Before peak, re-read last year's rescued orders.

You have about three weeks of quiet

Here's the calendar. UPS's broader 2026 demand surcharges on Ground Residential, Ground Saver and Air packages start October 25, according to Reveel's summary of the schedule. FedEx's equivalent fees start October 26, per Supply Chain Dive. The highest rates run from November 22 (UPS) and November 23 (FedEx) through late December, and the season doesn't fully end until mid-January. Thanksgiving is November 26 this year, so Black Friday is the 27th and Cyber Monday the 30th.

You can't change the carrier calendar. You can change what breaks inside it.

Most peak checklists are about servers, ad budgets, and inventory. Those matter. But some of the cheapest preparation you can do is sitting in last year's support inbox, and the next three weeks are the last calm stretch to do it.

Pull ten orders someone had to rescue

Go back through last November to January and find orders where a person had to step in. Not every complaint, just the ones where somebody on your team had to do something by hand to make the order come true. Search ticket tags, refund notes, the ops channel, or ask support who they remember.

You're looking for things like:

• An order confirmed for stock that wasn't really there.
• A promotion that stacked with another one in a way nobody intended.
• A split shipment where the customer only heard about half of it.
• An address problem caught by the carrier, not by you.
• A gift message, note, or option that didn't make it to the warehouse.
• A delivery promise the carrier was never going to keep in peak week.

If nothing was tagged, don't worry. Ask the two or three people who handled the most tickets last December to name the orders they still remember. People remember the painful ones surprisingly well, and the painful ones are exactly what you want.

Ten is enough. You're not building statistics. You're looking for the stories that will repeat when volume climbs and everyone is tired.

Follow each one backwards

For each rescued order, start at the rescue and walk back. Who noticed the problem? How? What did they do to fix it? And the important one: what was the first moment any of your systems knew something was wrong?

Very often, the answer is "much earlier than anyone noticed." The stock count disagreed hours before the confirmation went out. The warehouse rejected the gift note in a message that went to an inbox nobody watches. The carrier flagged the address the same day. The information existed. It just didn't reach a person who could act on it before the customer did.

Here's a made-up example, but it has a very common shape, so you'll probably recognize it. A customer emails on December 19 asking where the second half of their order is. Support finds it was split. The second box shipped from a different location on the 17th, with its own tracking number, and the customer got one email that only mentioned the first box. The split was decided at checkout, on the 14th. So the system knew five days before the customer asked, and the only thing missing was a sentence telling them.

That gap, between when the system knew and when a person found out, is what you're trying to close before peak. Our ecommerce page puts it simply: choose a recent order that needed rescuing, follow it backwards, and the most valuable first improvement usually shows itself.

Fix the two or three that will repeat

You won't fix all ten in three weeks, and you shouldn't try. Pick the ones most likely to happen again at peak volume and most expensive when they do. Then choose the smallest fix that would have caught it:

A rule. Don't let that promotion stack. Don't promise next-day delivery to that region in December.

An alert. When the warehouse rejects an order detail, a named person hears about it within the hour, not at the end of the week.

A message. If the order will split, tell the customer when it splits, not when they ask.

A script. If you can't fix it in time, write down for support exactly what to check and what to say. A known problem with a script is far better than a surprise.

While you're in there, check one more thing. If your shipping prices or delivery promises were set before the carriers published their peak schedules, make sure someone has decided whether they still hold after October 25. That's a decision, not a default.

And for whatever you can't fix, decide what you'll do by hand, on purpose. Name who watches for it, at what hours, and what they're allowed to do without asking: reship, refund shipping, upgrade the service. Manual work at peak isn't a failure. Manual work nobody planned for is.

Give support one page for peak

Finish by writing one page for whoever answers customers in November and December: the known issues, what each looks like from the customer's side, what to check, what to say, and who to call when it doesn't fit. Keep it short enough to read on a busy morning. Update it as peak goes on.

An entry can be three lines. "Split shipments: the customer gets one tracking email per box, and the second can arrive days later. Check the order's shipment list before saying anything is missing. If a box hasn't moved in four days, send it to the warehouse lead." That's the kind of thing a new seasonal hire can use on their first shift, and it's the kind of thing your experienced people are tired of explaining in chat.

If the rescues you found point to systems that don't talk to each other, between the storefront, inventory, warehouse and support, that's the work we do in ecommerce and fulfillment operations. Bring us one rescued order, and we'll follow it back with you.

Related work: Paddle8.

Write the review into a one-page brief your ops lead can keep: the ten orders, the three patterns, and the one change you will make before fees widen. That brief is the point of the exercise, not a longer deck.