaining list improves in quality. Lots of addresses that never open teach mail providers to distrust the sender, which lowers deliverability; once the non-openers have left, the same send lands in more inboxes. Set list size as the target and that improvement reads as damage, and you hold on to dead addresses at the cost of getting delivered.