Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- This week, pull gate-in and gate-out timestamps for your ten most recent intermodal moves; convert any lane whose average ramp dwell is under two days and hold any lane above three.
- Before converting a lane, get a drayage commitment in writing at both ends for the same volume, since the line-haul rate says nothing about whether a chassis and a tractor will be there.
- Re-base every customer delivery promise on door-to-door days including both ramp dwells, and add two days of buffer to any lane converting in the next 90 days.
- Treat rail as the default only above 700 miles; below that, model the ramp dwell against the line-haul saving before assuming rail is cheaper.
- Track ramp dwell and dray on-time weekly against a 95% target, and pull a lane back to truck if dwell exceeds three days for three consecutive weeks.
InTek's July intermodal report cited CSX intermodal volume up 9% (its largest growth contributor, tied to truck-to-rail conversion), Norfolk Southern intermodal revenue up 22% on 5% more volume, and Union Pacific domestic intermodal up about 19%. The shift reflects tightening truck capacity pushing freight to rail on lanes over roughly 700 miles.
The Analysis
Rail does not move your freight. It moves your box between two ramps, and a truck does the rest. That sentence is the whole story in InTek's July intermodal report, and it is the part the growth number skips.
CSX intermodal volume up 9%, and the report names it the largest growth contributor, tied to truck-to-rail conversion. Norfolk Southern intermodal revenue up 22% on 5% more volume. Union Pacific domestic intermodal up about 19%. The shift is on lanes over roughly 700 miles, and the cause is not a love of trains. It is that the truck did not show up.
Everyone is reading this as rail winning. I read it as freight being pushed onto a network with two more handoffs than the one it left, and handoffs are where things queue.
Count the handoffs on an intermodal move. A trailer gets loaded at the shipper's door. A tractor drags it to the ramp. It gates in and waits for the block to be made up. The train runs. It arrives, gets grounded, waits for a chassis, waits for a dray tractor, gets pulled to the consignee, gets unloaded. That is two more places to wait than a door-to-door truck, and each of those places has a gate that opens at a fixed hour and a lot with a fixed number of spots.
The 700-mile line in the report is not a rate threshold. It is the distance at which the line-haul saving finally pays for those two extra handoffs and their queues. Below it, the ramp eats the saving and you paid the dwell for nothing. Above it, rail wins on cost and loses on days. Both statements are true at once, which is why this decision keeps getting made badly.
The reason the freight is converting is physical and it is not going away this quarter. Trucks are tight, so a 1,100-mile lane that used to be a truck move becomes a rail move, and the shipper congratulates himself on the saving and then discovers the ramp. A ramp is a parking lot with a cut-off schedule. When volume at one railroad rises 9%, the queue at the gate does not rise 9%. It rises faster, because arrivals bunch around the block cut-off. Twenty extra boxes on a Tuesday morning is not twenty extra boxes spread across a week. It is a line of trucks down the access road.
Here is the second-order effect almost nobody prices in. The drayage that intermodal needs at both ends comes from the same driver pool and the same chassis pool as the port drayage that never converted, because a 300-mile inland move from a coastal port does not cross the 700-mile line and stays on a truck. So the conversion does not remove trucks from the road. It takes long-haul tractors off the interstate and turns them into short-haul tractors circling a ramp, competing with port drayage for drivers who must be home at night. A long-haul driver can sleep in the truck. A drayage driver cannot, so the job competes with local work at local wages, and that pool is smaller and far less elastic than the one it is being asked to replace. The escape valve is constrained by the exact shortage it is supposed to relieve.
Norfolk Southern's numbers say the quiet part. Revenue up 22% on volume up 5% means the railroad is pricing into the tightness, and it can, because it knows what the truck costs. The cheaper door is getting less cheap on the way in, and the shipper who models rail as a fixed discount is modelling last year.
Timing splits into two clocks. The conversion decision is made at bid, so it lands at your next contract cycle, not this week. The congestion arrives four to eight weeks later, when the drayage fleet at that ramp runs out of slack and the first appointments start slipping. By the time your transit times move, the cause is two months old and the fix is a phone call to a drayage dispatcher, not a rate negotiation.
Here is my own arithmetic, and the assumptions are mine, not the report's. Assume one train drops a block of 100 boxes at a ramp, and the drayage fleet serving that ramp can pull 40 a day at current staffing. The last box moves on day three. Those three days do not distribute evenly. They land hardest on whoever booked the latest appointment, which is usually the shipper who confirmed his order last. Now assume cargo value of $200,000 a container and an 8% carrying cost: three extra days is about $131. The carrying cost is nothing. The promise is everything. If you sold a six-day door-to-door and the ramp gives you nine, you did not lose $131, you lost the account's trust and probably a chargeback.
What the coverage misses is that intermodal conversion is a reliability trade disguised as a cost trade, and the metric that decides whether it sticks is not the rate per mile, it is the on-time percentage at the ramp gate. Shipments that arrive on the train but miss the dray appointment are, from the consignee's chair, lost freight. A railroad that holds transit but cannot turn a ramp will lose the conversion back to trucks the moment capacity loosens, and it will lose it quietly, one tender at a time. Watch ramp dwell, not volume growth. Volume growth is what a railroad reports when it is winning. Dwell is what it reports when it is struggling, and only one of those two numbers tells you whether your box moves on Thursday.
The counterfactual is clean. Had truck capacity stayed loose, most of this 700-plus-mile freight would still be on the road, and intermodal would be growing at roughly the rate of the economy. The conversion is a price signal working through a physical network, not a structural shift in how North America moves goods. If the railroads hold service through the next two quarters, some of it sticks, because shippers who redesign a lane around a ramp rarely redesign it back for a small rate difference. If ramps choke, the freight returns to trucks at a higher rate than it left, and the shipper pays for the round trip twice.
The traps are specific. Booking rail on the line-haul number and forgetting the drayage is a separate contract with a separate capacity problem. Assuming a ramp appointment is a reservation when it is often a queue ticket. Forgetting that a chassis is a piece of equipment somebody has to own and position, and that chassis pools tighten before anybody announces it. And the one that costs the most: converting a customer-facing lane to rail without re-basing the delivery promise, so you keep the old date and absorb the variance yourself.
Pick one lane this week, the longest one you have, and pull the actual gate-in and gate-out timestamps for the last ten moves. If the ramp dwell is under two days, convert and keep the promise. If it is over three, do not convert this quarter no matter what the line-haul quote says, because you are not buying a cheaper mile, you are buying a queue.
↑ Back to the key points