Key takeaway
A re-engagement campaign is a short sequence sent to the contacts who've stopped responding, and it ends one of two ways: they answer, or they come off the list. Treat it as a permission audit rather than a win-back. I define inactive on clicks, purchases and the replies my own systems record instead of opens, stage the send so it can't push complaints past 0.10%, and judge the result on the list I keep.
What a re-engagement campaign is.
A re-engagement campaign is a short sequence sent to the people on your list who have stopped responding, with two honest outcomes: they answer, or they come off the list. It's a permission audit rather than a revenue campaign. The list you keep afterwards is the product, and its price is a block of addresses you have to agree to lose before you press send. I've spent a decade watching senders run it the other way round: a rescue mission started after the numbers fell, judged on the click, and never run again. I've never seen that pattern break.
The mechanics are conventional: a defined segment of unengaged contacts, a short automated sequence rather than one broadcast, and every contact in it finishing in one of two states.
Yahoo asks senders to monitor inactive recipients and to consider sending a reconfirmation email to inactive subscribers in its sender best practices. That's the mailbox provider saying what it wants, and it isn't asking for a win-back. Framed as a win-back, the campaign is judged on how many people came back, and that number is always small. Framed as an audit, it succeeds either way. The exception is the sender who already has a reputation problem. If today's mail to the people who do read isn't reaching the inbox, this campaign isn't the first job. Running it first points the worst addresses on the list at a reputation that can't absorb them.
Decide what inactive means on your own schedule.
Inactive is a threshold you set. It's a length of time with no response, and it only means something relative to how often you send. Somebody who hasn't answered in ninety days has ignored roughly a dozen emails from a weekly sender and two from a monthly one, and those are two different people wearing one label.
The advice in circulation runs from a single month of silence to a full year. A spread that wide isn't a disagreement to settle by picking the middle. It says the number belongs to the sender rather than the category, so what's worth copying is the rule that generates one: pick the point where non-response has happened often enough to be a signal rather than an accident. That's a count of sends first and a number of days second.
Two inputs decide it, and both are yours. The first is how often you send. The second is how long a normal buying or renewal cycle runs in this business, because a contact who's quiet inside an eighteen-month renewal cycle is behaving as expected. A monthly B2B sender and a retailer sending daily can't share a threshold. The retailer's is shorter by a factor of several, and both of them are right.
A threshold is a guess until the first campaign has run against it. What makes it a policy instead of a guess is stating the number and the reason for it somewhere your stack can act on rather than somewhere a person has to remember, then moving it on evidence rather than on feel.
Stop defining inactive by opens.
Almost every version of this campaign builds its segment on opens, and the open is the one signal you no longer control. Apple's Mail Privacy Protection prevents senders from seeing whether a message was opened, so a segment defined that way is partly a report on which mail clients people use.
The damage is specific, and it runs in both directions. Apple loads remote content in the background whether or not the message is ever read, so people who read nothing still register as opens and never enter the segment at all. Anyone whose mail client doesn't load images runs the other way: they read everything, register nothing, land in the segment, and get deleted by the removal step at the end of the campaign. Build the segment on things a person did instead.
- Clicks: the last time they opened a link, which needs no image to load and is still theirs.
- Replies: the strongest thing a person can do inside the inbox, and the one almost nobody stores, because the record of it sits in your CRM, your helpdesk, or the mailbox it arrived in.
- Account activity: logins, sessions, or anything they did inside your product rather than inside their inbox.
- Purchases and renewals: the event the whole programme exists for, which outranks every other signal here.
Rank them, because they aren't equal, and set the threshold against the strongest signal the business actually records. A product with logins has a better answer available than clicks. A newsletter has clicks and replies, and that's fine as long as somebody chose the definition rather than inheriting it from a reporting screen.
One case this doesn't cover: a list that never clicks because nothing in the emails is clickable doesn't have an engagement problem. Re-engaging it won't fix the sends that made it quiet, and asking why they went silent gets an accurate answer nobody wanted.
Three moves, and the third one does the work.
Sequences in the wild run anywhere from one email to five, and the count isn't the interesting part. What matters is that three moves happen, in order, and most sequences only ever make the first two.
- The ask: remind them what they signed up for and where they did it, then give one obvious action.
- The choice: offer less email rather than none, so leaving the programme isn't the only way to stop the noise.
- The notice: tell them plainly that this is the last email unless they act, and then honour it.
The third email is the only honest one in the sequence and it's the one that performs, because it's the only message that gives a reader a reason to act now. It's also the one most senders cut, because writing it feels like giving up. That reading is backwards. The notice is what turns the campaign into an audit rather than a request, and a sequence without it has no ending, so the same dead segment is still there next quarter.
Two things get dropped into this sequence that don't belong in it. A discount buys a click from somebody who wants a discount, and proves nothing about permission. A cart reminder is a transaction prompt with a different trigger and a different programme behind it, and putting one here answers a question about a purchase when you asked a question about consent.
The limit: somebody who clicks the third email once isn't re-engaged. They're on probation, and the next threshold applies to them the same as it applies to everybody else.
Send it the way you would send to a cold list.
This send goes to your worst addresses, in bulk, in one go. That's the profile of a complaint spike, and it's the shape of send that finds abandoned addresses which have quietly been repurposed as traps. The campaign meant to protect your sending reputation is the one most likely to damage it.
Stage it. Start with the least stale cohort, hold the rest back, and read the result before the next batch leaves. The numbers to read against are published by the receivers themselves. Google asks senders to keep the spam rate below 0.10% in Postmaster Tools and never to reach 0.30%, and Yahoo sets the same 0.3% ceiling in its own guidance. M3AAWG's sender practices give the companion rule for the addresses that fail outright: remove an address that bounces consecutively at least two times over two weeks or more.
Then hold yourself to the result. If a batch moves the complaint rate, the campaign stops there and the cohorts still waiting come off the list instead of getting mailed. A win-back that costs the inbox placement of the list you kept has lost more than it could ever have returned, and that trade gets made every time the sequence is scheduled as one send to everybody.
Write the sunset rule down before you send.
A sunset rule is four things written in one place: the threshold, the sequence, what happens to somebody who doesn't answer, and when the whole thing runs again. The fourth is the one nobody writes down, and it's what turns a campaign into a policy, because a policy is something a system runs on a schedule without anybody deciding to run it. Quarterly is the right shape, since a list decays continuously and a rescue mission only starts once the damage shows up in the numbers.
Two things constrain the rule. Removal has to be real: the opt-out has to work on the first click, keep working long after the campaign is over, and take effect before the next send rather than at the next export. An email announcing that it's the last one, followed by three more, is worse than not sending it at all.
Removal also means suppression rather than deletion. Deleting the record loses the fact that this person asked to be left alone, and the address walks back in on the next import from the CRM or the event list. Suppression keeps the decision. Contacts are unlimited on every Nitrosend plan including Free, so a growing suppression list isn't something you pay for by the row.
The exception: the rule can't cover a contact who's unengaged in email and busy everywhere else in the business, and every list has a few. Name them as exceptions, with who decides. A policy with no named exceptions gets overridden informally, one account at a time, until nobody can say what it is.
Judge it on the list you keep.
The campaign has metrics and the list has a result, and the two get confused because the platform reports the first one. Four measures are worth reading afterwards.
- Reactivation rate: the share of the segment that answered, which is always small and isn't the point.
- Removal count: how many addresses came off the list, which is the campaign doing the job it exists for.
- Complaints and bounces after: the two numbers the receivers are watching, so read the complaint rate and the bounce rate on the sends that follow.
- Engaged-list performance: how mail to everybody who stayed performs from here on, which is where the return sits.
The reactivation rate is the number every version of this campaign ends on, and it's the least useful of the four. I'd report the removal count first and the reactivation rate last, and expect an argument about it. A campaign that wins nobody back and takes four thousand dead addresses off the list has succeeded. The uncomfortable half is worth saying out loud: a shrinking list with rising engagement is a healthier asset than a growing one nobody reads, and almost no reporting is built to tell you that. That's not a house opinion either: engagement is the primary signal now, and a chronically unengaged list out-loses a smaller engaged one on every metric that matters.
The fourth measure needs several sends before it shows anything, so this campaign can't be judged in the week it runs. Whoever asks for a number on the Friday gets handed the reactivation rate, which is roughly how it became the headline.
Make the policy something your stack can run.
A sunset rule is a handful of numbers and a short sequence. In most teams it lives in a builder canvas somebody drew once, and the reasoning behind it lives nowhere. Nine months later the threshold is still ninety days because ninety days is what the template said, and the person who could argue with that has moved on.
A rule short enough to state in a sentence should be short enough to run as one. Nitrosend is MCP-first: every capability is an API endpoint and an MCP tool before it's a screen, so segments, campaigns, flows, and suppression are things an agent creates, changes, and explains back to you. They're built against the click, purchase, and reply activity your own systems push in. Every existing email platform was designed for humans clicking buttons, and an assistant added to one of those still has to drive the same builder canvas a person would, screen by screen. Quarterly list hygiene is exactly the work nobody wants to click through four times a year, whoever is doing the clicking.
If you've got a segment that stopped answering and no written rule for what happens to it, that's the job Nitrosend was built to take. State the threshold, the three emails, and the suppression once, then let an agent build it from the activity your systems send in and run it again every quarter. Start free, unlimited contacts included, and write the sunset rule this week.
Sources
- Yahoo, Sender Best Practices: monitoring inactive recipients, sending a reconfirmation email to inactive subscribers periodically, and keeping the spam rate below 0.3%.
- Google, Email sender guidelines: keeping spam rates reported in Postmaster Tools below 0.10% and never reaching 0.30%, and unsubscribing recipients who don't read your messages.
- Apple, Use Mail Privacy Protection: the feature prevents senders from seeing whether a message was opened.
- M3AAWG, Sender Best Common Practices v3: removing an address that bounces consecutively at least two times over two weeks or more.
- Apple, Mail Privacy Protection: Protect Mail Activity downloads remote content in the background by default, regardless of whether you engage with the email.
Common questions
A click on the last email of a re-engagement sequence takes that contact back out of the inactive segment, and that's all it proves. One click on a final notice isn't a restored relationship, so treat that person as being on probation: the same inactivity threshold applies to them from that day, and if they go quiet again they're in the next campaign with everybody else. If the sequence offered a lower frequency, move them onto it, because the reason they stopped answering is usually the volume rather than the sender.
There's no universal number. Published advice spans one month of silence at one end and a full year at the other, a spread wide enough to tell you nobody else can set this for your list. Derive it instead: how many sends has this contact ignored, and is that enough of them to read as a decision rather than a coincidence? A weekly sender hits that point far sooner than a monthly one, and a long buying or renewal cycle pushes it back out, so the sends come first and the calendar second.
Enough to make three moves. The ask reminds them what they signed up for and gives one obvious action. The choice offers less email rather than none. The notice says plainly that this is the last email unless they act. The count matters far less than whether the third move happens, and most sequences stop after the first two, which is why they never actually end.
Remind the reader what they signed up for and where they did it, then give one action that takes a single click. Offer a lower frequency as an alternative to leaving. In the last email, say plainly that it's the last one unless they act, and then honour that. The sequence is asking one question, which is whether this person still wants the mail, and every sentence in it should be serving that question.
No. An offer tests price sensitivity, and the only thing this campaign is measuring is permission, so a click bought with money answers a question you didn't ask. The contact who takes the deal and goes quiet again is back in the same segment next quarter, and you paid for the round trip. Discounts belong in the programmes built to sell. Leave this one asking a plain question.
No, and deleting them is the version that goes wrong. A deleted contact leaves no trace of having asked to be left alone, so the next sync from your CRM or your event list treats that address as new and it gets mailed all over again. Suppressing the contact stops the sending and keeps the record of why, which is the part you need six months later.
Once, deliberately, and then stop. Mail that nobody reads damages the reputation of everything else you send, so a message you're entitled to send can still be a message you shouldn't. If you run the campaign, the removal at the end has to be real: an email announcing it's the last one, followed by three more, is worse than not sending it at all.
Judge it on the removal count and on the list's behaviour afterwards, not on the reactivation rate. Zero contacts recovered and four thousand dead addresses retired is a good outcome, because the addresses were the problem. What you're watching for after that is the complaint and bounce rates on the next few sends, and whether mail to the people who stayed performs better than it did before you started.