.jpg)
American factories grew for the ninth straight month in September, and plenty of plant managers will have closed the month with their production targets met. Many of them likely finished it further behind their customers than they started. The Institute for Supply Management's survey, released October 1, shows order backlogs swelling even as output rose, and a production target on its own is poorly built to show that gap.
The target counts units against a calendar. Customers keep a different tally, built on whether the order they placed turned up when they needed it, and the two can drift apart for months while the plant whiteboard stays green. That comfort may be the best reason to trust the number a little less.
What September's Survey Revealed
The survey's backlog of orders index jumped 4.6 points to 56.4 in its ninth consecutive month of growth, on a scale where anything above 50 signals expansion. Production came in at 56.7, so output kept growing too. Customers' inventories sat at 41.6, a level purchasing managers class as too low.
Demand has come back faster than supply. PNC's economists described a strong rebound after the sector spent most of the previous three years contracting, while ISM's supplier deliveries index rose to 59.0, a sign of slower deliveries. When material runs short, a plant tends to build whatever it has parts for rather than whatever customers ordered first.
Where the Product Mix Goes Wrong
Mix is often where the drift starts. Take a hypothetical snack plant in Ohio that owes its planners 40,000 cases a week. It hits the number by running 9,000 more cases of its bestselling flavor than anyone ordered, since that line hardly ever jams. A grocery chain's short run, due Thursday, slips to next week, yet the weekly report still reads 100 percent.
Big retailers tend to notice that kind of week. Many now score suppliers on on-time, in-full delivery and penalize the misses. Definitions vary, though. A McKinsey survey of 24 North American retailers and manufacturers found no standard version of the metric, and 92 percent said one would create value.
The largest customers are the sensible place to start. Agreeing with each on what "on time" and "in full" mean, then posting that score beside the daily count, lets the shift see the customer's version of the week while it can still act.
Score Volume, Mix and Sequence Together
Timing compounds the damage. Some lean practitioners use a less forgiving score, build to schedule, which multiplies volume, mix and sequence so a perfect count can't paper over a scrambled week. A line that makes all its volume, gets the mix 80 percent right and builds in sequence three times in four lands at 60 percent, probably about what its customers would have guessed.
A plant running on one count tends to catch that gap late, because the count lives on a whiteboard updated twice a shift. That delay is the practical case for factory productivity software, a category of tools that logs output by the hour and by product while the shift is still running. A line lead can then see at 10 a.m. that the mix has wandered, rather than reading about it in Friday's review.
It pays to start small. One line serving a demanding customer is plenty for a first month, scored daily against a schedule frozen 24 to 48 hours out, so a reshuffled sequence shows up in that score the same day.
Why Overbuilding Passes for Success
Overbuilding is the opposite failure, and it can look far healthier than it is. When a target reflects what the equipment can make instead of what customers have ordered, crews run ahead and bank units against a slow week. That runs against the Toyota Production System, built around making only what customers order, when they need it.
Taiichi Ohno, the engineer credited with building that system, ranked overproduction as the most serious waste on the grounds that it breeds other waste and then hides it. Surplus stock ties up cash and floor space. A bad batch made three weeks early may surface only when a customer opens the case.
Changeovers explain why plants drift toward the easy products. Because every switch costs setup time, long runs of familiar items flatter the count while short, awkward runs drag it down. Reporting changeover minutes as their own number, instead of letting them vanish into a low count, shows exactly what the hard runs cost.
Short Crews, Expensive Targets
Labor is the other hidden cost. One fabricated-metals company in the ISM survey described "a severe shortage of workers, limiting our production output," a gap that looks unlikely to close soon. A report released at Ford's skilled-trades summit in Detroit points to about 1.7 million skilled-trades openings a year through 2035, with training programs producing roughly 55 workers for every 100 needed.
Short-staffed plants still hit their targets, just more expensively. Past 40 hours a week, federal overtime rules put a nonexempt worker's time at time-and-a-half or more. Those extra hours often come out of the maintenance windows and training that protect next quarter's output. The month's count looks much the same either way.
Fatigue adds a risk the count rarely shows. Long hours and back-to-back shifts can raise the chance of injuries and accidents, partly because tired people miss things. Fatigue even played a part in the 2005 Texas City refinery explosion.
A simple trigger helps here. Three straight weeks above an agreed overtime share usually points to a staffing problem, since a genuine busy patch rarely lasts that long. Preventive maintenance windows deserve protection while that gets fixed.
The Calendar Bends More Than Output
Soviet industry called the end-of-month rush shturmovshchina, usually translated as storming. Work crawled through the early weeks while plants waited on materials, then crews stormed through the last days on overtime to make the plan. Goods built in that rush had a reputation for poor quality. Plenty of Western plants run a milder version every month.
Looking at 4,877 American manufacturers across two decades, a team led by Kai Hoberg of Kühne Logistics University in Hamburg found inventories running about 4 to 6 percent lower in the fiscal fourth quarter. Stocks then climbed back. The cuts were deepest where companies had cash-flow targets to meet, which suggests the shelves were partly being managed for the annual report.
Storming shows up in a plant's own data. If the last week of the month regularly accounts for well over a quarter of its output, the schedule is probably bending to the calendar, and next month's first days tend to pay for it.
When the Rate Outruns the Process
Boeing offers a costly example of what can happen when output outpaces its process. The FAA froze 737 Max production at 38 jets a month in January 2024, weeks after a door plug blew out of an Alaska Airlines plane. Boeing had to wait until last October to win approval to raise 737 Max output to 42 a month, once inspectors had gone through its manufacturing process.
Customers absorbed much of the delay. By March, CNBC reported, Southwest expected just 46 of the 79 Max jets it had planned on for 2024. United paused pilot hiring as new planes ran late.
Rivals Don't Work to Your Target
Some of the falling behind happens against competitors, and few internal targets register it. The International Federation of Robotics' latest count puts the global stock of industrial robots at about 5 million, after factories added more than 600,000 units last year, an 11 percent rise.
China took 59 percent of those installations, while the EU's total fell 11 percent. A plant that meets the same target every month with the same process is keeping pace with last year's version of itself and nobody else.
Part of the problem lies in how targets get set. Many start from last year's actuals plus a few percent. Back in 1980, the economist Martin Weitzman wrote about that habit as the ratchet principle, under which a good year mostly earns a harder target for the following one. Targets built from the order book, and revisited each quarter, move when customers do.
Put the Count Beside the Order Book
The count deserves its place on the board, just never on its own. Before the next monthly review, plant leaders can pick the line serving their most demanding customer and post its build-to-schedule score and the age of its oldest open order beside the count for four weeks. Any gap between those figures and the count is something the customer already knows.
Backlogs grew in September for a ninth straight month, the same nine months in which American factories kept expanding. How many of those plants posted a green month, and how many of their customers had already started taking calls from a competitor?