How to Protect Margin, Delivery Performance, and Returns Capacity
Most peak season planning starts with a volume forecast: how many orders, how many markets, how much capacity tao book. That's necessary, but it isn't what determines how peak actually goes and what happens to the assumptions sitting underneath once real volume starts moving through.
Many cross-border programs, even mature ones, carry assumptions that hold at normal volume but haven’t been tested at peak volume: A duty calculation that was right in July. A backup carrier that's never actually carried a live shipment. A returns process built for a steady trickle rather than a real surge.
Normal volume leaves enough slack to absorb a small error before anyone notices it. Peak magnifies that slack. Most peak failures aren't new problems. They're existing weaknesses exposed at a volume and speed that make them harder to correct: small gaps multiplied across far more parcels, with less time to catch them and less room to reroute around them once they’re caught.
The real question worth asking before Q4 volume moves isn’t "Are we ready for more volume?" It’s "Which of our assumptions have been tested only at normal volume?" Three assumptions, involving peak costs, backup routing, and returns programs, are worth pressure-testing specifically, because they're expensive discoveries once parcels are already in motion.
Pressure-test the economics before promotional pricing is locked
By the time peak volume ships, your pricing is likely set. Promotional plans, ad budgets, and checkout pricing for Q4 typically get locked well before the season starts, based on landed cost assumptions from earlier in the year. That gap between when the number gets set and when it gets tested is where margin exposure lives.
For instance, UPS has set its 2026 peak surcharges: Additional handling and large-package fees run $8.75–$11.90 and $96.25–$117.50 per package respectively outside the core peak window, rising to $11.90 and $117.50 from November 22 through December 26, with demand surcharges on standard ground and air services layered on top of that. Other major carriers have also raised 2026 peak season fees. And keep in mind that in August 2025, the United States ended duty-free treatment for shipments valued at $800 or under, moving most of that volume into formal customs entry requirements. International postal shipments follow a separate, still-evolving duty-collection mechanism rather than the same rule as commercial parcels, so the applicable treatment and costs depend on how a given shipment actually moves.
A landed cost model that hasn't been updated since these changes were announced is running on outdated assumptions. And at peak volume, that cost gap gets multiplied across many more parcels in a much shorter window than it would at any other time of year.
Delivered Duty Paid (DDP) shipping, where duties and taxes are calculated and collected at checkout instead of billed to the customer at the door, is what makes this visible in time to act on. DDP makes duty exposure knowable before delivery, not after. Accurate modeling turns that visibility into protected margin, since those costs get priced in up front instead of surfacing downstream as an unexpected charge at the customer's door. At ePost Global, we model duty and surcharge exposure by market as the inputs change, so a brand's peak pricing reflects what a shipment will actually cost to move, not what it cost last quarter. Our guide to DDP and landed cost goes deeper on the mechanics behind this.
Are your promotional and peak pricing assumptions built on landed cost data confirmed after this year's duty and surcharge changes, or are they from last year’s planning cycle?
Pressure-test the routes before you need the backup
International shipping programs usually have some type of a backup plan: a second carrier relationship, an alternate route on file, a note in a runbook. Peak season asks “Has that backup ever moved real volume, or does it exist only on paper?”
Coverage is whether a viable alternate route exists for a given market at all. Readiness is whether that alternate route has actually carried traffic for you, has current rates, and can absorb volume without a renegotiation or a new integration. A brand can have coverage without readiness, a carrier relationship that's technically active but has never actually routed a shipment. That gap stays invisible until the day the primary route fails and someone finds out, in real time, whether the backup actually works.
The same problem shows up in delivery timelines. A carrier's performance in a specific market can soften for weeks before it shows up on a service level agreement (SLA) report, and at peak, there's little runway left in the season to notice and correct it.
Even a successful reroute isn't automatically equivalent to the route it replaced. If the alternate carrier's system doesn't hand off cleanly to the one your customer has been watching, a parcel can keep moving physically while the tracking handoff and the customer-facing status information quietly degrade. Parcel movement and visibility aren't the same thing, and a program that monitors only the first one can miss a real "where is my order?" (WISMO) problem in the second.
The Canada Post strike shows the difference in practice. It ran 32 days, from November 15 through December 17, 2024, directly through peak holiday shipping. Brands whose Canadian volume ran through Canada Post exclusively had no options available. ePost's network, on the other hand, rerouted 47,000 affected shipments within 48 hours with zero SLA failures. That was possible because the routing capacity was already active before the strike started, not assembled in response to it. That's the gap between a contract and a tested route, in practice.
Do you know which of your top peak-season markets would lose real backup coverage under peak-level constraints?
For the markets that do have backup coverage, has that alternate route carried traffic for you recently, or does it exist only on paper?
Treat post-peak reverse volume as part of the peak plan
Most peak plans end at delivery. Operationally, they shouldn't. The same sales surge that creates forward volume in November and December creates a return volume that follows it. A forward volume forecast creates a later reverse volume requirement, and most peak plans treat the two as separate problems on separate timelines rather than as one capacity-planning question.
International returns already carry more complexity than domestic ones: return freight that can rival the original outbound cost, customs clearance in reverse, potential re-import duties, and longer transit times that tie up inventory for weeks instead of days. Refusals are an additional issue, though they're more likely when duties are collected at the door instead of at checkout, the scenario a DDP program is built to prevent. None of this is new information. What's less examined is whether the capacity to process this complexity, at the volume that peak season generates, has been planned for the same way forward capacity has.
How much reverse volume a given peak season creates depends heavily on category and return profile. The reverse volume requirement is a direct corollary of the forward forecast, and treating it as a separate, later problem is a planning gap worth closing before the season starts.
At ePost Global, our partnership with ShipWise gives brands a prebuilt cross-border returns path into Canada and the EU, so reverse capacity doesn't have to be assembled after post-peak volume has already started arriving. Our guide to returns and refusals in cross-border eCommerce covers the broader framework this section builds on.
Has expected post-peak return volume been modeled against your current reverse logistics capacity and staffing, the same way forward peak volume was modeled, or does that modeling not exist yet?
Five questions that tell you whether the plan is actually ready
These are answerable today, before peak volume moves. If you can’t satisfactorily answer any, you have a gap worth closing now.
- Are your promotional and peak pricing assumptions built on landed cost data confirmed after this year's duty and surcharge changes, or carried forward from last cycle?
- Do you know which of your top peak-season markets would lose real backup coverage under peak-level constraints?
- For the markets that do have backup coverage, has that alternate route carried traffic recently, or does it only exist on paper?
- Is your tracking and customer-communication path tested across your alternate routes, not just your primary one?
- Has expected post-peak return volume been modeled against your current reverse logistics capacity and staffing?
The decision before volume moves
A gap revealed by any of the five questions above is a planning conversation you need to have right now. The same gap found in November or after the season ends is a customer, margin, and service problem at a moment when there's very little room left to fix it.
If you're not sure where your program stands, the ePG Market Readiness Assessment is a place to start. And once you know where your own gaps are, 8 Questions to Ask Your International Carrier Before Peak Season is the next conversation, the one to have with whoever is actually moving your peak volume.
FAQ - Peak Readiness
What does "peak readiness" mean if our SLAs held last year? An SLA measures whether a shipment arrived in the promised window. It doesn't measure whether the backup route that got it there had been tested before it was needed, whether customers were satisfied with the service, or whether the pricing behind it reflected current duty and surcharge costs. A program can post clean numbers and still be carrying assumptions that haven't been checked at peak volume.
How is this different from checking carrier capacity and cutoff dates? Capacity and cutoffs are forecasting questions, and they matter. This guide is about a different question: Of the assumptions built into your pricing, routing, and returns program, which ones have actually been validated under real conditions, and which are still just contracted or assumed?
