Batch size
lot size · small batches · batching work · work in batches
How much work is accumulated before it moves to the next step. Smaller batches move through a process faster and surface problems earlier, at the cost of more frequent handovers.
In practice
Invoicing monthly instead of weekly means a billing error takes up to a month to appear and applies to every invoice in the batch. The work is identical either way; the exposure is not.
The common mistake
Assuming large batches are more efficient because the setup cost is amortized. That is true of the step and false of the whole process, because the accumulated work is sitting idle while it waits.
Batch size is how much work piles up before it moves. It is the least discussed variable in most operations and one of the few that can be changed on a Monday without anyone's permission.
The intuition says bigger batches are efficient. There is a setup cost per batch, so a larger batch spreads it over more units. That is correct about the step and wrong about the process, and the gap between those two is where most of the waiting in a business lives.
The arithmetic that gets missed
Ten items processed as one batch of ten: every item waits for all ten to be done before any of them moves. The first item finished sits idle for the duration of the other nine.
The same ten processed one at a time: each moves as soon as it is ready. The last item finishes at roughly the same moment either way, but the average item finishes in about half the time, and the first one finishes almost immediately.
Total work is unchanged. Average time in the system halves. This is Little's law arriving from the other side — inventory in the system is what sets wait time, and a batch is inventory by another name.
The larger effect is on error
A batch is a bet that everything in it is correct, and the bet is settled all at once.
Process a month of invoices and discover an error, and the error is in a month of invoices. Process them weekly and it is in a week of them. The size of the mistake is set by the batch size, not by the mistake.
More important is the delay before you find out. In a large batch the feedback arrives once, at the end, long after the decision that caused the problem. In a small batch it arrives repeatedly and early, when the cause is still recent enough to identify. That is the real argument for small batches and it is a feedback loop argument rather than a throughput one.
Where the batches are hiding
Most are not called batches, which is why they persist.
Reporting cadence. A monthly report is a month-sized batch of information. Anything it would have told you is a month old on arrival.
Approval cycles. Work accumulating while it waits for a weekly meeting is a batch, and its size is set by the meeting schedule rather than by anything about the work.
Large projects. A six-month project delivered at the end is one batch of six months. Every assumption in it is tested simultaneously, at the point where changing any of them is most expensive.
Hiring in waves. Three people at once is a batch: the same onboarding mistakes are made three times before the first correction, and the team absorbs all of the disruption in one quarter.
What smaller batches cost
They are not free and the tradeoff is real.
More handovers, each with its own overhead and its own chance of something being dropped. More setup, where setup is genuinely expensive. More coordination, since things arrive continuously rather than at known moments.
Which gives the actual rule, and it is not "smaller is better". It is: reduce batch size until the handover cost starts to bite, and if the handover cost bites immediately, reduce the handover cost instead. That second half is where most of the gain is. The businesses that run small batches well did not simply choose to; they made the transition between steps cheap enough that frequent transitions stopped hurting.
Concept web
Open the full webQuestions
Why are smaller batches faster?
Because work in a large batch sits idle waiting for the rest of the batch. Total work is the same either way, but the average item's time in the system roughly halves, and the first item finishes almost immediately.
What is the downside of small batches?
More handovers, more setup and more coordination. The rule is to reduce batch size until handover cost starts to bite — and if it bites immediately, to reduce the handover cost rather than accepting large batches.
Where do hidden batches occur in a business?
Reporting cadence, approval cycles, long projects delivered at the end, and hiring in waves. None of them are called batches, which is why they persist.