Throughput management: protecting the rate of saleable output

Counting only what the plant can actually sell

Throughput should be measured as good units completed and accepted, not as units started, machine hours run or stock moved between departments. The distinction matters because the three commonly diverge: a plant can raise hours run while good output falls, if the extra hours produce material that later fails inspection or sits half-finished. Define the count point once, at the step where the product becomes saleable, and take it from the same source every day. Where different departments each report their own figure, senior meetings degenerate into reconciling numbers instead of dealing with the cause of the shortfall.

Local efficiency is the most common enemy of plant output

Every department measured on its own utilisation has an incentive to keep running, which produces long runs of whatever is easiest, stock in front of the next process, and machines busy on work nobody needs this month. None of that reaches the invoice. The corrective is to measure non-constraint areas on schedule adherence and on keeping the constraint supplied, and to reserve rate measures for the resource that actually sets output. Expect resistance, because efficiency measures are comfortable and long established. Changing what is reported to senior management is usually the only thing that changes behaviour on the floor.

The hourly check that catches a shortfall while it is recoverable

Compare planned against actual output at short intervals through the shift, at the pacing resource, recorded by the team doing the work. When the count falls behind, the reason is written down at the time and while it is still known. This produces two things: a supervisor who intervenes within the hour rather than explaining the next morning, and a cause list that is worth analysing because it was captured before memory smoothed it. The scheme fails wherever the reaction to a recorded loss is blame, since the reasons then become uniformly vague and the record loses all diagnostic value.

Deciding what to sacrifice when the rate is short

A plant behind on rate has a limited menu: run through a break with relief cover, extend the shift, defer a changeover by extending the current run, move work to a slower alternative route, or accept the miss and tell the customer. Each option has a cost that lands somewhere else — overtime spend, delayed products, a longer changeover later, higher scrap on the alternative route. Someone has to be authorised to spend that cost during the shift, otherwise the decision waits for the morning meeting and the opportunity is gone. Set the authority level in advance and record which lever was used.

Reading the weekly pattern rather than reacting to yesterday

Daily output moves around for reasons that are mostly noise, and a management team that reacts to every dip generates churn without improving anything. Look at the run of days: whether shortfalls cluster on particular shifts, particular products, particular days of the week or after specific changeovers. Clustering points at a cause with a fix, while scattered variation points at general process instability that needs a different response. Bring the loss reasons captured hourly to this review, because the pattern in the reasons is usually clearer than the pattern in the totals.

Frequently asked questions

Is higher throughput always the right goal?

No. Throughput is worth increasing only where the extra output can be sold or where it relieves a genuine constraint. Pushing rate on a resource that is not limiting the plant simply builds stock and consumes material and labour early. When demand is below capacity, the operational goal shifts to reducing cost and holding delivery performance, and pursuing rate for its own sake in that situation makes the numbers look better while the cash position worsens.

How does throughput relate to the plant's efficiency figures?

They answer different questions and often move in opposite directions. Efficiency asks how well a resource used its available hours; throughput asks how much saleable output the plant produced. A department can improve its efficiency by running long batches of easy work while the plant's output of what customers ordered falls. Keep both, but be explicit that efficiency is diagnostic for a local area and throughput is the figure the business is actually judged on.

Who should own the daily throughput number?

The production manager owns delivery of it, but the number itself should be produced from a single agreed source that nobody in the reporting line can adjust. Independence matters because a figure the owner can edit will drift towards whatever avoids an uncomfortable conversation. Where the data comes from manual counts, spot-check them against stock movements or dispatch records regularly enough that everyone knows the check happens.