Peak season pricing has already started. UPS’s peak charges for additional handling, large packages and over-limit parcels took effect on September 27, and its residential demand surcharges start on October 25, rising to $0.75 per Ground Residential and Ground Saver package between November 22 and December 26, according to its published surcharge table. FedEx’s residential surcharges begin on October 26 and peak at $0.80 per package, Supply Chain Dive reported, up from $0.65 last year. USPS has asked regulators to approve a temporary holiday increase from October 4 that averages about 6%. And on September 22, FedEx announced a 5.9% average general rate increase for January 4, 2027, with several services and surcharges rising faster than the headline.
Most brands respond to this the same way every year: they renegotiate, rate shop, or absorb it. Those are carrier-level levers, and they matter. But a large share of shipping cost is locked in earlier, at the moment an order is routed. Which location fills the order, and whether the order leaves as one box or three, is an order operations decision. By the time a shipping system picks a carrier, most of the cost is already set.
Every package carries its own charges
Carrier surcharges are priced per package. That sounds obvious until you look at what it means for a multi-item order. If an order ships as one box to a residence in early December, it picks up one base rate, one residential charge and one demand surcharge. If the same order ships as two boxes from two warehouses, it picks up two of each.
Using UPS’s own published numbers, an order split into two Ground Residential packages during the November 22 to December 26 window pays $1.50 in demand surcharges, double the $0.75 a single box would pay, before counting the second base rate and the second residential charge. Multiply that by every split order in the busiest five weeks of the year and the pattern becomes visible on the P&L.
Split shipments also cost more than postage. Each extra package means another pick, another box, more dunnage and more labor at the dock. For the customer, it means two tracking numbers, two deliveries and a higher chance that one of them arrives late. None of these costs shows up on the carrier invoice as a single line item, which is part of why they persist.
Where split shipments come from
Split shipments are common, and they grew as fulfillment networks spread out. Narvar’s data showed the share of orders shipped in more than one package rising from 16.7% in 2019 to 22% in 2020, holding at 21.1% in 2021, as brands added fulfillment locations, ship-from-store and dropship during the pandemic.
In most operations, splits come from a handful of causes. Inventory is spread across several warehouses, 3PLs or stores, and no single location holds every item on the order. The closest location is out of one SKU, so the router sends that line somewhere else. Some items ship from a vendor or supplier directly, so they were never going to travel with the rest of the order. Oversize or hazardous items need their own packaging. And in many systems, routing logic evaluates each line on its own, choosing the “best” location per item, which produces a perfectly optimized set of lines and an expensive order.
The last cause is the one most brands can fix fastest, because it lives in their own rules.
The real tradeoff: distance, boxes and the promise
Some splits are worth paying for. Three forces pull against each other on every multi-item order.
Distance drives the base rate. Carrier pricing climbs with zone, so shipping the whole order from a far warehouse can cost more than shipping two boxes from two nearby ones. Box count drives per-package charges, including the peak surcharges above. And the delivery promise constrains both: a consolidated order that misses the date the customer was shown costs more in the long run than a split that arrives on time.
Package size adds a fourth variable. Carriers bill the greater of actual weight and dimensional weight, which UPS calculates by dividing length times width times height by 139 for daily rates. Consolidating several items into one large box can push it over dimensional or handling thresholds. During peak, UPS’s additional handling charge runs $8.75 to $11.90 per package and FedEx’s runs $8.80 to $11.85, according to the sources above. A consolidated box that trips one of those charges can cost more than the two smaller boxes it replaced.
So the right answer is order-specific. A routing decision that only looks at inventory availability, or only at distance, will get it wrong in predictable ways. The routing layer needs to see the whole order and weigh the options against rules the business sets.
A worked example
Picture a brand with an East Coast warehouse, a West Coast 3PL and a handful of retail stores that ship online orders. In early December, a customer in Denver orders a jacket and a pair of boots. The East Coast warehouse holds both. The West Coast 3PL holds the jacket but is out of the boots. A store two hours from Denver has the boots.
A router that picks the closest location for each line sends the jacket from the West Coast 3PL and the boots from the store. That is two packages, two sets of peak surcharges, two picks, two boxes, and two tracking emails for the customer.
A router that evaluates the whole order first sees that the East Coast warehouse can fill both items. It then checks whether that single shipment, from farther away, still meets the delivery date the customer saw at checkout. If it does, one package goes out, at a higher zone but with one set of per-package charges. If it does not, the router falls back to the split, because missing the promise in the second week of December costs more than a second box.
Neither answer is right every time. The point is that the decision is made on purpose, for the whole order, against the promise, with the cost of each option visible. That is what brands mean when they say routing is a margin lever.
Routing rules that cut splits during peak
The brands that manage this well tend to use a similar set of rules, adjusted for their catalog and network.
First, prefer a single location that can fill the whole order within the promise. Before splitting anything, check whether one warehouse, 3PL or store holds every line and can still hit the delivery date the customer saw. Only if none can does the router consider splitting.
Second, set explicit split rules. Decide when a split is allowed: when the promise would otherwise be missed, when the order value justifies the extra cost, or when a line is oversize and ships alone regardless. Write those rules down and put them in the router, so the decision is consistent at 2 a.m. on Cyber Monday.
Third, decide what happens to a missing line. When one item is out of stock everywhere nearby, the choice is to ship what is available now and send the rest later, hold the order until everything can ship together, or source the missing line from a farther location. Each has a cost. Make the choice deliberately, by category or order value, and set it in the system.
Fourth, position inventory for how customers actually buy. If two SKUs are frequently ordered together, stocking them in the same locations before peak removes splits at the source. Pre-peak inventory rebalancing is one of the few moves that lowers both shipping cost and delivery time.
Fifth, treat known oversize items as their own shipment on purpose. Planning for them avoids forcing them into consolidated boxes that trip handling charges, and it keeps them from blocking the rest of the order.
Sixth, route on accurate inventory. Every rule above depends on the router knowing, per location and in close to real time, what is actually available. When counts lag, the router sends lines to locations that cannot fill them, and the order gets rerouted and split after the fact. For the mechanics, see What Is Real-Time Inventory Sync.
When a split is the right call, say so
Some orders will still split, and some should. When they do, the customer experience depends on communication. Narvar notes that poor communication can erase the benefit of a faster partial delivery, and that unexpected multiple boxes read as confusing. The fix is operational: the order record should know it is split from the moment it is routed, so the confirmation email, the order status page and the tracking updates all say “arriving in two packages” before the first box shows up. That only works if the routing decision and the customer-facing status come from the same system of record.
Measure it weekly during peak
Split shipments are easy to ignore because the cost is spread across thousands of orders. A few metrics make it visible. Track split rate, meaning the share of multi-item orders that ship in more than one package. Track packages per order, since that is the unit the carriers bill. Track surcharge spend per order, by carrier and by week, from October through mid-January. And break all three down by fulfillment location and sales channel, because splits usually cluster around one node that is chronically short of a few SKUs.
Reviewed weekly, those numbers tell an operations team where to move inventory and which routing rules to tighten before the most expensive weeks arrive.
Where Pipe17 fits
Pipe17 is the Order Operations Platform. It connects sales channels, fulfillment and back-office systems, and orchestrates every order from checkout to delivery, with routing rules the business team sets and changes without engineering. Orders from every channel arrive in one place, Pipe17 sees per-location inventory across warehouses, 3PLs and stores, and it routes each order to the locations that should fill it. That routing decision sets the minimum number of packages an order becomes, before a shipping system or 3PL chooses a carrier and prints a label.
Carrier rate shopping still has its place, and it stays with the shipping software or the 3PL. What changes is the order of decisions. Decide where the order ships from and how many pieces it leaves in first, and every downstream carrier decision starts from a cheaper baseline.
The takeaway
Peak surcharges, the 2027 rate increases and dimensional weight rules all charge by the package. That makes packages per order one of the most controllable shipping costs a brand has, and the lever that controls it is order routing. Brands that fill orders from one location whenever the promise allows, split only on purpose, and route on accurate inventory will pay fewer surcharges this peak and carry that discipline into 2027.
