Last-mile in cross-border B2B: the part everyone underestimates

At Mahaya we keep seeing the same pattern across mid-market exporter clients: the operational story they tell themselves ends at the destination port. The vessel arrives, the container clears, and the rest is "delivery." In their pricing, in their lead-time promises, in their margin model, the segment between the destination terminal and the receiving dock at a B2B customer is treated as a single number, sometimes a single line item, sometimes a footnote. And then it eats the margin.

This note is the version of the conversation we have when an exporter comes to us asking why a profitable lane on paper has turned into a break-even one in practice. Nine times in ten, the answer is in the last 80–400 kilometers, the segment that the planning sheet treated as a constant.

Why B2B last-mile is structurally different

Consumer last-mile is the case most people have in their head: a parcel courier, a doorstep, a tracking number, a 24–72 hour window. The economics are crowded but legible. B2B last-mile in a cross-border context is none of that.

The receiver is a warehouse with appointment windows of 30–120 minutes, often booked 48–96 hours in advance, with penalties for late arrival that can run $150–$450 per missed dock slot. The vehicle is a 40-foot trailer, not a van. The driver may need a specific license type for the destination region. The cargo is palletized, not parcelized, and damage on the last segment is more expensive because the units are larger. The receiving dock has lift constraints, height constraints, and a forklift driver whose shift ends at a specific time. Every one of these creates a hard edge that the planning sheet does not see.

The unit economics also flip. On the ocean leg, the cost per unit weight falls roughly proportionately with shipment size; doubling the volume halves the per-unit ocean cost almost linearly. On the last mile, the cost per unit is dominated by time-on-stop and asset turn, not weight. A two-pallet delivery and a six-pallet delivery to the same customer can have within 18% the same final cost, because the truck and the driver were going there either way. The exporter who hasn't internalized this asymmetry mis-prices the small-order tail of their book.

Where the variability really lives

The lazy answer is "traffic." Traffic is the smallest part of the variability. The real variability lives in five places:

• Customs and last-mile handoff: a delay in the customs broker's release to the trucker can push a delivery by a full business day, because the truck slot is gone and the next one is tomorrow. This is the failure mode we've watched create the most surprise on a final invoice.
• Chassis or trailer availability at the destination port. In the worst weeks of 2024 in the U.S. inland network, this single factor was responsible for 5–9 day delays on a meaningful share of containers. The condition has improved but not gone away.
• Receiver appointment-window mis-syncs. The trucking partner books a window. The vessel arrives 14 hours late. The window is now wrong, the next available window is 36 hours out, and the cargo sits.
• Driver/equipment regulation. Cabotage rules in Europe, hours-of-service caps in the U.S., specific permit requirements in parts of Latin America and the Gulf. These don't change often, but when they're missed they're missed expensively.
• The receiver themselves. The customer who doesn't open Saturdays, the customer with a forklift outage, the customer whose intake clerk is on vacation. None of these are the exporter's fault. All of them are the exporter's cost.

The pattern this connects to is the one we sketched in our field note on logistics planning: the inland leg is the failure mode most planning sheets handle as a constant when it is structurally the most variable segment of the journey.

What "good" looks like operationally

The exporters whose last-mile costs are reasonably predictable share a few practices. The list is not exotic, but the practices are not common:

They price the last mile as a stochastic line, not a deterministic one. Each lane has a base last-mile rate and an expected variance, both tracked. The variance number sets the inventory buffer and the price quoted to the customer. This connects directly to the question we cover in our note on inventory positioning — how much buffer is the right buffer at which node.

They have a named trucking primary and a named secondary on every destination region they ship to regularly. Not three forwarders racing each other to the bottom on quote; two carriers with whom they have an actual relationship, who will take a call when the appointment window slips. The price is 3–8% higher than the spot market on average. The reliability is meaningfully better, and the dispute resolution at quarter-end is on a different planet.

They monitor the upstream signals that will hit them downstream. If the destination port congests, the chassis pool tightens, and the last-mile delivery rate gets worse within 5–10 days. We pick up the upstream monitoring question in our port congestion brief; the operational point here is that the last-mile team should be reading those signals, not just the destination ops team.

And they instrument the segment. A last-mile lane with no event timestamps — no tendering, no PU/PD, no POD — is a lane where root-cause analysis cannot happen. We've seen clients run an entire fiscal year unable to answer the question "which of our customers cost us the most in last-mile variance" because the data wasn't there.

The framing question for the buyer

The frame we suggest to mid-market exporters running 200–2,000 containers a year: take the worst destination region you ship to, by margin erosion. Pull the last six months of invoices on that lane. Separate the line items into ocean, port-related, and last-mile. Then look at the last-mile line specifically and compute what percentage of total delivered cost it represents and what its coefficient of variation is. We almost always find the percentage is somewhere between 15% and 35% (people underestimate this), and the variation is wider than the variation on any other segment.

That number is the answer to whether the last mile deserves more operational attention than it currently has. On most of our clients' books the answer is yes, and the structural fix usually pays back inside two quarters. The UNCTAD trade and logistics reporting at the macro level shows the same picture: the share of total logistics cost concentrated in the post-port leg has been rising in absolute and relative terms since 2021. Treating it as a footnote was always going to age badly.