The most useful finding in this year's email fatigue research is that most subscribers aren't fatigued. Only 24% of opted-in consumers say they'd unsubscribe over receiving too many messages. Fifty-eight per cent say they'd welcome more.
That's from Optimove's 2026 fatigue report, and it sits awkwardly against every article warning you to cut back before Q4. It doesn't mean frequency is free — the send-volume data is unambiguous that it isn't. It means the thing being measured as fatigue is mostly something else.
So: what the frequency curve actually looks like, where its knee sits, and how to plan a peak season against it.
The frequency curve has a shape
The clearest number available comes from a Salesforce Marketing Cloud analysis of 19 billion email sends in 2026.
| Emails per week | Unsubscribe rate |
|---|---|
| 1–2 | 0.07% |
| More than 5 | 0.58% |
Roughly eight times the attrition. And 0.58% sits above the 0.5% level generally treated as the healthy ceiling — so a brand sending daily through December isn't near the edge, it's past it.
Klaviyo's 2026 data across 8,500 e-commerce brands gives the monthly view: the category average unsubscribe rate has climbed to 0.31%, and brands sending more than twelve promotional emails a month recorded rates as high as 0.48%.
Twelve a month is roughly three a week. Which is a perfectly normal November for a retail brand — meaning most e-commerce senders cross that line every year without noticing they crossed anything.
For context on where you sit: median unsubscribe rates in 2026 sit around 0.22%, below 0.5% is the healthy ceiling, and below 0.2% is strong. E-commerce runs higher than average by design.
But the fatigue story is more interesting than "send less"
Here's where the two datasets pull against each other, productively.
Optimove's 2026 finding — 24% would leave over volume, 58% would welcome more — points at relevance rather than frequency as the operative variable. The report's own conclusion is that subscribers aren't fatigued by communication; they're fatigued by poor communication. Repetitive, impersonal, mistimed.
The reframe The question isn't how many emails you can send before people leave. It's how many emails worth opening you can produce — and for most brands that number is considerably lower than their planned send count.
A caution on reading these two datasets as a single story. The 2026 question asks what people would do; earlier waves asked what they had done, and produced far gloomier numbers. Different questions, different answers, and stated intent is a weaker signal than behaviour. The Salesforce figures are behavioural, which is why they carry more weight in a planning decision.
What survives both readings: volume alone doesn't drive people away, but volume of undifferentiated promotional content does — and Q4 is when the differentiation usually collapses.
Sentaro
AI-powered email security that blocks phishing, BEC, and zero-day attacks before inbox delivery.
Best for: Teams on Google Workspace or Microsoft 365
The bulk unsubscribe, which isn't about you
A behaviour worth knowing about because it changes how you read your own numbers.
An April 2026 analysis of newsletter subscriber behaviour found 31% of unsubscribers cited too many newsletters cluttering the inbox and 28% cited repetitive or low-quality content — but the striking figure is that 18% unsubscribed in bulk. Not one sender at a time: a systematic inbox cleanse terminating multiple relationships at once.
Which means a meaningful share of your December unsubscribes were caused by everyone else's over-sending, not yours. Somebody hit their limit with the inbox as a whole and cleared out fifteen senders in ten minutes, and you were one of them.
Two practical consequences. First, you're competing on inbox share rather than on your own restraint — being disciplined doesn't protect you if the rest of the category isn't. Second, the way to survive a cleanse is to be recognisably worth keeping in the two seconds someone spends deciding, which is a positioning problem as much as a frequency one.
Your worst send probably wasn't the one that lost them
The finding that most changes how you diagnose churn.
Subscriber behaviour analysis describes unsubscribes spiking after previously well-performing campaigns, once cumulative fatigue crosses a threshold. Fatigue accumulates; the departure is triggered by whichever email happens to arrive at the tipping point.
So the send with the unsubscribe spike is often not the send that caused it. If you review Q4 by looking for the campaign with the worst opt-out rate and rewriting it, you're fixing the wrong email — and quite possibly one of your better ones.
Look at the two or three weeks preceding instead. Cadence, repetition and relevance over that window are the causal variables; the final campaign is just the one holding the parcel when the music stopped.
Don't cut frequency uniformly
The instinct after reading the numbers above is to reduce sends across the board. That's the wrong response, and it costs revenue for no protective benefit.
Tolerance varies enormously by engagement. Someone who opened three of your last four emails and bought last month will absorb daily sending through December without complaint. Someone who hasn't engaged since March is the person who unsubscribes — or worse, complains — the moment volume rises.
So vary cadence by tier rather than cutting uniformly:
- Recently active and purchasing. Your highest tolerance. Peak-season frequency is fine here and this is where the revenue is.
- Engaged but not recently. Moderate cadence. Send the genuinely differentiated moments, skip the filler.
- Long dormant. Reduce or suppress entirely for the season. These are the people generating your complaint rate, and complaints are a deliverability problem rather than just a churn one — the mechanism in re-engaging a dormant list without wrecking deliverability.
That last tier matters more in Q4 than at any other time, because complaint rate is measured daily and the enforcement thresholds don't relax for seasonality. Increasing volume to a disengaged segment during your highest-revenue month is the single most expensive mistake available in email — it can cost you inbox placement for the segments that were working. The wider picture is in why legitimate email still lands in spam and landing in the inbox consistently.
Doing this properly is a segmentation exercise, and it's the highest-return one available before peak.
Let them choose instead of leaving
Earlier Optimove research found 90% of consumers want the ability to customise the frequency and type of messages they receive. Very few brands offer it.
An unsubscribe is binary and permanent. A preference centre turns the same impulse into a smaller outcome: fewer emails, or only the ones about a specific category, or a pause until January. All of those retain a customer that a plain unsubscribe link loses.
Three things that make one work:
- Offer it on the unsubscribe page, not buried in account settings. That's the moment of decision, and it's the only moment most people will ever look.
- Include a seasonal pause. "Take me off holiday emails but keep everything else" is exactly the option a lot of December unsubscribers actually want, and almost nobody gives it to them.
- Honour it visibly. A preference that's quietly ignored converts a retained subscriber into a complaint, which is worse than the unsubscribe you avoided.
What earns the extra send
If the operative variable is relevance rather than volume, the planning question changes from "how many can we send" to "which of these deserves to exist."
A rough test for each planned December email: could a subscriber tell it apart from the one you sent four days ago, from the subject line alone? If not, it isn't an extra send — it's the same send again, and it counts against you at full weight while returning less.
What consistently earns its place:
- Genuine deadlines. Shipping cutoffs are the strongest email you'll send all quarter, because they're useful rather than promotional — the reasoning in delivery promises that decide peak-season sales.
- Genuinely different offers, not the same discount reframed.
- Behaviourally triggered messages, which are relevant by construction — the flows in email automation 101, and increasingly the personalisation approaches in how AI is changing email personalisation.
- Segment-specific sends that only some of the list receives, which raise relevance and lower per-person volume simultaneously.
What rarely earns it: the third reminder about a sale everyone already knows about, and the "last chance" that isn't. Subject lines can only do so much when the underlying message is a repeat, though they still decide whether the good ones get seen — the craft in writing subject lines that get opened matters more when the inbox is at its most crowded.
The bill arrives in January
One consequence of heavy Q4 sending that rarely gets planned for, because it lands after everyone has stopped looking at the campaign reports.
Email lists decay at roughly 22.5% a year under normal conditions. A December that pushes frequency well past the curve doesn't just produce unsubscribes in December — it produces a January list that is smaller, less engaged and more expensive to reach, at exactly the moment you need it to carry a quiet quarter.
The people who left are also not a random sample. The ones who unsubscribe under pressure are disproportionately the ones who were only mildly engaged — which sounds like good riddance until you remember that mildly engaged is where next year's customers come from. Your best customers won't leave over frequency. Your future ones might.
Worth doing two things about it. First, budget for the attrition rather than discovering it: model your December list size at your planned cadence using the curve above, and see whether the incremental revenue actually covers the subscribers it costs. For many brands the fifth weekly email in December is net negative once January is priced in, and nobody ever runs that calculation because the campaign report only shows the revenue side.
Second, plan a January that earns people back rather than continuing the December pattern into a month with no deadline attached to it. A quieter, more useful January is how a list recovers — and it's cheaper than the alternative, which is discovering in March that your reachable audience shrank while you weren't watching.
What to plan for this Q4
- Count total sends per subscriber, campaigns plus automation, and compare against the frequency curve rather than against last year's plan.
- Set a ceiling by tier instead of one number for the list.
- Suppress the long dormant for the season. Peak is the worst possible time to test their tolerance.
- Ship a preference centre with a seasonal pause before November if you don't have one. It's a small build with a direct effect on December attrition.
- Watch complaint rate daily, not unsubscribe rate weekly. Unsubscribes cost you a subscriber; complaints cost you the inbox.
- Diagnose churn on the preceding fortnight, not on the campaign where the spike appeared.
- Test cadence properly in January, when the stakes are low — the approach in an A/B programme that compounds. Q4 is a bad time to learn anything.
And hold this alongside the demand picture in what early signals say about holiday spend and the sequencing in the 90-day Black Friday timeline. If email frequency is being pushed up to compensate for expensive paid acquisition, that's worth naming as a decision rather than letting it happen — the same trade-off discussed in peak-season paid media. Where the constraint is production capacity rather than strategy, that's an e-commerce lifecycle resourcing question.
The short version
One to two emails a week produces a 0.07% unsubscribe rate; more than five produces 0.58% — eight times higher and above the healthy ceiling. Twelve promotional emails a month is where e-commerce brands start seeing 0.48%, and twelve a month is a normal November. But the 2026 fatigue research says only 24% would leave over volume while 58% want more, which points at relevance rather than frequency as the real variable. Roughly a fifth of unsubscribes happen in bulk inbox cleanses that have nothing to do with you. Churn spikes after good campaigns because fatigue is cumulative, so diagnose the fortnight rather than the send. Don't cut frequency uniformly — vary it by engagement tier and suppress the dormant entirely for the season. And offer a seasonal pause, because a lot of your December unsubscribers only wanted December off.
The question was never how many emails you can send. It's how many you can make worth opening — and that number is set in October, not December.
Planning a heavier December than your list can carry?
We set cadence by engagement tier so peak frequency lands on the people who want it.
Plan my Q4 cadence →Frequently asked questions
How many emails per week is too many?
The clearest available figure comes from a Salesforce Marketing Cloud analysis of 19 billion sends in 2026, which found subscribers receiving one to two emails a week from a single brand unsubscribed at 0.07%, while those receiving more than five a week unsubscribed at 0.58% — roughly eight times higher. Klaviyo's 2026 data across 8,500 e-commerce brands adds a monthly view: brands exceeding twelve promotional emails a month recorded unsubscribe rates as high as 0.48%, against a 0.31% category average.
Are subscribers actually fatigued by email volume?
Less than the coverage suggests. Optimove's 2026 fatigue research found only 24% of opted-in consumers would unsubscribe because of receiving too many messages, while 58% said they would welcome more because they find them useful. The conclusion drawn is that subscribers are not fatigued by communication itself but by communication that is repetitive, impersonal or irrelevant. Volume is the trigger people report; relevance appears to be the actual variable.
What is a healthy unsubscribe rate?
Median rates in 2026 data sit around 0.22%, with below 0.5% treated as the healthy ceiling and below 0.2% considered strong. E-commerce runs higher than average at roughly 0.27% to 0.31%, reflecting more frequent promotional sending. Crossing 0.3% is generally read as a signal to review cadence and content rather than as an emergency, but sustained movement toward 0.5% indicates a frequency or segmentation problem rather than a content one.
Why do unsubscribes spike after a successful campaign?
Because fatigue accumulates rather than arriving with a single message. Analysis of subscriber behaviour describes unsubscribes spiking after previously well-performing campaigns once cumulative fatigue crosses a threshold, which means the send that triggers the departure is often not the send that caused it. The practical consequence is that attributing churn to the last campaign misdiagnoses the problem, and cadence over the preceding weeks is the more useful thing to examine.
Should you reduce email frequency during the holiday season?
Not uniformly, because tolerance varies enormously by how engaged someone is. Recently active subscribers absorb higher frequency without complaint; people who have not engaged in months are the population that unsubscribes or complains when volume rises. Varying cadence by engagement tier protects the list far better than an across-the-board reduction, and it preserves the revenue that peak-season frequency exists to generate.
