Re-engagement: winning back subscribers who stopped

A re-engagement campaign aimed at the wrong list does damage rather than nothing, so the definition comes first.

The obvious definition — has not opened in ninety days — is no longer sound. Apple Mail's Privacy Protection routes messages through Apple's proxies and preloads remote images, tracking pixels included, before the recipient does anything, so an untouched mailbox can report opens indefinitely. Corporate security gateways fetch images for the same technical reason. Building a dormancy list on opens therefore selects partly on which mail client somebody uses.

Define it on actions that need a person: clicks, replies, purchases, bookings, logins. Then set the window against your own purchase cycle rather than a borrowed number. A weekly newsletter can reasonably treat three months of silence as dormant. A Surat textile wholesaler whose customers order twice a year cannot, and a homestay taking annual bookings certainly cannot. Getting the window wrong in the strict direction means writing a win-back campaign to people who were behaving exactly as expected — which reads, to them, as a business that has not been paying attention.

A re-engagement series works better than a single message, because the first one may simply not be seen — these are, by definition, people who have not been looking.

Three messages over roughly two to three weeks is the common shape. The first asks the question directly: we have noticed you have not read anything for a while, do you still want this, here is exactly what you would keep receiving. The second stops asking and demonstrates — the single most useful thing you have published or the offer with the widest appeal, chosen to be worth opening on its own merits rather than as an appeal to sentiment. The third is the last call, and it states the consequence plainly: if we do not hear from you, we will stop sending, and here is the one link that keeps you on.

The sequence should end. A win-back series with no defined last message becomes an indefinite campaign to people who have already answered by not responding, and it is indistinguishable from the problem it was meant to solve.

The reflex is to lead with money, and it answers a different question than the one being asked. A discount tests price sensitivity. Re-engagement is testing interest, and the two produce different lists.

It also has three costs. It trains the readers who respond to wait for the next discount, which changes the economics of every subsequent campaign. It spends margin on people who would have bought anyway. And it recovers a group defined by wanting a deal rather than wanting to hear from you, which is the segment least likely to still be engaged next year.

The alternatives are cheaper and select better. Remind them what they originally signed up for, which is frequently the actual problem — the newsletter has drifted from what was promised at the form. Offer a lower frequency, since volume is a common reason for going quiet. Offer a narrower topic. Each of these gives a reader a way to stay that fixes the reason they stopped, which a discount does not.

The design that works puts two clear options side by side: keep me subscribed, and unsubscribe me. Equal prominence, no ambiguity about which is which.

This feels self-defeating and is not. The people who take the exit were already gone; the only question was whether they left through your link or through the spam button, and the second option damages delivery for the entire list. A clean unsubscribe at the end of a win-back series is the successful outcome of the campaign for that recipient, not a failure of it.

One detail worth getting right: make the 'keep me' action a single click that visibly works, and treat it as a genuine engagement signal that resets their status. If somebody confirms they want your mail and then receives the third 'we are about to remove you' message anyway because the series was scheduled rather than conditional, you have demonstrated that clicking made no difference — which is a worse position than before you asked.

A dormant segment is, by construction, the audience most likely to complain and least likely to engage. Sending all of it at once concentrates the worst signal you have into a single burst, which is precisely what a mailbox provider is watching for.

So break it into batches over several days or weeks, and watch complaint and bounce rates between batches with an explicit threshold at which you stop. Google's guidance to keep spam rates below 0.30%, and preferably under 0.10%, is the reference point; a dormant batch running well above your normal rate is a signal to halt rather than to push through the remainder.

Send it from your established, well-authenticated sending identity rather than a new domain. A fresh domain has no reputation to spend, and this is the mail that will spend the most of it. Expect a bounce spike too — some of these addresses have been dead for months, and the bounces are useful information you should suppress on immediately rather than retry.

When the series ends, act on the result. Addresses that did not respond go to suppression, and suppression rather than deletion, so a later import cannot bring them back.

The reason to actually do this — rather than leave them on the list because a bigger number feels better — is that a dormant address is not free. There is the direct cost per message or per subscriber. There is the reputational drag: every send that includes them produces engagement statistics worse than your real audience would produce, and that is what filtering reads. And there is spam-trap exposure, because providers recycle some abandoned mailboxes into addresses that exist only to identify senders mailing unmaintained lists.

That last cost is the one to internalise. The price of keeping people who will never read your mail is not paid by them. It is paid by the subscribers who do want it, in the form of your messages arriving in their spam folder instead of their inbox.