Black Friday is on 27 November and Cyber Monday on 30 November, which means you have about seven weeks.
Which is enough time to protect yourself and not enough time to fix everything, so this post is organised around that constraint. What is already too late, what you can still do, what to send, and the list of things that will quietly cost you January.
I have written the long version of peak prep before, aimed at people reading it in August. Take this as the October version, written for where you are right now.
Warming a brand new domain or subdomain. A meaningful warm-up runs four to eight weeks minimum and longer for big programmes. Starting one now means your new domain hits peak with almost no history, which is the worst possible combination. Park it until January.
Migrating ESP. If it is not already done, it is a January project. Doing it in November means a reputation reset, unfamiliar tooling and no fallback during your highest-stakes period.
Moving DMARC to enforcement for the first time. Worth doing, not worth doing in the next three weeks without clean reporting behind it. If your record is still at p=none, monitor now and tighten in the new year.
Rebuilding your template. The redesign will not make or break peak and it will consume the capacity you need for the things that will.
Building a new acquisition engine. People joining your list in the next three weeks are the ones most likely to complain during peak. Acquire if you can, and do not count on them.
None of that is a counsel of despair. It is the list that frees up the next three weeks for work that will land.
Find out where you are landing, per provider. Before you change anything. If you are already struggling at one provider, everything below gets more urgent and your volume plan needs rethinking.
Verify the list you intend to mail. And verify it again in early November if you are mailing anything that has been sitting idle. Bounces are the fastest way to tell providers you do not know who is on your list, and peak is when that costs most.
Decide who is NOT getting your peak campaigns. The single highest-value hour of work available to you between now and November. More on this below.
Ramp your volume gradually, starting now. If you normally send 300,000 a month and plan 1.5 million across peak week, you cannot make that jump cold. Step it up through October so the shape of your sending changes gradually rather than overnight.
Warm the engaged end of your list first. Any volume increase should go to your most engaged people before it goes anywhere else, so the extra sending generates positive signals rather than silence.
Check your authentication, then leave it alone. Confirm SPF, DKIM and DMARC are passing and aligned. Fix anything broken in the next fortnight, then freeze.
Set up your monitoring before you need it. Postmaster Tools, SNDS, Yahoo Sender Hub, and a baseline reading written down with today’s date against it so you can tell in November whether a number is unusual.
Write the thresholds that stop a send. Agreed now, in writing, with a named person who can call it. Deciding in the moment produces panic sending.
Pre-approve your contingency comms. Site down, stock gone, code not working, delivery promise missed. Templates signed off in October are worth a fortune at 9pm on the 27th.
The instinct at peak is to email everybody, because it is the one week of the year where the maths appears to justify it. It is also the week where your complaint rate is rising anyway, every other sender is in the inbox with you, and providers are watching volume patterns closely.
A large dormant segment included in a peak blast drags down placement for the entire send, which means the people who were going to buy do not see it either.
Your dormant group. Nothing at all, inside or outside the email, over a long period. They carry your risk and they will not rescue your numbers.
Anyone acquired in the last few weeks through a weak mechanism. Discount pop-ups and competition entries especially. No relationship, high complaint propensity, worst possible timing.
People who bought the thing you are promoting. Obvious, routinely missed, and irritating in a way that generates complaints rather than unsubscribes.
Anyone in a live service or delivery issue. A promotional email to somebody whose order is late converts a fixable problem into a public one.
Segments where you have historically high complaint rates. You already know which ones. Peak is not the moment to test whether they have improved.
The single best thing you can send in the next three weeks, and almost nobody does it.
A short email, sent in early November, saying some version of: the next few weeks will be busier than usual and mostly about our sale. If that is not for you, press this and we will leave you out of it, and you will still hear from us normally in December.
It converts complaints into preferences. The people who would have hit the spam button during peak now press a button that costs you nothing. You trade a reputational event for a suppression flag.
It improves the segment you keep. Everybody left in your peak audience has been given a chance to leave and chose not to, which is a meaningfully more engaged group to be sending to.
It signals confidence. A sender who makes it easy to opt out of a sale reads as a business that respects the inbox, and that lands harder than any amount of brand copy.
And it retains people you would otherwise lose. Somebody who finds peak overwhelming can stay subscribed for the eleven months where you suit them, instead of unsubscribing in November and never coming back.
Send it in early November, not late. Before the volume starts, so the offer makes sense.
Make it one click. A link that suppresses them from the peak campaign tag. Not a preference centre with eight options.
Be specific about the dates. Out of sale emails from the 20th to the 1st, back to normal after that. People say yes to a defined period far more readily than to a vague one.
Honour it exactly. If one promotional email slips through to that group you have done more damage than if you had never offered.
And bring them back deliberately. A short note in December confirming normal service has resumed.
Every inbox in November is a wall of identical discount emails, which means the differentiator is almost never the size of the offer.
Early access for your most engaged. Rewards the people carrying your deliverability, generates a strong engagement cluster before your big sends, and warms your reputation at precisely the right moment.
The anti-Black-Friday email. We are not discounting, and here is why. Works brilliantly for brands with a quality or sustainability position, and it is the most forwarded email of the season every year.
A buying guide rather than an offer. Help people decide what they need before you tell them what is cheap. Useful, gets opened by people not ready to buy, and positions you ahead of the decision.
What is NOT in the sale. Counterintuitive and it works, because it tells people what to buy now at full price and it is a properly useful piece of information.
Service and reassurance content. Delivery cut-offs, returns windows, what happens if it is wrong. Removes uncertainty at the moment people are spending money with businesses they do not know well.
One email that sells nothing at all. In the middle of peak week. It will be the best-performing thing you send and it buys you goodwill for the three asks either side of it.
A proper thank you afterwards. On the 1st or 2nd, to people who bought, with no ask attached. The end of an experience is weighted heavily in memory, and most brands end peak by asking for one more thing.
Do not email your whole database because it is Black Friday. Covered above, and it is the one I would most want you to take away.
Do not reactivate dormant contacts for peak. The reactivation campaign that seems sensible in October is a complaint event in November.
Do not buy, rent or append data. Ever, and especially not six weeks before the highest-scrutiny period of the year.
Do not increase frequency without increasing value. Volume without consequence is what trains people to stop seeing you.
Do not manufacture urgency you cannot back. A deadline that extends twice teaches your audience that your deadlines are decorative, and that lesson lasts well past December.
Do not bury the unsubscribe to protect your numbers. You will trade unsubscribes for complaints, which is the worst trade available in email.
Do not make infrastructure changes in November. The freeze exists for a reason.
Do not let flows run unchecked underneath your campaigns. Your automations do not know it is Black Friday. Somebody is receiving a browse abandonment, a replenishment reminder, a newsletter and three promotions in the same 48 hours.
Do not stop sending on 1 December. Going quiet after a volume spike is its own signal, and the January hangover is made worse by it.
Complaint rate per send, per provider, daily. Your early warning signal. Gmail's ceiling is 0.3% and you never want to be approaching it, particularly when volume is high.
Bounce rate on anything you have not mailed recently. A spike means the list was staler than you thought.
Placement by provider, not blended. Strong Gmail numbers will happily conceal a collapse at Outlook or Yahoo inside a healthy-looking aggregate.
Deferrals and throttling. Repeated deferrals from one provider mean you are being slowed down, which is a reputation signal rather than a technical one.
Unsubscribe rate against the same email type last year. Not against your newsletter, which is a different animal.
Revenue per thousand sent, by segment. Which tells you whether the wide sends earned their place or just added volume.
Complaint and unsubscribe cost by segment. Put the two numbers next to each other and the argument about going wide next year resolves itself.
Engagement in December and January, by cohort. The real cost of peak shows up six weeks later, and nobody looks.
How the opt-out group behaved afterwards. If they stayed and re-engaged in December, you have the evidence to make the opt-out permanent.
Placement before, during and after. Three readings, same method, so next October you know exactly what peak costs you and can plan around a number rather than a feeling.
Peak is not won in peak week. It is won by the exclusions you set in October, the ramp you start now, and the discipline not to email everybody because the date says you should.
Seven weeks is enough to protect what you have. Verify the list, decide who is not getting it, ramp gradually, offer people a way out, send a few things that are not asks, and freeze the infrastructure in November.
Then measure it properly in January, so that next year you are making decisions from your own data rather than from whatever everybody else is doing.
How to Prepare Your Email Deliverability for Black Friday and Peak Season
Email Behaving Badly: Why Compliance Will Not Save Your Deliverability