About this case. It describes a common situation and a sensible approach, written without naming any customer. It reports no measured results, so set your own baseline and targets and measure your own outcomes.
The situation
Picture a home-and-lifestyle online store selling through its own website. Most of the year it sends a modest amount of email: order confirmations, shipping updates, a weekly newsletter. Then the festive season arrives. For about six weeks, order volume multiplies, the marketing team schedules daily promotions, and the list that mostly sat quiet is suddenly asked to read a lot of mail.
Last year went badly. The team sent its biggest promotion of the season to the whole list on a single morning. The send was ten times larger than anything the sending domain had done in months. Several mailbox providers slowed delivery, order confirmations got stuck behind promotional mail in the same queue, and customers who had just paid wrote in asking whether the order had gone through. Some promotional mail went to the spam folder, and the open rate on the hero campaign was a fraction of normal. The team now wants a plan that makes the next peak boring.
Diagnosing last year
A review of the previous season found four root causes:
- A volume cliff. Mailbox providers learn a sender's normal pattern. A tenfold jump on a single morning looks like a compromised account or a list purchase, and filters respond cautiously.
- Cold segments. The promotion went to everyone, including subscribers who had not opened or clicked for a year. Disengaged recipients are the most likely to mark mail as spam or to hit abandoned addresses.
- Shared lane. Order confirmations and promotions travelled through the same domain and queue. When promotional mail was throttled, transactional mail waited behind it.
- No early warning. The team learned about the problem from customers rather than from dashboards.
The plan: eight weeks before the peak
Week 8 to 6: separate and authenticate
The first change was structural. Order and shipping mail moved to a dedicated subdomain (for example orders.example.com) and promotions to another (offers.example.com), each with its own SPF, DKIM and DMARC records. The team checked that the From domain aligned with the signing domain on both. Our guide to mailbox provider requirements lists what Gmail and Microsoft expect from bulk senders, and DKIM setup covers the signing step.
The team also added a one-click unsubscribe header to promotional mail and tested that the endpoint responded correctly, because the large providers now expect it for marketing mail.
Week 6 to 3: build volume gradually
Rather than waiting for the peak, The store raised promotional volume in steps, sending first to its most engaged subscribers. The ramp below is an example of the pattern, not a fixed schedule; the right pace depends on your list and provider mix, and you should watch results at each step before moving on.
| Step | Who receives it | What you check before the next step |
|---|---|---|
| 1 | Subscribers who opened or clicked or bought in the last 30 days | Delivery, bounces, complaints by provider |
| 2 | Add those active in the last 90 days | Same checks, plus placement in a few test inboxes |
| 3 | Add those active in the last 180 days | Complaint rate and unsubscribes by segment |
| 4 | Peak sends to the full engaged audience | Hourly monitoring on send days |
The warm-up logic is the same as for a new IP address: build a track record on engaged recipients, then widen. Our IP warm-up guide explains the principle in more detail. If the store moves to a new dedicated IP for promotions, this ramp doubles as the warm-up.
Week 4 to 2: clean the list and plan the segments
- Validate addresses that joined through checkout and popups, and remove those that are invalid or role-based.
- Define a sunset segment. Subscribers with no click or purchase for a long period receive a final re-permission message and are then suppressed if they do not respond. Because Apple's Mail Privacy Protection inflates opens, the team uses clicks and purchases, not opens, to judge engagement (we discuss why in the end of the open rate).
- Plan the calendar. Spread promotions across days and time zones instead of stacking them, and keep a reserve of send capacity for transactional mail.
Week 2 to peak: set the monitors
The store connected delivery events to its own dashboard through webhooks and set alerts for each stream: complaint rate, bounce rate, deferral rate by provider, and the time from order to confirmation. It also set up Google Postmaster Tools for both subdomains. On send days, a named person watched the dashboard for the first hour of every campaign, with the authority to pause.
Peak week: how a send day runs
- Pre-flight (the day before). Check that authentication passes on a test message, links work, the unsubscribe header is present and the segment count matches expectations.
- Staged send. Release the campaign in waves, not at once, starting with the most engaged segment. Hold the next wave until the first shows normal delivery and complaint signals.
- Protect transactional mail. Order confirmations travel on their own subdomain and are never rate-limited by the promotion's schedule.
- Decide by data. If deferrals rise at one provider, slow that provider's wave instead of stopping everything.
- Debrief. Record what happened by provider and segment while it is fresh, so next year's plan starts from facts.
Mistakes this plan is designed to avoid
- Treating the festive season as a normal campaign. A send that is ten times your usual volume is a different kind of event and needs a different kind of preparation.
- Mailing the whole list because it is there. Subscribers who stopped engaging a year ago are the likeliest to complain and the least likely to buy.
- Sharing one queue between receipts and offers. If a provider throttles promotional mail, anything behind it waits too.
- Changing too much at once. Moving to a new IP, new domain and new template in the same week makes it impossible to know what caused a problem. Make structural changes early and creative changes late.
- Judging by opens. Machine-generated opens flatter every campaign. Use clicks, purchases and complaint rates to decide what to send next.
How the team would judge the season
We do not quote result numbers here. A real team should judge the season on a small set of measures, set in advance:
| Measure | Why it matters |
|---|---|
| Order confirmation delivery time | Customers who have just paid need immediate reassurance |
| Complaint rate per promotion | Gmail guidance aims for under 0.1% and treats 0.3% as a ceiling |
| Deferral and bounce rate by provider | Early sign of throttling or list problems |
| Clicks and revenue per recipient, by segment | Tells you whether wider segments earn their risk |
| Unsubscribe rate by segment | Shows fatigue before it becomes complaints |
What the team learned
- Peaks are won in the weeks before. The best send-day decision is the one made in week six: building volume gradually.
- Engagement beats size. A smaller send to engaged subscribers often earns more than a larger send that annoys the rest.
- Keep the receipt lane clear. The two messages customers care about most, the confirmation and the shipping update, must never wait behind a promotion.
- Watch, then decide. A person with a dashboard and the authority to pause is cheaper than any recovery.
Adapting this to your store
- Pick your peak date and count back eight weeks.
- Split transactional and promotional mail onto separate authenticated subdomains.
- Build a ramp from your most engaged segment outwards, checking results at each step.
- Define a sunset policy using clicks or purchases rather than opens.
- Stage large sends in waves and keep a pause rule written down.
- Set up delivery webhooks, Postmaster Tools and a named person to watch on send days.
Planning a seasonal peak? Tell us your volumes and dates and we will help you build a ramp.
Related reading
Frequently asked questions
Does this case name a customer?
No. It describes a common situation and approach without naming any customer, and it reports no measured results.
Why is a sudden tenfold increase in volume risky?
Mailbox providers learn a sender's normal pattern, and a sudden jump can look like a compromised account or a purchased list, which makes filters cautious.
How far ahead should I start preparing for a peak?
The scenario uses eight weeks: separate and authenticate in the first weeks, build volume gradually, clean the list and set monitoring before the peak.
Should I use opens to decide who to mail?
Prefer clicks and purchases. Apple Mail Privacy Protection inflates opens, so they are a weak engagement signal.