When delivery dates slip, the instinct is to blame the cutter. It is the biggest, most expensive, most visible piece of equipment in the shop, so it feels like the thing that must be limiting output.
It usually is not. In most laser shops the machine is busy for a fraction of the hours it is switched on, and the constraint sits somewhere much less impressive: the bench where parts are cleaned, the person who nests files, the packing table, or the fact that nobody can start a job until a customer approves a proof.
Buying a faster cutter in that situation produces a familiar and expensive result: the pile moves. The machine finishes sooner, and the same pile of unfinished work now sits in front of the finishing bench, one step further downstream. Capacity went up in one place, output did not change at all.
This article is about finding the step that actually limits output, and about the uncomfortable fact that in a laser shop it is almost never the laser.
Before anything can be planned, the shop's real time map has to be drawn. Almost no shop has one, and almost every shop is surprised by it.
For a typical job, the machine is involved in one of these steps — and the rest happen around it:
Read down the "machine running" column: the machine is working in two rows out of ten. That is not a criticism of the machine — it is the nature of a job shop. But it explains why machine speed is such a weak lever on delivery performance, and why "our machine cuts at X" tells a customer almost nothing about when their order will arrive.
The map also shows something more useful: most of the steps are manual, and manual steps vary. Variation, not slowness, is what creates queues — a bench that takes a predictable hour is easy to plan around; a bench that takes twenty minutes or three hours depending on the job is where schedules die.
The bottleneck is not the step that feels hardest or the one that costs most. It is the step that everything else is waiting for.
Three symptoms identify it, and they are all visible without any software:
Work accumulates in front of it. There is a pile, a queue, a list, or a stack of pallets at that station, and it does not clear. Other stations look tidy; this one always has something waiting.
It runs at its maximum while others idle. The finishing bench is working through lunch while the machine sits unused after a batch; or the nesting desk is the reason the machine was idle this morning.
Speeding it up speeds up the shop. This is the test that matters. If an extra hour at that step produces an extra hour of shipped output, it is the constraint. If it produces nothing but a bigger pile at the next step, the constraint is downstream and you have found a symptom, not the cause.
That third test is the reason capacity planning is iterative rather than a one-off calculation: fix the bottleneck, and a different step becomes the constraint. The shop that does not re-measure after a change will keep investing in the step it already fixed.
You do not need a planning system to find your constraint. You need one week of honest observation.
Count the pile, daily. Every morning, walk the shop and write down how many jobs are waiting at each station. Six lines a day, five days. The station whose number keeps growing is the constraint; the station whose number oscillates is fine.
Follow three real orders from enquiry to dispatch. Pick a simple job, a normal one and an awkward one. Write down every time the job stopped and what it was waiting for. The pattern across three orders is usually unambiguous, and it is often a wait nobody had noticed — waiting for a proof approval, waiting for material, waiting for a fixture, waiting for the person who knows how to nest that particular product.
Ask the operators what they wait for. People know exactly what annoys them and they are almost always right. The question to ask is not "what is slow" but "what do you stand around waiting for" — those are different answers, and the second one is the useful one.
Time one batch at each station, once. Not to build a model, just to see the ratio. Shops are consistently surprised to discover that the station they consider trivial consumes as much time as the one they consider the main event.
The output of that week is a single page: a list of steps, the typical wait in front of each, and one sentence on why. That page is worth more than any software licence, because it is specific to the shop and it can be argued with.
In laser shops, the constraint tends to live in one of five places. Each has a characteristic signature and a characteristic fix — and the fix is rarely "buy something".
The pattern is worth stating plainly: three of the five fixes cost nothing but discipline. Templates, standard layouts, a checklist and a named owner are cheaper than any machine and usually produce more output.
If there is one prediction worth making about a laser shop without seeing it, it is that the finishing bench is the constraint — or will be as soon as cutting gets faster.
The reason is structural. Cutting is fast, repeatable and unattended once running. Finishing is manual, variable and hard to speed up: cleaning edges, removing masking, sanding, polishing, assembling, checking. A machine can cut a sheet of parts in the time it takes one person to finish a fraction of them.
So the arithmetic of adding a second cutter is often: more parts, same bench, longer lead time. The shop looks busier, ships the same amount, and wonders why revenue did not move.
The fixes are unglamorous and effective:
For shops doing high-mix work — which is most of them — the machine's real capacity limit is not cutting speed but how long it takes to change from one job to the next.
A machine that cuts for twenty minutes and takes thirty minutes to set up has a capacity that is dominated by setup. Doubling cutting speed changes almost nothing; halving setup changes everything.
The principle is the one every lean programme teaches and every small shop ignores: separate what must happen while the machine is stopped from what can happen while it is running. Files, material, fixtures and decisions can all be prepared while the previous job is still cutting. Most shops do them after the machine stops, which is why the machine appears to be the constraint when it is really waiting for its operator.
Before a second machine, the honest question is whether the first one can run more hours — because the fixed cost is already paid, which makes those hours the cheapest capacity available (this is the same argument as in the cost-per-part article, seen from the other side).
Unattended running deserves a specific caution in a laser shop: it is only appropriate for work that is proven, on material that will not move, with fire safety provisions that actually work. A shop that leaves an unproven job running overnight is not adding capacity; it is adding risk.
Outsourcing is usually treated as all-or-nothing — either the shop does the work or it does not. The more useful approach is to outsource the operation that is the constraint, and keep in-house the operations where the shop adds value.
If the constraint is finishing, sending out edge polishing buys capacity immediately without capital. If the constraint is the machine on a particular material, sending out that material preserves the customer relationship and the schedule. If the constraint is a process the shop is bad at, outsourcing it is cheaper than learning it badly.
What not to outsource is the step that makes the shop distinctive — the design work, the material knowledge, the fit with the customer's process. Outsourcing the constraint is capacity planning; outsourcing the identity is a different decision entirely.
After a change, one number tells you whether it worked: the queue in front of the constraint. If it is shorter and shipping is the same or higher, the change worked. If the queue moved elsewhere, you have found the new constraint — which is progress, not failure, but only if you notice it.
Two metrics to distrust:
Machine utilisation on its own. A machine at high utilisation may be the constraint, or may be working through a backlog caused by something else. Utilisation tells you the machine is busy; it does not tell you the shop is productive.
Output without lead time. Shipping more while lead times grow is not improvement; it is a shop accumulating promises it cannot keep.
The pair that matters is output and lead time together, measured at the same time. A shop that ships the same volume in half the time has doubled its real capacity without buying anything.
Capacity in a laser shop is not set by the cutter's speed. It is set by the step everything else waits for — and in most shops that step is manual, unglamorous, and cheap to improve once it has been identified.
The method requires no software: draw the time map, count the pile for a week, follow three orders, ask what people wait for. Then fix the constraint, re-measure, and expect the constraint to move.
The shops that grow smoothly are not the ones with the fastest machines. They are the ones that know, at any moment, which bench the work is piling up in front of — and treat that bench as the thing worth investing in.
Hot News2026-11-16
2026-11-12
2026-11-10
2026-11-09
2026-11-06
2026-11-05