Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By the end of this week, pull your last eight weeks of car orders and compare requested placement date with actual; if the gap is two days or more, add that gap to your order lead time instead of absorbing it in the yard.
- Rebuild safety stock on the two lanes with the widest placement gap to cover 1.5 times the current gap in days, and set the review date 30 days out.
- If you run private or leased cars, audit where the fleet sits at the end of each week for the next month; anything idle more than 72 hours at an origin you no longer ship from is a repositioning decision, not a waiting decision.
- Add one receiving day per week at your destination for the next quarter and measure whether placement-to-unload time drops by at least 24 hours.
- Write the escalation threshold down now: if cars online on your serving railroad rises again while speed falls for two consecutive weeks, go to your commercial contact within 48 hours rather than waiting for the invoice.
A newly updated dashboard tracks average train speed, dwell time and cars online for all US Class I railroads, giving shippers real-time visibility into network performance. The tool arrives as intermodal and bulk shippers weigh routing amid congested ports and tariff-driven volume shifts.
The Analysis
A dashboard is a mirror, not a road. The newly updated Class I rail dashboard now publishes average train speed, dwell time and cars online across the US railroads, and it lands in a month when intermodal and bulk shippers are redrawing routings because ports are congested and tariff-driven volume has moved.
I have spent enough hours at origin yards to be suspicious of all three numbers in the same way. They are true. They are current. And none of them is the number that decides whether your car gets spotted on Thursday.
Here is where you actually wait, what that wait costs, and which single decision takes you out of the queue.
Three numbers, and each one measures a different piece of the same journey. Average train speed is line-haul: terminal to terminal, wheels turning, yard time excluded. Dwell is the time a car spends sitting between arriving at a yard and leaving it. Cars online is a headcount, the number of cars physically on the network at that moment. Side by side they give you a decent picture of a railroad. Put them against your own shipment and you still have a picture of a railroad, which is not the same thing as a picture of your car.
The reason this is worth reading closely right now is that the volume has moved. Congested ports and tariff-driven shifts have pushed cargo onto different gateways, which means the origin of a rail move has changed for plenty of shippers even when not a line of their contract changed. Rail equipment is positional. A car is not fungible across origins the way a dollar is fungible across accounts. The empties are sitting where last year's demand was, and the yard now receiving the volume was staffed for the volume it had before. So dwell climbs at the new origin, and it climbs for a reason that has nothing to do with how well that yard is run: it is absorbing a shape of traffic it was not built for. Speed falls on the new lane for the same reason, because more stops and more meets.
Who feels this first depends less on company size than on who owns the asset. Ship bulk in your own cars and you feel it as per-diem and as equipment stranded at the wrong end of the system, but you have a contract lever to pull. Ship intermodal in railroad-supplied equipment and you feel it as a slot that does not exist on the day you wanted, and your lever is a phone call and a relationship. Small shippers are the ones the dashboard cannot see at all: four carloads a month disappears into a network average, and no line on that page will ever move because of you. Very large shippers have their own instrumentation and their own remedies. The exposed party is the middle, big enough that a week of dwell is real money, too small to have a lane-level view or a service commitment worth enforcing.
The transmission order is worth knowing because it decides when you act. Dwell moves on the dashboard first, usually within a day or two of the physical congestion, since it is measured at the terminal and terminals fill visibly. Speed averages move more slowly, because a network average is a window that has to drift. Your invoice moves last; demurrage and per-diem are billed after the fact, which means the first hard evidence you get arrives roughly a month after the decision that caused it. The symptom at your own dock arrives in between, and it does not look like a delay. It looks like nothing on Monday, nothing on Tuesday, two cars on Wednesday.
Here is what the dashboard does not say, and this is the part I would underline. When cars online rises while speed falls, the network is not merely busy. Loaded cars are accumulating, which means the railroad is using track as a warehouse, and the inventory being warehoused is somebody's working capital. Yours, if you are the shipper. Read that pair of lines as a balance-sheet signal rather than an operations signal and the picture changes: a rising cars-online count against a falling speed is a statement that your goods are being stored for free on someone else's real estate, at your cost of capital. The second thing it leaves out is that dwell is blended. High dwell at origin means your car has not left the yard yet. High dwell at destination means your car arrived and you cannot get it in. Identical number, opposite remedy, opposite person to call. Collapsing two queues into one figure is the least useful form of the most useful data on the page.
Let me put a number on it, with the assumptions out in the open because none of these come from the article. Assume you move thirty cars a month on one lane, assume the average car in your mix carries material worth 150,000 dollars, and assume your cost of carrying inventory runs about eight percent a year. That works out near 33 dollars per car per day. If the current congestion adds two days to the cycle, thirty cars times two days is sixty car-days, about 2,000 dollars a month, or 24,000 dollars a year, before any per-diem or storage charge your contract may add on top. None of those four figures appear in the article. The mechanism does. The arithmetic is mine, and if your car value or your carrying cost is half of what I assumed, the answer halves with it. Run it with your own numbers before you decide it is small.
The action that decouples you is not watching the dashboard harder. Visibility does not add a car, a crew or a track. What decouples you is moving the decision earlier than the congestion. Order your empties against a transit assumption that includes the current dwell rather than last quarter's, and pull the order date back by the amount of dwell you are seeing instead of arguing about it afterwards. If you own or lease private cars, the fleet decision is the real lever, because it is the only place in the chain where you hold the asset instead of queuing for it. At destination the fix is usually inside your own fence: receiving hours, dock doors, and whether your team can take a car on a Saturday. Half of the dwell that shows up as a railroad number is a shipper-side appointment problem wearing a railroad costume.
The trap is treating a published average as a promise about your lane. It is not a promise, it is a population statistic, and your car is one observation in it. The other trap is reading speed and dwell as one story. They are two queues with two owners. Pull the number that corresponds to the queue you are actually standing in, call the person who owns that queue, and do it this week, because by the time it reaches your invoice the argument is already over.
↑ Back to the key points