Lean for job shops is the practice of applying flow, quick changeover, and visual scheduling in a custom, make-to-order shop where every job is different and much of the work may never repeat. Because you cannot forecast or level what you have not quoted yet, job-shop lean leans on controlling work-in-process and making the schedule visible rather than on takt and long runs.
Job shops get told, more than any other kind of plant, that lean is not for them. Every job is a one-off, the argument goes, so there is nothing to standardize and nothing to flow. That is half true and it is the wrong half. The parts change; the process of making parts does not. A job shop still cuts, mills, welds, inspects, and ships, and every one of those steps has waste in it. This is the job-shop version of the general lean playbook the same waste hunt, aimed at where custom work actually loses time.
What is lean for a job shop?
Lean for a job shop means removing waste from the process of building custom work, even though the products themselves keep changing. The customer value definition holds: a customer pays for the finished part, not for the days it spent sitting in a queue between the saw and the mill, not for the hour a machinist spent hunting for the right fixture, and not for the rework caused by an ambiguous drawing. Those are the wastes, and in a job shop they are enormous.
The reframe that unlocks job-shop lean is to stop looking at the part and start looking at the job's journey. Follow one order from the moment it lands to the moment it ships, and clock how much of that elapsed time the part was actually being worked versus waiting. In most job shops the part is being cut, machined, or welded for a small fraction of the time it is in the building. The rest is queue. That queue is the target.
Can lean even work when every job is different?
Yes, because the waste is in the flow between operations, and the flow repeats even when the parts do not. A job shop is usually laid out by function: all the saws in one area, all the mills in another, welding in a third, inspection in a fourth. Every job zig-zags through those departments in its own order, and at each handoff it joins a queue behind other jobs. The part is unique; the traffic jam is identical every time.
Lean attacks that traffic jam with tools that do not care whether the part repeats. Cutting setup time helps every job. Controlling how many jobs are on the floor at once helps every job. Making the schedule visible helps every job. None of these require you to build the same thing twice, they require you to build anything through a shop with less waiting, less searching, and less rework. That is why the job-shop objection is backwards: high variety is exactly the condition where flow discipline pays the most.
Job shops are also the backbone of U.S. metalworking, not a fringe case, a large, mostly small-shop sector building parts on a job or order basis.
| U.S. machine shops (NAICS 332710), a classic job-shop sector | Figure | Primary source |
|---|---|---|
| Estimated industry employment in machine shops | ~226,000 | U.S. Bureau of Labor Statistics (OEWS) |
| Work defined by the Census Bureau as machining parts on a job or order basis, generally low volume | By definition | U.S. Census Bureau (NAICS 332710) |
Where does a job shop lose time?
A job shop loses time in four places, and none of them is the actual machining. First, queue time between departments, which usually dwarfs everything else. Second, setup, because runs are short, often a lot size of one, the changeover can take longer than the cut, so setup is a huge share of the labor. Third, searching: for the fixture, the program, the material cert, the right drawing revision, the traveler that walked off with the last job. Fourth, rework, driven by ambiguous drawings, missing information at the machine, and quality problems found at final inspection instead of at the operation that caused them.
A simple value stream map of a representative job makes this undeniable. Draw the job's path, write the touch time and the wait time at each step, and total them. The ratio of value-added time to total lead time in a first-time job-shop map is often shockingly low, and it points every improvement that follows.
How do cells help a custom shop?
Cells help even when parts do not repeat, because most job shops still have part families hiding in their mix, brackets, shafts, plates, weldments, that share a rough routing. Grouping the machines those families need into a cell arranged in process sequence, collapses the department-to-department queue for anything that fits the family. A job that used to wait in four queues now flows one small loop.
Not everything fits a cell, and that is fine. Job shops usually end up with a hybrid: a couple of cells for the recurring families, plus the functional departments for the true one-offs and the odd geometry. The cells drain the volume that queues worst, which frees up the functional area for the genuinely unique work. You do not have to cellularize the whole shop to get most of the benefit.
How do you schedule a job shop visually?
You cap the work-in-process and make the queue visible, rather than trying to compute a perfect schedule. Job shops are notorious for overloading the floor: sales releases every job the moment it is booked, the shop fills with work-in-process, and every job slows down because it is stuck behind every other job. The counterintuitive fix is to release fewer jobs to the floor at once. Holding a job in a pre-release queue and letting it onto the floor only when the shop has capacity, a work-in-process cap, sometimes called CONWIP, shortens lead time for everything, because jobs on the floor stop fighting each other.
Then you make status visible. A physical or digital job board that shows every active job, its current operation, and whether it is on track does more for a job shop than most scheduling software, because it turns the daily question "where is my order" into something anyone can answer at a glance. This is visual management aimed at the schedule, and it pairs naturally with a short daily huddle at the board, the core of a daily management routine.
How does quick changeover pay off in a job shop?
Quick changeover pays off more in a job shop than almost anywhere, because your lot size is often one and your setup can exceed your run. When you machine a single part, every minute of setup is pure overhead spread over that one piece, there is no long run to dilute it. Cut the setup and you directly cut the cost and the lead time of every job.
The SMED method applies cleanly: pre-set tools in holders offline, prove out the program before the machine is free, stage the fixture and material at the machine ahead of time, and standardize the teardown-and-build routine so it does not depend on which setter is on. Capturing that routine as standard work matters double in a job shop, because setup knowledge is the classic tribal knowledge that lives in one veteran's head and walks out the door at retirement.
What is a job-shop lean sequence?
Job shops get the fastest results by attacking queue and setup before anything else. Work this order.
- Map one representative job. Follow a typical order door to door and record touch time versus wait time at each step. The value-added ratio is your baseline and your wake-up call.
- Cap work-in-process on the floor. Stop releasing every booked job immediately. Hold jobs in a pre-release queue and let them onto the floor only when there is capacity. Lead time drops before you change anything physical.
- Stand up a visual job board. Show every active job, its current operation, and its status, so status questions stop pulling people off the work.
- Attack the worst setup with SMED. Pick the machine that changes over most, move setup work offline, and standardize the routine.
- Cell your biggest part family. Find the recurring family that queues worst and put its machines in a small flow loop, leaving true one-offs in the functional area.
- Run a daily huddle at the board. A short standing meeting on late jobs and blockers turns the board into a habit and catches slips the day they happen.
Job-shop lean and high-mix, low-volume lean share most of these moves; the difference is that a job shop's work may never repeat, so it relies on visual scheduling and WIP control where an HMLV plant can also lean on repeating intervals.
Where does real-time data fit in a job shop?
Job shops run on questions paper cannot answer fast enough: where is this job, is it late, what did that setup actually cost. Travelers and end-of-shift tallies are always behind, so the shop manages by walking the floor and asking, which is itself a form of waste. Digitizing job tracking and capture at the machine, so operation status and real setup and run times land in the system as they happen, turns the visual board into live truth and the quote-versus-actual comparison into fact, without ripping out the ERP or estimating system a shop already runs. When CLS moved production logging off paper, problems surfaced during the shift instead of the next morning. See how the connective layer fits on our features overview and start by mapping one job.