← Back to Field Notes

March 21, 2026 · The Bloomfield Team

How Late Shipments Actually Start

A shipment goes out three days late. Someone asks the production manager what happened. The answer focuses on the last 72 hours: the job took longer than expected, the machine went down, first article failed. Every explanation points at the end of the chain.

The chain started breaking weeks earlier. Late shipments in job shop manufacturing almost never originate where they become visible. The dock is where you see the problem. The origin sits two to four weeks upstream in quoting, order entry, purchasing, or planning. Production speed cannot fix what was already broken before the job hit the floor.

The Quoting Phase: Where Lead Times Get Set Wrong

Late deliveries often begin the moment the lead time is quoted.

An estimator receives an RFQ for 200 precision-ground shafts. Customer wants four weeks. The estimator checks ERP capacity, sees an opening in grinding three weeks out, confirms four weeks. Quote goes out. Customer places the order.

What the estimator did not check: 4340 alloy steel in the required diameter is running six-week lead times from two of the shop's three suppliers. The third can deliver in three weeks at a 15% premium that was never included in the quote. The estimator quoted based on processing time and assumed material availability. That assumption created a delivery commitment the shop cannot keep without absorbing a cost increase or missing the date.

A 2023 Precision Machined Products Association study found that 23% of on-time delivery failures in surveyed job shops traced back to lead time commitments that did not account for current material availability. The estimator had no way to check material lead times during quoting because that data lived in purchasing's email threads.

The quoting process already carries pressure to respond fast. Adding a material lead time verification step slows the quote. Skipping it creates a delivery promise built on an assumption that may be wrong. The fix is making material availability data accessible to the estimator at the moment they build the quote, which requires connecting systems that currently operate in isolation.

The Order Entry Phase: Where Details Get Lost

Customer approves the quote and sends a PO. Someone in the front office creates the job record, enters quantities, assigns operations, sets the due date.

During this handoff, critical details from the RFQ and PO get lost or simplified. A note on the RFQ specified individual wrapping. The order entry person did not see it because it was on page three of a six-page email thread. Parts get produced correctly, packed in bulk. Customer rejects the shipment. It comes back, gets repacked, ships again four days late.

Another common failure: the customer's PO references a drawing revision that differs from the revision quoted. Order entry logs the PO and part number without comparing revisions because the quoting file lives in a different system. The job launches with the wrong revision. First-article inspection catches it, consuming a day. Correct revision gets programmed, set up, and run, consuming two more days. Three days late because of a detail missed at order entry.

These are small administrative gaps between quoting and production, places where information crosses a system boundary and loses fidelity. Each gap adds hours or days. Stacked across a month, they account for a meaningful percentage of late deliveries.

The Purchasing Phase: Where the Clock Starts Without Anyone Noticing

Material procurement is the most common upstream cause of late shipments, and it operates on a different timeline than production. A job with three weeks of processing time and a five-week delivery commitment looks comfortable in the ERP. It looks comfortable until the purchasing agent realizes material will not arrive until week three, leaving zero margin for the actual machining.

The root cause is timing. Material is ordered after the job is created in the ERP, which happens after the PO is received, which happens after the quote is approved. Each handoff adds one to three days. By the time the purchase order reaches the supplier, a week or more of the delivery window has evaporated with no production progress.

Some shops carry inventory on frequently used bar stock, sheet metal, and hardware. That helps for repeat work. It does nothing for one-off jobs or specialty materials, which are precisely the jobs with the longest material lead times and the highest risk of late delivery.

A purchasing manager at a 45-person aerospace shop put it plainly: "We get the PO on Monday. I enter the material requisition Tuesday morning. The buyer places the order Wednesday. The supplier confirms delivery in four weeks. The job is due in five weeks. We need three weeks of floor time. That leaves us one week of negative margin on the schedule, and nobody knows it until the material actually arrives and someone in planning does the math."

That math was doable on the day the quote was sent. Nobody did it because the quoting system, the purchasing system, and the scheduling system are three separate tools with no connection between them.

The Scheduling Phase: Where Conflicts Hide

New job enters the schedule. The planner works backward from the due date, slots it into available capacity. On paper, it fits.

The schedule shows Machine 4 available Tuesday for a 6-hour first operation. What the schedule does not show: Machine 4 is running a job 4 hours behind estimate and will not finish until Wednesday morning. The operator who ran this part type before is on vacation next week; the replacement needs an extra 2 hours of setup. Machine 4 is due for scheduled maintenance Thursday, taking it offline for a full shift.

Every one of these conflicts is knowable. The ERP has actual versus estimated progress. The attendance system has the vacation schedule. The maintenance system has the PM calendar. None of these systems feed into scheduling in real time. The planner builds from the ERP's static capacity view. The floor reveals conflicts one at a time as the week unfolds.

A schedule built on incomplete information generates late deliveries because missed conflicts consume the buffer time that was supposed to protect the ship date. Buffer gone, job runs late. When several jobs run late simultaneously, triage begins, and some work gets pushed further to make room for the most urgent. One scheduling gap cascades into three late shipments.

Tracing the Chain

A late shipment that looks like a production problem on Friday afternoon is the final symptom of a sequence that started weeks earlier. Lead time quoted without checking material. Drawing revision missed at order entry. Material ordered three days late. Schedule failed to account for a machine running behind on the prior job.

Each upstream event removed a day or two of margin. By the time the job reached the floor, the remaining time was too tight to absorb any production variability. The machinist ran the parts in exactly the estimated time. The job still shipped late because it arrived at the machine three days later than it needed to.

Measuring on-time delivery honestly is the first step. Tracing each late shipment back through the chain to its origin is the second. When shops start categorizing late deliveries by root cause, the distribution surprises them. Production speed accounts for less than 40%. The majority originates in the administrative and planning processes that precede production.

The data for this analysis exists across the shop's systems: quoting records, purchase orders, material receipt dates, job start dates, operation timestamps, ship dates. Connecting that data reveals the upstream patterns that no amount of floor-level expediting can fix.

Fixing the Origin Points

Once the root cause distribution is visible, fixes become specific. If 25% of late deliveries trace to material procurement timing, integrate material lead time data into quoting so lead times reflect actual supply chain conditions. If 18% trace to scheduling conflicts with jobs already in process, build a scheduling tool that shows actual progress against plan and flags conflicts before they consume buffer time. If 12% trace to missing information at order entry, build a structured handoff between quoting and production that verifies critical details before the job launches.

Each fix addresses a specific link. The combined effect is a delivery process that stops generating the problems it spends most of its energy reacting to. Production stops fighting fires. The shipping dock stops being the place where upstream failures become visible for the first time.

The tools to connect these upstream processes and surface risks before they become late shipments are buildable now. The hardest part is accepting that the answer to "why was this late" is rarely "production was slow." It is almost always "we did not have the right information at the right time, three weeks ago."