Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- Before committing any lane over 700 miles to rail, get the block size per arrival in writing; if it exceeds your dock's daily trailer capacity, secure a holding yard or reject the conversion.
- Measure door-to-door transit variance across four weeks rather than the average, and keep any load with a firm delivery date and a tolerance under one day on truck.
- Name the chassis supplier and the pool agreement in the contract before the first container moves; a shared pool with no named allocation is not capacity.
- Re-plan receiving for block arrivals: if a block is twenty trailers, staff and assign doors for twenty in one shift, and clear the yard within 24 hours.
- Set a re-quote trigger: if intermodal revenue per unit keeps climbing at the pace implied by 22% on 5%, rerun the truck comparison within 30 days instead of waiting for the annual bid.
North American intermodal demand stayed strong in mid-July, with CSX reporting intermodal volume up 9% (citing truck-to-rail conversion), Norfolk Southern's intermodal revenue up 22% on 5% higher volume, and Union Pacific domestic intermodal up about 19%. The gains reflect tightening truck capacity pushing freight onto rail for lanes beyond roughly 700 miles.
The Analysis
CSX moved 9% more intermodal volume and put the gain down to truck-to-rail conversion. Norfolk Southern's intermodal revenue rose 22% on 5% more volume. Union Pacific's domestic intermodal was up about 19%. The common explanation for all three is that truck capacity tightened and shippers pushed freight onto rail beyond roughly 700 miles.
I have spent enough time around ramps to be suspicious of a good line-haul story. A train that leaves at nine at night and covers 800 miles by the second morning is not the hard part. Railroads have done that for a century and they are good at it. The hard part is the eleven miles at each end, and those eleven miles do not grow when the train does.
Look at the three numbers again and notice they are not the same kind of number.
Three different denominators sit in that paragraph, so do not rank them. CSX gave volume. Norfolk Southern gave revenue against volume. Union Pacific gave domestic intermodal, which cuts out the international piece. Three spoons out of the same barrel.
The pair that says something is Norfolk Southern's. Revenue up 22% on volume up 5% works out to about 16% more revenue per unit, and that is division on two published figures, not a yield the company reported. You do not get 16% per unit by running fuller trains on the same mix. You get it from mix and from price. Which means the tightness is already being charged for.
CSX's 9% is a cleaner volume story. More boxes moved. Union Pacific's 19% says the same on domestic lanes. Those two tell you conversion is happening. Norfolk Southern tells you conversion is now priced, and that is a different stage of the same cycle.
The threshold is the interesting part. Why 700 miles and not 400? Because under that distance the drayage at both ends eats the line-haul saving, and every extra terminal handling eats another day. So the number is not a property of rail. It is a property of how far a truck has to run at each end, what local drayage charges, and how many days of transit the shipper will tolerate.
That is why the conversion is not evenly spread. It happens on lanes where the dray is short and the delivery window is soft. It does not happen where the dray is long, and it does not happen where the customer counts days and takes a penalty off the invoice when they slip.
Who feels it first is a shipper with a three-day promise, a distribution center that takes twelve trucks spread evenly through the day, and a customer measuring on-time performance. Their freight converts because the truck rate moved, the promise does not move with it, and the receiving dock was laid out around arrival times that no longer exist.
Who is fine is a shipper with ten-day windows and a yard big enough to park trailers. Rail's extra day costs them nothing they were selling, and the saving is real. Also fine: anyone whose origin and destination both sit within forty miles of a ramp with on-dock rail and a chassis pool that is not shared with three other importers.
The middle group is the one that gets quietly hurt, and it is the largest group. Importers who clear at a port 150 miles from the loading ramp. Their dray is long, their chassis is whatever the pool has that morning, and their buffer is one day.
Mid-July is early in this cycle. Conversion talk runs about six weeks ahead of the physical reality, because the first cargo that moves is the cargo that was already flexible. The dwell shows up afterwards, when the volume that has no choice starts arriving on the same ramp.
The symptom order is consistent. First the chassis gets hard to find at the origin ramp, because the pool is shared and everybody converted in the same month. Then appointment times at the destination ramp stretch from one day to two. Then the dray carrier stops quoting a one-day window. Then a delivery date slips by a day and nobody can explain which hand it slipped through. By the time the train itself is late, you are three problems deep and the train was never the problem.
Here is what a conversion does that the volume numbers cannot show. Rail does not smooth arrivals. It batches them.
Forty truckloads a week spread across five days is eight a day, which any decent dock can take. Forty truckloads a week arriving as two blocks of twenty is a different building. You need twenty trailer spots, twenty chassis, and a crew that can unload twenty in a shift, and then you need nothing for two days. The weekly average is identical. The operation is not.
Most distribution centers are staffed and laid out for the average. That is why a conversion that looks like a 15% saving on paper turns into overtime, yard congestion and a receiving backlog in the first month, and why the saving is usually booked in month one and gone by month three.
Run the sum. Assume forty loads a week on an 800-mile lane, my assumption, and that rail adds two days door to door, which is the usual shape of the trade at that distance. Two extra days at forty loads a week is eighty load-days of freight in transit at any moment. At twenty pallets a load, that is 1,600 pallet-days of inventory you are now financing and insuring, and not one pallet of it is under a roof.
Then look at the dock. If those forty arrive as two blocks of twenty, your peak day is twenty trailers against a normal eight. A dock with eight doors on two shifts can physically do that. A dock with four doors cannot, and no amount of planning changes the number of doors.
The saving on the line is real. It is spent at the two ends.
The trap is signing a rail conversion off a rate comparison and never touching the receiving calendar. The second trap is the chassis: it is a shared pool, it is not yours, and it empties in exactly the week when everyone else converts. The third is measuring ramp to ramp instead of door to door, which quietly hides the two days that break your promise.
The honest test is variance, not average. Rail's average transit is acceptable on most lanes past 700 miles. Its tail is not, and the tail is what your customer sees on the one shipment that matters.
Call the ramp this week and ask two questions. What is the block size on this service, in trailers per arrival? And who supplies the chassis at origin, under what agreement? If nobody can answer the second question in one sentence, the rate is not the number you should be negotiating.
↑ Back to the key points