The Bottling Line Wasn't the Bottleneck

Picture a bottled water plant on a hot Thursday in July. The filling lines are running. There is water, there are bottles and caps, and the finished pallets are wrapped and stacked in the staging lanes, every one of them already sold. The plant is behind on shipments anyway.

Ask how much capacity that plant has and the answer usually comes back from the filler, the machine that fills the bottles and often sets the pace of the whole line: cases an hour at its demonstrated rate. It is a real number, and it is measured at the start of a chain that ends at the gate. None of those pallets earns a dollar until it is on a trailer that has left the property.

So here is this week's question: how many loads can that plant ship in a day?

Every plant has two capacity numbers

ASQ's glossary defines throughput the way Eliyahu Goldratt did, as "the rate the system generates money through sales," and adds a second reading: "the conversion rate of inventory into shipped product." By either reading, a pallet of water in a staging lane is still inventory. It counts when it ships.

Between the filler and the public road, that pallet passes through a chain of steps, and each step has its own ceiling. The line bottles it. The warehouse stages it. The dock loads it. Then the yards take over: an empty trailer has to be at the right door when loading is ready, the loaded trailer has to be pulled and parked where the driver will look for it, the paperwork has to exist before the truck does, and the driver has to get checked in, hooked up and out the gate. The yards' number is the loads a day the gate, the dock office and the spotters can release with the people and trailers they have. Count every step in loads a day and the plant ships at the rate of the slowest one. Call that slowest number the plant's realized production capacity, because it is the only production that turns into revenue.

Anyone who read Goldratt's The Goal remembers Herbie, the slowest kid on the scout hike, who set the pace for the whole troop no matter how fast the boys at the front could walk. Alex Rogo's fix was to take the weight out of Herbie's pack and split it among the kids who could carry it. Time saved anywhere other than the constraint, Goldratt argued, buys the system nothing.

It is natural to look for Herbie on the production floor, because that is where the expensive machines are. I would check the other end first. The yards are the last machine on the line. They rarely appear on the line diagram, and their speed gets set by things nobody puts in a capital plan: a driver waiting on paperwork, an empty trailer parked at the wrong door, a spotter who never got the move.

What happened at Primo Brands

Primo Brands bottles water, and it is a YardFlow customer, so read what follows as a vendor's numbers, each labeled by kind. Their team put the goal in writing: take on more volume without adding people in the dock office. My own read of the problem is simpler. The plants could bottle the water. The limit was how many loads the facilities could turn and release in a day.

The work started with the driver: check in from a phone, sign a digital bill of lading, get a text with the trailer's spot. In drop and hook the loaded trailer is waiting before the driver arrives, so the driver's minutes are not dock minutes. They become capacity in two places. At the window, a few people in the dock office can only check in, print and sign for so many drivers an hour. At the door, the arriving driver often drops the empty the dock loads next, so a truck stuck at the window is an empty that has not reached a door. Moving check-in and paperwork onto the driver's phone is Rogo's move: it takes weight out of the dock office's pack and hands it to someone who can carry it.

Three things came out of it, and they are three different kinds of evidence, so I am going to keep them apart.

Read the 5% from a capacity planner's chair. The dock office did not grow, and it handled every one of those extra loads. The workflow was not the only thing that changed at those sites in 2025, and a year-over-year number cannot say which change did the work, or promise 5% at your plant. What it does give a capacity planner is a result in the right unit: product shipped, with the same people in the dock office.

If the faster turn is behind any of that 5%, it also tells you where the ceiling was. A faster turn can only add shipments at a plant where turns were holding shipments back. Where the bottling line is the constraint, a faster turn shortens the wait and the freed-up capacity sits unused.

Don't divide 48 by 24

It is tempting to look at 48 and 24 and decide the plant doubled. It did not, and the reason is worth a minute.

Little's Law says that in a stable process, the average amount of work inside it equals the rate it flows times the time each unit spends there. Halve the time each truck spends in the turn, keep the same number of trucks in it at once, and the turn by itself can move twice as many. The turn is one step, though. The line still runs at its rate, the warehouse still picks at its rate, carriers still show up when their schedules let them, and demand is whatever it is. Speed up the constraint and the plant ships more until it reaches the next ceiling. That is how a turn can take half the time while the plant ships only a little more.

It also means most saved minutes never become a load. The minutes that turn into shipments are the ones that were standing between a finished load and a truck that could have taken it.

The rest are worth something too, mostly to the carrier. Jake priced that side on October 18: David Correll of MIT's FreightLab told Congress in 2021 that the drivers he studied drove about six and a half of the 11 hours the law allows, and ATRI put 2023's detention bill at $3.6 billion in direct expenses and $11.5 billion in lost productivity. Those are carrier numbers, and they do not belong in a shipper's business case. Your share arrives through tender acceptance, which was edition 4.

Walk one load backward from the gate

When a plant cannot ship what it makes, the usual answers are expensive: another shift, overflow storage down the road, a 3PL, more trailers in the pool, a new distribution center, and eventually another line or another building. Sometimes that is the right call. But if the yards were the ceiling, a new line just makes more product to wait at the same gate.

Before that check gets signed, take one load and walk it backward from the gate:

1. When loading was ready to start, was an empty trailer already at the door?
2. When loading finished, did the spotter get the move right away, or find out later?
3. Was the loaded trailer parked where the driver was told it would be?
4. Did the bill of lading and the release exist before the truck arrived?
5. Of the driver's time on site, how much was spent waiting on the facility?

Those five answers set the plant's realized production capacity.

The plant you did not build

Here is the arithmetic that should get a CFO's attention. Take a network of 20 plants of the same size. If each one ships 5% more on the days the yards hold shipments back, the network ships 20 × 5% on those days, a whole plant's worth of outbound loads, from the plants it already has.

The working version is on yardflow.ai if you are reading this somewhere else. The days matter more than they look. A ceiling in the yards binds on the busy days, and for a water bottler those are summer days, which is also when an overflow truck and a rented warehouse cost the most.

Two conditions apply before anyone books it: someone has to want the extra loads, and the plants have to be able to make them. If either one is false, the yards are not your constraint, and as Jake put it in August, you should say so plainly and run the cost case.

The Monday morning question

Production capacity has a number and an owner, usually someone in manufacturing with a capital plan. Freight has a number and an owner in transportation. Realized production capacity, the rate at which a plant turns finished goods into departing loads, sits in between.

Who at your company owns the number for how many loads each plant can ship in a day?

Reply on LinkedIn and tell me who it is, or that it is nobody. Both answers are useful.

Sources

1. ASQ Quality Glossary, "Throughput" and "Theory of constraints": asq.org/quality-resources/quality-glossary/t
2. Eliyahu M. Goldratt and Jeff Cox, The Goal (North River Press, 1984). Herbie is the scout hike chapter.
3. Michael R. Godfrey and Brent Bandy, "Applying Little's Law and the Theory of Constraints," Six Sigma Forum Magazine, February 2005: asq.org
4. MIT Center for Transportation and Logistics on David Correll's testimony to the House Transportation and Infrastructure Committee, November 17, 2021: ctl.mit.edu
5. ATRI, "New Research Documents Substantial Financial and Safety Impacts from Truck Driver Detention," September 10, 2024: truckingresearch.org
6. Primo Brands: the 48 to 24 minute turn is YardFlow's measurement in the initial pilot at one site; the ~5% is YardFlow's analysis of Primo Brands' operating data for the sites running the driver workflow, second half of 2025 against a year earlier; the quote is from Primo Brands' Business Solutions Manager, Inventory to Deliver, in writing.
7. Jake Koppinger, Yards First, "We cut the turn in half. That was the smallest thing that happened.", August 14, 2026, and "The trucks are already here. Turn them.", October 18, 2026: yardflow.ai/newsletters/yards-first
8. The network scenario is arithmetic on the assumptions shown, not a customer result.