First, locate the spike: which send, which flow, which list source
Pull unsubscribes for the last few weeks with three attributes attached: the message that carried the exit link they used, the segment or tier they belonged to, and the source recorded when they joined. Then ask three questions in order. Is the spike on one campaign, on several campaigns in a short period, or on an automated flow that runs every day? Within that send, is it spread across segments or concentrated in one, such as people who have never bought or people in the least engaged tier? And among those who left, does one sign-up source or import account for far more than its share?
Read the reasons people give where the platform collects them, and read complaints beside unsubscribes, because the two move together when expectations were broken and apart when the exit path is the problem. If complaints are rising while unsubscribes are flat, the problem is more likely an exit that is hard to use, which emails going to spam covers. The answer to the three questions usually points to one row of the table below.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Exits rise across segments on a change of format, topic or frequency | The sign-up wording against what was sent recently | Content or frequency different from what the sign-up promised | Restore the promised content and cadence, or ask subscribers to opt into the change |
| Promotional sends lose subscribers mainly among people who have never ordered | Unsubscribes by purchase history on promotional sends | Offers sent to subscribers who never buy | Send non-buyers the content they joined for; keep hard offers for buyers and engaged tiers |
| One source loses subscribers at many times the rate of the rest | Unsubscribes and complaints by sign-up source or import | Imported or co-registered addresses that never expected the mail | Stop mailing the source as part of the main list; re-permission or remove it |
| A cluster of exits during a sale season, spread across sends | Sends per subscriber per week against the usual pattern | Cadence raised for a seasonal push without warning | Cap weekly sends; announce the season and offer a lighter option |
| Exits steady on campaigns but high on one automated flow | Unsubscribes per flow message; gaps between messages; overlap with campaigns | A flow sending too many messages too close together | Widen the gaps, cut the weakest messages, hold flow entrants out of campaigns |
Fix, verify and prevent each cause
The mail no longer matches the promise
Every subscriber agreed to something specific: a fortnightly digest, recipes, early access, the sentence next to the form. When the programme drifts, adding a sales email, switching topic or sharply raising frequency, people who joined for the original promise leave in a cluster. The list building and consent guide traces most churn to this gap.
Fix: compare the wording on each sign-up point with the last month of sends, and either return to the promise or tell subscribers about the change and let them choose, routing those who want less into a lighter stream.
Verify: unsubscribes on the next comparable sends return to their earlier level and complaints do not rise in their place.
Prevent: keep the sign-up promise in the campaign brief template, so every planned send is checked against what subscribers were told.
Offers aimed at people who never buy
A list contains readers who joined for advice and have never placed an order, and a run of discount emails sent to everyone tells them the newsletter has become a catalogue. Their exits concentrate on promotional sends while buyers barely move.
Fix: use the lifecycle and engagement segments in the segmentation guide so that non-buyers receive the content they joined for, with offers woven in lightly, and reserve hard-sell sends for buyers and the most engaged tiers.
Verify: on the next promotional send, unsubscribes among never-purchased subscribers fall while the offer's conversion rate (CVRShare of visits or clicks that convert.) among buyers holds.
Prevent: require an audience decision for every promotional send, rather than the whole list by default.
Addresses that never expected the mail
Imported spreadsheets, co-registration offers and lists collected at events contain people who gave an address for one thing and receive another. They unsubscribe fast, and those who do not complain. If one source accounts for a disproportionate share of exits, it is a consent problem showing itself through the unsubscribe report.
Fix: move the source out of the main list into a holding segment, send one re-permission message, and keep only those who confirm, as the list building guide describes for imports; the ICO's direct marketing guidance explains when consent is needed for electronic marketing in the UK.
Verify: exits on the main list return to normal once the source is separated, and the confirmed remainder behaves like other subscribers.
Prevent: record a source on every address and stage every import before it joins regular sends.
A seasonal cadence nobody announced
Peak trading tempts every team to add sends: a preview, a launch, a reminder, a last chance, an extension. Each looks reasonable alone; together they multiply what subscribers receive in a week, and exits spread across all of them. The campaigns hub argues that promotional sends should sit inside the promised frequency rather than on top of it.
Fix: set a cap on marketing sends per subscriber per week for the season, send the extra messages only to the engaged tier, and announce the busy period in advance with a one-click option to pause until it is over.
Verify: exits per send during the next peak stay close to the off-season level, and the pause option is used instead of unsubscribe by a visible share of people.
Prevent: plan the seasonal calendar a quarter ahead and count sends per subscriber, not sends per campaign. The tutorial on running a sale or launch by email builds that count, the pause option and buyer suppression into the sequence before the first email goes out.
A flow that crowds the inbox
Automated flows run without anyone pressing send, so a sequence that fires several messages in a few days, or overlaps with the weekly newsletter, can produce a steady stream of exits nobody connects to a decision. Welcome series are the usual suspect, because new subscribers receive both the series and regular campaigns.
Fix: read unsubscribes per message in the flow, remove or merge the messages that lose the most people, widen the gaps, and hold people in the series out of campaign sends until it ends, as the welcome series tutorial advises.
Verify: unsubscribes per entrant fall over the next month while the flow's click-through (CTRShare of impressions that result in a click.) holds or rises.
Prevent: review each live flow's exits per message on a calendar, and set a priority rule for subscribers who qualify for two flows at once.
Prevention: expectations written at sign-up and cadence rules
Unsubscribe spikes are cheap to prevent because almost every cause is a decision made before the send. Two written documents do most of the work. The first is the promise: for each sign-up point, the exact sentence a subscriber saw about content and frequency, kept where campaign planners can read it. The second is a set of cadence rules: the normal number of marketing messages per subscriber per week, the seasonal maximum, whether flow entrants also receive campaigns, and which segments may receive promotional sends. With both in place, the pre-send check becomes a comparison rather than a judgement call. Where the normal cadence itself is in doubt, an email frequency test sets it on evidence, with unsubscribes as a guardrail rather than an afterthought.
The exit path deserves its own check. A visible unsubscribe link in every marketing message and the one-click headers described in the one-click unsubscribe standard keep people who want to leave from reaching for the spam button, and Google's sender guidelines document both for bulk senders. Offer a lighter option beside the exit, such as a monthly digest, and make sure it genuinely reduces what the person receives. Finally, report unsubscribes per send, per flow message and per source every month as a list-health indicator (KPIA metric tied directly to an objective.), so the next drift shows up as a small rise in one place rather than a spike across the list.
A cookware brand's November exits
A cookware brand with 54,000 subscribers normally sends one recipe newsletter a week and loses about 0.15% of recipients per send. In November its unsubscribe rate jumps to 0.9% on four consecutive sends. Located by send, the exits fall on six promotional emails added for a sale period, on top of the weekly newsletter. Located by segment, 70% of the exits come from 31,000 subscribers who have never ordered and joined through a recipe-guide download promising "a weekly recipe". The team stops sending the remaining sale emails to never-purchased subscribers, sends them one note explaining the sale and offering a "recipes only until January" option, which 3,400 choose, and keeps the sale sends for customers and the active tier. Unsubscribes on the next three sends return to 0.2%, and sale revenue is within 4% of the forecast.
Common mistakes
Reading the list-wide rate instead of locating the send. Rewriting copy when the audience was the problem. Hiding the unsubscribe link to protect list size. Offering a "fewer emails" option that changes nothing. Merging an import into the main list before it has been asked. Adding seasonal sends one by one without counting the total per subscriber.
