Choosing a batch size is choosing how much you are willing to be wrong by. A large batch spreads setup cost across many units and locks in the assumption that the design, the demand, and the process are all correct. A small batch keeps the option to change, and pays for it in setup time.
Most buyers think about batch size as a purchasing decision about quantity. It is actually a decision about how expensive it will be to change your mind, and that cost is decided at the moment the batch is released, not later when the change is needed.
Manufacturing businesses that run mixed work live with this trade every day, which is why they tend to be sceptical of round numbers chosen for administrative convenience. On work such as titanium components, where material cost dominates the decision, the choice is narrower than the formula suggests.
The invoice for a change is rarely the largest cost. When a design changes after a batch is released, the visible charges are rework, scrap, and a new setup. The invisible costs are larger: capacity consumed by rework that was planned for new work, learning that has to be repeated, and the delivery risk created by inserting an unplanned job into a loaded schedule.
There is also a cost on the buyer's side that never appears anywhere. A change consumes the supplier's willingness to absorb the next surprise, because goodwill is finite and is spent on exceptions.
The practical implication is that change cost is not linear in batch size, it is step-shaped. Below a certain quantity, a change can be absorbed inside the existing run. Above it, the change requires a new run, and the cost jumps.
The textbook batch size minimises the sum of setup cost and holding cost. It assumes a stable design, predictable demand, and a known process. In custom fabrication, all three assumptions are weak, and the formula produces a number that is too large.
It also ignores the cost of the queue. A large batch occupies a machine for longer, which delays every job behind it. That delay is not in the formula, but it is real, and it is paid by other orders.
A better starting point is the smallest batch that justifies the setup for the operation that is hardest to change. That operation is usually the one with the longest setup or the tightest tolerance, and it sets the floor for everything downstream.
The most useful tool for managing change cost is not a batch size at all. It is a schedule of freezes: periods during which the specification will not change, followed by windows in which changes are collected and released together.
A freeze converts scattered changes into one event. Instead of five amendments arriving on five days, each disrupting a run, one release absorbs all five before the next batch starts. The saving is not in the change itself but in the number of times the process has to restart.
Staging works alongside it. A small first batch validates the design; a larger second batch exploits the validated setup. The first batch is deliberately uneconomic, and it buys information that makes the second batch safe.
Large batches are not a mistake in every case. They are correct when the design is mature, the demand is contracted, the process is proven, and the holding cost is low relative to the setup cost.
They are also correct when the part is simple and the setup is dominant, because there is little to learn from a small run. A plain bracket with one bend and two holes does not benefit from a proving batch; the process is understood before it starts.
The mistake is applying the same logic to a complex assembly with cosmetic requirements and several welded joints. There, the learning is the expensive part, and a large first batch converts a cheap lesson into an expensive one.
These five questions usually produce a batch size that is smaller than the buyer expected and larger than the supplier would choose alone, which is a reasonable place to land.
A change made before material is cut costs almost nothing. The same change after cutting costs the material. After forming it costs the tooling time. After welding it costs the inspection and possibly the whole part.
This progression is why the timing of a decision matters more than its content. Buyers who want flexibility should ask when the last cheap moment is, and then make their decisions before it. A supplier who can name that moment is a supplier who understands the process.
Beyond that moment, the honest answer is that the change is possible but expensive, and the schedule will move. A supplier who says yes to every late change without adjusting the date is quietly planning to absorb the cost somewhere else, usually in another customer's order.
In a shop running many different parts, batch size also determines how well the schedule can be packed. Small batches fit into gaps; large batches require clean windows. A mix of both is usually better than either alone, because gaps exist whether or not there is work to fill them.
This is one reason mixed shops can be competitive on small quantities despite higher unit setup. The setup is real, but the alternative is an empty window, and an empty window costs the same as a setup that produces something.
Buyers of fabricated parts will find the trade differs by material, because cost and cutting parameters both push toward smaller, better-proven batches. The company background, the customisation process, the range under other parts and stainless exhaust systems, and the dealership terms are worth reading before the batch question is settled at quotation.
