Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- Every week the AAR print lands, put your own box count for the same week beside it. If your ramp runs more than five points above the national number, treat your lane as tight whatever the headline says.
- Book fixed weekly ramp appointments for the next eight weeks now, at the hours you actually need, instead of taking what is left each Thursday.
- Know your chassis per-diem days per box for the last quarter and put a hard ceiling of two days on it before it reaches whoever owns the pool.
- Move the time-critical 10 to 15% of SKUs to the road for the peak weeks and leave the slow movers on rail. Do the split on the calendar, not on the rate.
- Write the trigger down: if dwell passes 48 hours or per-diem days pass two for two consecutive weeks, shift 20% of that lane back to road inside seven days.
The Association of American Railroads reported total U.S. weekly rail traffic of 523,900 carloads and intermodal units for the week ending July 18, up 3.4% year-on-year. Intermodal volume reached 297,017 containers and trailers, up 7.2% YoY, while carloads eased 1.2% to 226,883. Year-to-date combined traffic rose 3.4% to 14,176,139 units.
The Analysis
523,900 units. That is the whole U.S. rail network for one week, carloads and intermodal together, up 3.4% on the same week last year. The number underneath it is the one to keep: 297,017 containers and trailers, up 7.2%, against 226,883 carloads, down 1.2%.
A national weekly total is a strange thing to plan against. It is an average of a country. Put it in physical units and it stops being abstract: those 297,017 boxes over seven days is about 42,431 a day, about 1,768 in any given hour. One box roughly every two seconds, somewhere on the network, all week, including at four in the morning.
That is the pace a ramp has to hold. When a weekly print moves up, it does not move up evenly. It moves up at the four or five terminals that were already the tightest.
What stands out is not the 3.4%. It is the 8.4-point gap between intermodal at plus 7.2 and carloads at minus 1.2. Same railroad, same track, same crews, moving in opposite directions. What that describes is a network being re-cut around boxes. A carload of anything bulk, grain or chemicals or aggregates, and a stack of 40s do not need the same siding, the same dwell, or the same yard track. When the mix tilts, the constraint does not stay put.
Follow the physical consequence. More boxes and fewer carloads means more lifts, more gate moves, more chassis turns, more drayage, on the same footprint. A terminal does not get bigger because the mix changed. What goes first is appointment availability. Then free time gets enforced instead of tolerated. Then the chassis pool runs short in the afternoon. Then the drayage rate for that zip code moves. None of that appears in a carload print. All of it appears on your invoice.
Size decides who absorbs it. An importer running 200 boxes a week through one ramp has a dedicated drayage fleet, a chassis pool it pays for, and a person whose entire job is the gate. They feel this as a scheduling annoyance and a worse turn time. An importer at 15 boxes a month through the same ramp has none of that and takes whatever slot is left on a Thursday afternoon. The difference between those two is not volume. It is whether anyone is awake at six in the morning when the appointments open.
Direction changes the answer. Importers feeding a fixed inland DC can shift volume between ramps and between modes on two weeks' notice. Exporters with a fixed plant and a fixed dock cannot. Their outbound box leaves from where the plant is, and the nearest ramp is the nearest ramp. An exporter running a seasonal peak is the most exposed of all, because the peak is precisely when the network is least willing to be flexible.
A weekly print is a lagging number by the time you read it, and the ramp feels the shift two to three weeks after the volume actually moves. What you notice first is fewer green slots at the hours you want and more of them at five in the morning or eleven at night. Then dwell. Then a call about a box that missed a cut-off. The rate is the last thing to move and the least informative when it does.
Here is the comparison nobody runs. Take the year-to-date total of 14,176,139 units and divide by the number of reporting weeks so far. I am assuming twenty-eight, and the release does not state it, so treat that as my assumption and not a fact from the report. That gives roughly 506,000 units a week as the year's own pace. This week is about 3.5% above it. Line that against the 3.4% year-on-year and something the headline misses comes into view. The growth is not a spike off one soft week last July. It is a level shift that has been running all year. For planning, those are different animals. A spike you wait out. A level shift you re-plan around.
The print also hides where the growth is not. Carloads down 1.2% means the bulk shipper is losing crew time and siding space to boxes, quietly, a little at a time. That shipper is not in your competitive set, but it is on your railroad, and it is paying for the re-cut in service quality rather than in rate.
Scale it to your own book. If a tenth of those 297,017 weekly boxes had to go back on the road, because a customer pulled a program or a ramp lost a week, that is about 4,243 extra loads a day hunting for tractors. My assumption is one driver, one box, one turn, which is roughly what long haul is. No fleet is sitting idle to absorb 4,243 loads a day. That is the honest reason this conversion is happening, and the honest reason it will not reverse quickly, whatever diesel does next month.
Without the shift, those boxes would have had to move by road, and the road has spent two years saying it cannot. That is the counterfactual, and it is why I do not expect this to unwind. The pitfall is assuming your ramp behaves like the national average. It does not. A system number up 3.4% can sit on top of one terminal up 20% and another flat, and you only care about one of them. So the useful move this week is not to read the print. It is to find out which of those two terminals you are on, and what your own last eight weeks look like against it.
↑ Back to the key points