Peak does not fail in November. It fails in August, while everybody is looking at creative and arguing about discount depth, and the failure only becomes visible on the one weekend where being visible matters most.

Almost every peak email guide gets published in October, which is roughly two months after the point where the advice would have been useful. By October, your reputation is what it is, your list is what it is, and your options have narrowed to choosing a subject line and hoping. Nobody writes the August post because August does not feel urgent, and that is precisely why August is where the work belongs.

So none of what follows is about offers, calendars, countdown timers or how many sends to run on the day. Plenty of people write that and some of them write it well. What follows is the infrastructure underneath all of it, which is the part that decides whether any of the clever stuff ever reaches a human being.

Black Friday falls on 27 November. Count the weeks back from there, and you will see why the timing of this post is the argument it is making.

One more thing before we start, because it shapes everything else. Peak planning is a risk exercise, not a growth exercise. Every decision you are about to make adds load to a system, and your job is to know how much load that system will carry before you add it. There are three risks in play and they are worth naming separately, because teams usually only think about the first one.

• Risk to your deliverability. Volume, complaints and reputation, and whether your mail arrives at all across the weekend that pays for the quarter.

• Risk to the customer experience. Whether the person on the other end feels served or besieged, and whether your order confirmations survive your own marketing.

• Risk to the following quarter. What you inherit in January from decisions you made in November, which is the cost nobody prices in at the time.

Hold those three in your head as you read. Everything below is either measuring one of them or reducing one of them.

The environment

The inbox you are landing in, and the person reading it

Almost every peak plan gets built as though the recipient is sitting in the same inbox they occupy in June, in the same mood, with the same tolerance. They are not, and skipping past that is how technically competent programmes still manage to annoy everybody.

Your email in late November is not competing with your last email. It is competing with every brand that person has ever bought from, all shouting on the same morning, several of them three or four times a day.

What happens to inbox volume

Proton’s 2025 analysis of retail sending found inbox volume rose by around 93% across the holiday peak, with some department stores sending three or four messages a day, every day. Sinch reported that volumes across its platforms rose roughly 30% year on year on Black Friday itself, and around 32% across November as a whole. Braze recorded more than fifty billion messages across the BFCM period in a single year.

Sit with what that means for one human being. Their inbox roughly doubles, the extra volume is almost entirely promotional, and it arrives in a fortnight. Whatever attention you were getting in October, you are getting a fraction of it in November, and your beautifully considered campaign is a single line in a very long list.

How people shop now, which is not how they shopped five years ago

• The single day is over. Sinch’s consumer research found nearly half of shoppers want promotions at least a full month before Black Friday, which has turned a weekend sprint into something closer to a six week campaign season.

• They plan, compare and wait. People increasingly decide in advance what they intend to buy and hold out for the discount, which means your November email is often confirming a decision rather than creating one.

• Deal fatigue is real. By the third week of constant discounting, urgency stops working because everything is urgent, and the brands still shouting are simply louder rather than more persuasive.

• Email is still where they want it. The same research found more than half of consumers prefer email for peak promotions, and a much larger share rely on it for order and delivery updates. The channel is not the problem, the volume is.

They are also being scammed, and it makes them suspicious of you

Peak is open season for fraud. Bitdefender’s analysis of Black Friday themed spam found that roughly three quarters of it was outright scam rather than aggressive marketing, up from around seven in ten the year before.

Which changes how your email is read. Somebody who has already deleted four fake delivery notifications that morning is scanning for reasons to distrust a sender, and the things that make an email look suspicious at peak are the same things peak encourages: a new sending domain, a mismatched from name, urgent countdown language, an offer that seems too good, and a reply-to address nobody recognises. Every shortcut you take on authentication or sender identity is being judged by a reader who is primed to be wary.

The wider world does not pause for your campaign

November is also the month when household budgets are tightest before Christmas, when financial pressure is most present, and when whatever else is happening in the news is competing for the same attention. People are more discount-driven and less brand loyal at peak, which is exactly why so many of them joined your list in the first place, and it is worth remembering that a shopper hunting for money off is not in a receptive mood for a brand relationship.

None of that means send less to everyone. It means the tone that works in June, the frequency that works in June and the assumptions that work in June should all be re-examined, because you are talking to somebody in a different state of mind.

Complaints go up at peak, including yours

The part that turns all of the above into a deliverability problem rather than a marketing one.

Spam complaint rates rise across the whole industry during peak, and they rise disproportionately. Validity found that while sending volume climbed steadily through the season, complaint volume spiked harder than volume did over Black Friday and Cyber Monday. The Certified Senders Alliance, looking across Q4 data from mailbox provider partners, found the same paradox: the more campaigns get sent, the more the risk indicators climb, with spam trap hits and complaints rising together, and the resulting reputation damage taking weeks or months to recover from long after the short term revenue has been banked.

Four things drive it, and none of them is about your creative.

• People are clearing the decks. Peak is when inboxes become unmanageable, so it is also when people purge, and the spam button is the fastest tool available to them.

• The complaint button is being used as an unsubscribe. If your opt-out takes more than one obvious action, somebody in a hurry will take the easier route, and the easier route damages you.

• They have forgotten who you are. Anybody you have not contacted since last November is receiving mail from a stranger, and strangers get reported.

• It is happening on a phone. Mobile clients put the junk control within very easy reach, and somebody triaging in a queue is not hunting for your footer.

What this changes about how you measure the season

Your ordinary benchmarks do not apply in November, so comparing peak week against an average week will have you panicking about a rise that was always coming, or relaxing about one that was not. Set a peak specific expectation for each type of email before the season starts, using last year as the comparison rather than last month, and adjust it for anything that changed in between.

Non-negotiable one

Audit first, because you are measuring risk

Not as a nice-to-have, and not in October when the plan is already signed off. You cannot plan a volume increase against infrastructure you have not inspected, in the same way you would not book a hundred covers into a kitchen without checking whether the ovens work.

The framing that matters is that a deliverability audit is a risk assessment in simple terms. You are about to put the heaviest load of the year through a system, into an environment that is already pushing your complaint rate upward, and the audit tells you how much of that load the system will carry. Everything you decide afterwards, how far you reach, how fast you ramp, how much you send, is a decision about risk that you cannot make sensibly without these answers.

So the audit has to cover the following, and each item is there because it is a place risk hides:

• Authentication (ovbs). SPF, DKIM and DMARC in place and aligned, and whether your DMARC policy is sitting at enforcement or quietly doing nothing at p=none. A record that exists is not the same as a record that protects you.

• Your domain and subdomain architecture. What shares reputation with what. If your promotional sends and your order confirmations are travelling on the same reputation, peak is the season where that decision comes due.

• Where your mail is landing right now, by provider. Not your delivered rate, which counts spam-foldered mail as a success. Gmail, Microsoft, Yahoo and Apple all behave differently, and a blended figure hides whichever one is failing.

• Complaint and bounce rates cut by segment and by source. One acquisition source is almost certainly producing a disproportionate share of both, and you want its name before you start increasing volume.

• Every automation that will collide with your peak campaigns. Abandoned basket, browse, post-purchase, replenishment, and win-back flows all keep running underneath your campaign calendar, and somebody is about to receive six emails in a day from you without anyone deciding that should happen.

• Your engagement make-up. Specifically how large your dormant group is, because that group is the ballast you will be dragging into the busiest sending period of the year.

Some of what the audit finds will take weeks to fix and weeks more to take effect, which is the whole reason it goes first. Authentication changes need to propagate and settle. Suppression needs time to show up in your reputation. Subdomain separation needs a warming period of its own. None of that can be done in the week before you launch, and attempting it in November is how people turn a manageable problem into an emergency.

Non-negotiable two

Know your reputation before you touch your volume

Most teams have no idea what their sender reputation looks like, which means when something goes wrong in peak week they cannot tell whether they caused it, whether it was already like that, or whether the whole internet was having a bad day. Without a baseline you are debugging blind at the worst possible moment.

Delivered is not inboxed, and the gap between those two words is where most peak disappointment lives. Your ESP will tell you 99% delivered while a meaningful share of that mail sits in a spam folder nobody opens on a Friday in November.

Where to look

• Google Postmaster Tools. Domain and IP reputation, spam rate, authentication pass rates and delivery errors, straight from the provider that will handle the largest slice of your consumer list.

• Microsoft SNDS and the Smart Network Data Services picture. Less generous with data than Google, still worth having, and important if any part of your audience is B2B.

• Seed testing and inbox placement tooling. Imperfect, and worth running consistently anyway, because the value is in the trend rather than any single reading.

• Feedback loops. So complaints come back to you as data you can act on rather than disappearing into a provider's decision about you.
  

Apple gives you nothing at all, so a portion of your audience is invisible by design and you should stop expecting a complete picture. Knowing where the blind spot sits is more useful than pretending it is not there.

I would recommend doing the Email Deliverability Certification ASAP.

The Ramp

Build up gradually, starting long before you think you need to

Peak volume for most senders is somewhere between three and five times their ordinary output, arriving inside a fortnight, aimed at a larger list than usual. From a mailbox provider's point of view that pattern is indistinguishable from the behaviour of somebody who has just bought a list, and providers do not grade on intent.

Warming is not only for new domains and new IPs. Established senders need a ramp too, because reputation is built on consistency and a sudden step change is the opposite of consistent. The way to think about it is that you are teaching a provider that this new volume is normal for you, and teaching takes repetition over weeks.

How to run the ramp

• Increase in steps, not jumps. Move your volume up in stages across September and October rather than arriving at peak volume on the first day of the campaign.

• Watch complaints and bounces at every step. If either moves in the wrong direction, hold at that level until it settles rather than pushing through and hoping.

• Ramp with your best audience first. Extra volume should go to your most engaged people before it goes to anybody else, because they generate the positive signals that make the next step safe.

• Give any new sending domain months, not weeks. If you are separating streams onto a new subdomain, and you should be, it needs its own warming period. Standing one up in November is sending your most important mail from a stranger.

The mistake worth naming is the one where a team does everything right through September, then abandons the plan on the first slow sales day and triples volume overnight in a panic. Write the ramp down, agree it with whoever will panic, and make it something you have to consciously break rather than something you can quietly drift out of.

The big one

Do not go wide

Of everything in this post, the instruction people find hardest to follow is the simplest one. Peak makes businesses email everybody, and emailing everybody is the single most expensive thing you can do to a sender reputation.

The logic in the room always sounds reasonable. It is Black Friday, everyone likes a discount, we have four hundred thousand addresses and we are only using ninety thousand of them, so surely this is the week to use the rest. Somebody senior says it, nobody has a number to argue back with, and the send goes wide.

What follows is predictable. The dormant portion of the list receives mail for the first time in a year, a meaningful number of them mark it as spam because they have forgotten who you are, and your complaint rate spikes at the exact moment your sending volume is at its maximum and your reputation is under its heaviest load. Placement then degrades across the whole programme, which means the engaged people who were going to buy from you stop seeing your emails somewhere around Saturday morning.

You did not gain the dormant group. You lost part of the engaged one, during the four days of the year when losing it costs the most.

If you are going to widen, widen early and slowly

There is a version of reaching further that works, and it involves doing it in September when a mistake is cheap rather than in November when it is not. Take one tier out from your engaged core, send to it, watch what happens to complaints and placement for a fortnight, and only then decide whether the next tier is safe. By the time peak arrives you will know exactly how far you can reach, because you will have tested it under conditions where being wrong did not matter.

Safeguarding

The protections that need to exist before the season starts

Peak is not the time to be making judgement calls, because judgement gets worse when everyone is tired and the numbers are being watched hourly. Everything below should be a rule that runs without anybody having to remember it.

• Separate your transactional stream completely. Order confirmations, shipping notices and password resets must not share a sending reputation with your promotional campaigns. If a hard week of marketing takes down your order confirmations, you have turned a deliverability problem into a customer service crisis in the middle of your busiest trading period.

• Suppress the dormant group in September. Not in November, because suppression needs time to show up in your reputation and doing it late gives you the loss without the benefit.

• Build and test your exclusions before the calendar is loaded. Recent purchasers out of the same offer, anyone in a live service conversation, anyone who has just bought the thing you are about to discount. Test them by running a send and checking who was excluded rather than trusting the interface.

• Cap frequency across campaigns and automations together. Your flows do not pause because your campaign calendar got busy, and the person receiving both is the one who complains.

• Agree on a kill switch and who holds it. One named person, a documented way to stop a send inside minutes, and a rule about what triggers it. Deciding that at eight on a Friday evening is how bad sends finish sending.

• Move monitoring to daily for the whole season. Complaint rate, bounce rate and placement checked every day, not reviewed the following Tuesday when the damage has already compounded.

• Freeze all infrastructure changes from four weeks out. No new ESP, no new sending domain, no template system migration, no clever new tool. Change introduces risk and you have no time to absorb it.

Stop doing

Things that used to be survivable and are not any more

The rules moved when Gmail and Yahoo introduced enforced sender requirements, and a lot of peak habits date from before that. Several of them were merely risky and are now actively dangerous.

• Emailing the whole database because it is Black Friday. Covered above, and worth repeating because it will be suggested to you at least twice.

• Waking up lapsed data just for peak. A reactivation campaign in November is a complaint generator wearing a Santa hat. If you want to attempt one, it belongs in September with a small cohort and a lot of monitoring.

• Buying, renting or borrowing lists for the season. Faster route to enforcement than almost anything else you could do, and it puts your entire domain at risk rather than one campaign.

• Standing up a new subdomain in the run-up. No history, no reputation, maximum volume. All three at once is a choice, and not a good one.

• Reporting delivered as though it means inboxed. It has never meant that, and at peak the gap between the two widens because everyone is sending more.

• Using open rate to define your engaged segment. Privacy protections and pre-fetching mean a chunk of those opens never involved a human. Build the segment on clicks and on meaningful actions outside the email.

• Hiding or softening the unsubscribe to protect list size. The person who cannot find the unsubscribe link uses the spam button instead, and one of those two outcomes damages you.

• Judging this year against last year without adjusting for what changed. Open rates, tab placement and measurement have all moved. Comparing raw numbers across years produces confident conclusions from broken arithmetic.
  

The hangover

What peak does to your programme in January

The part almost nobody writes, because it happens after everyone has stopped paying attention.

Peak floods your list with consequential opt-ins. Thousands of people who handed over an address to get money off something, who did not sign up for a relationship and were never asked whether they wanted one. In December that looks like the best list growth of the year and gets reported as a triumph.

Then Q1 arrives, those people ignore everything, your engagement rates fall off a cliff, and somebody asks why email has gone quiet. The answer is that you did not grow your list in November, you diluted it, and the bill lands eight weeks later when nobody connects the two events.

What to do about it, decided now rather than in December

• Give peak buyers their own orientation flow. They arrived through a discount, so treat them as somebody who has not yet agreed to hear from you regularly, and earn the second opt-in rather than assuming it.

• Do not merge them into the main programme automatically. Hold them separately until they have shown you something, and let behaviour rather than a signup date decide when they join everyone else.

• Tag the cohort so you can see it later. In February you will want to know how the November intake behaved compared with everybody else, and you cannot answer that retrospectively if you did not label them at the time.

• Report engaged list size in January, not total list size. Otherwise the growth chart tells a story your revenue does not support, and you will spend the quarter defending it.
  

The countdown

Working backwards from the weekend

Rough, and worth adjusting to your own sending frequency, but the sequence matters more than the exact dates.

1. Sixteen weeks out, which is now. Run the audit and take your reputation baseline. Nothing else starts until you know what you are working with.

2. Twelve to fourteen weeks out. Fix authentication, get DMARC to enforcement, separate transactional from promotional, and define what engaged means for your audience.

3. Ten to twelve weeks out. Suppression live, exclusions built and tested, dormant group dealt with, and any new subdomain started on its warming schedule.

4. Eight to ten weeks out. Begin the volume ramp with your most engaged people, and start widening one tier at a time while mistakes are still cheap.

5. Six weeks out. Send at something close to peak volume at least once, so the first time you do it is not on the day it matters.

6. Four weeks out. Freeze infrastructure changes. Whatever is not in place now is next year’s project.

7. Two weeks out. Daily monitoring begins, kill switch confirmed, and everyone knows who decides what.

8. Peak week. Monitor, do not build. The only decisions you should be making are about stopping things.

9. The week after. Peak buyers into their own orientation flow, cohort tagged, and a note written to yourself about what you would do differently.

Where I would leave it

Peak is not won in November. By the time the weekend arrives you are simply collecting on decisions you made in August and September, and the businesses that do well are rarely the ones with the cleverest campaign. They are the ones whose email arrived.

Which is a slightly deflating thing to read in a marketing post, and also the most useful. Everything that determines your peak result is boring, it is infrastructure, and it is available to you right now while there is still time to change the outcome.

Audit your own email deliverability

To give you the BEST chance, take my email deliverability certification now and you'll be in tip top shape by November.