Job shop manufacturing: machines grouped by process, jobs queuing between them

Selling capability rather than a product

A job shop sells hours on equipment plus the skill to set it up, which is a different proposition from selling an item. The commitment is to keep a spread of processes available and staffed so that any incoming drawing can be quoted. Work arrives as single pieces, small runs, repairs, prototypes and overflow subcontracted from larger plants, and it comes from many customers rather than a few. That mix is both the protection and the problem: no single customer can sink the business, and no single job repeats often enough to be properly learned. Nearly every operating decision follows from accepting a product mix you do not control.

Buying by the job, and owning machines that will attempt anything

Material is purchased against individual jobs, usually in small quantities from stockholders and service centres who charge more per unit than a mill would but deliver quickly and cut to size. That premium is the price of not committing to volume, and it is normally the right trade. Equipment is general purpose and long-lived, chosen for range instead of speed. The real capital sits partly in the machines and partly in accumulated tooling, fixtures and measuring equipment that allow the shop to attempt unfamiliar work. That stock of capability is invisible from the doorway and expensive to rebuild once it has been allowed to decay.

Where informal scheduling stops working, and why setup decides quality

A small shop schedules in someone's head and it works. Beyond a certain number of concurrent jobs on shared machines it stops working, and the symptom is that nobody can say when a job will finish without walking the floor. What is required is data capture at operation level, so recorded times feed back into future estimates, and a schedule that acknowledges finite capacity. Quality is governed by setup rather than by running, because each job is a fresh opportunity to get it wrong. Verifying the first piece before a run continues is the highest-value inspection in the building, and it is the one most often abandoned under pressure.

Cash parked in queues, and suppliers willing to sell small quantities

There are no finished goods and little raw stock, yet work in progress is substantial because jobs spend most of their elapsed time waiting between operations. Every hour of that queue consumes money and floor space without producing progress, which is why delivery performance and cash performance improve together here. Upstream, the shop depends on material stockists, treatment houses and specialist subcontractors prepared to take small quantities at short notice. Those relationships are worth more than any price list, since a heat treatment supplier who fits you in on a Thursday decides whether a promised date survives contact with reality.

Estimates that were never true, and the ceiling people impose

Work is quoted from estimated times, and when those estimates come from optimism rather than recorded history the shop sells jobs below cost without ever learning which ones. The second recurring failure is expediting: a hot job jumps the queue, delaying several others, each of which becomes hot in turn until the sequence is decided entirely by volume of complaint. Growth is bounded by skilled people rather than machines, because an unstaffed machine produces nothing and capable setters take years to develop. Queueing imposes its own ceiling, since loading closer to full capacity stretches lead times sharply, so chasing utilisation quietly destroys a delivery reputation.

Frequently asked questions

Why does a busier job shop deliver more slowly?

Because jobs arrive irregularly and take varying amounts of time, so machines cannot be loaded to the brim without work piling up in front of them. As load rises, waiting time grows much faster than the extra work would suggest, and it grows worst where the mix is most variable. The practical implication is that a shop wanting dependable dates has to run with deliberate slack at its busiest resources, and recover the cost through better pricing rather than through fuller machines.

How should a job shop set its charge-out rates?

From the actual cost of providing each capability, including the machine, the space, the tooling and the person, divided by the hours it can realistically be sold rather than the hours it exists. Shops that divide by theoretical availability underprice systematically, because setup, maintenance, quoting and idle time are all real. Rates should differ by work centre, since a large machining centre and a manual bandsaw plainly do not cost the same to keep available for a customer who may or may not need them.

What data is worth collecting first in a shop like this?

Actual time per operation against the estimate, recorded by job. Nothing else changes decisions as quickly, because it exposes which work is priced wrongly, which processes are slower than assumed, and which customers absorb disproportionate effort. Collecting it needs to be light enough that operators actually do it, so start with start and finish times at each operation rather than an elaborate scheme that produces beautiful data for the first month and nothing afterwards.