Find where the work gets stuck.

Suppose a testing service receives 840 items during an eight-hour day. Customers complain that results arrive late, although management says the intake team can handle 960 items per day and the main processing equipment has even more capacity.

Before diagnosing the problem, define the promise. Is every item due within four hours of arrival? Is there a same-day deadline? Are urgent jobs handled separately? What proportion misses the target, and when does that happen?

Throughput, turnaround time and on-time delivery are different measures. Completing 840 items eventually does not establish that each customer's deadline was met.

For this exercise, focus on the time from receipt to delivery of a completed result. That makes both waiting and administrative release relevant, alongside processing speed.

A simple process might be:

Arrival → intake → processing → review → result delivery

For each stage, identify the input, the people or equipment required, the usable processing rate and the rule for passing work forward. Include queues between stages. Ask whether some jobs repeat a step, follow a different route or require a specialist.

That map helps you distinguish three problems. Work may arrive faster than a stage can handle it. Completed work may wait for a scheduled handoff. Or the final approval may depend on someone who is available only at certain times.

Those problems require different interventions. “Improve efficiency” does not yet tell you which one exists.

Map the process, compare demand with usable capacity and find the constraint before evaluating new equipment. Otherwise you may buy more capacity at a stage that is already waiting.

Assume 540 of the day's 840 items arrive evenly during the first three hours. Intake can process 120 per hour, continuously, with no initial backlog.

The early arrival rate is 180 per hour. Intake therefore accumulates 60 waiting items each hour, leaving a queue of 180 at the end of hour three.

The remaining 300 items arrive evenly over five hours. Intake now has 60 units of spare capacity per hour, so the accumulated queue takes three more hours to clear. It disappears at hour six under these assumptions.

Daily capacity of 960 exceeds demand of 840. Yet the morning arrivals wait because capacity is not available when they need it. For an item joining the back of the 180-item queue, the intake wait is about 1.5 hours at 120 per hour, before its own processing and subsequent steps.

The daily average hid a timing problem. Compare the same unit over the interval relevant to the customer's promise.

Suppose the main equipment can process 200 items per hour, but intake releases work to it only every two hours. An item that has completed intake may wait for the next release even when the equipment is idle.

Increasing equipment speed does not remove that release delay or the morning intake queue. A smaller transfer batch might reduce waiting, provided it does not create disproportionate setup work or compromise the process requirements.

Then inspect the final stages. Results may wait for review or for a scheduled email export. A faster middle stage offers little benefit if completed work still sits until the same afternoon release.

This is the distinction I would keep returning to in the interview: how much time does the job spend being worked on, and how much time does it spend waiting for permission, people or other jobs?

For the morning queue, explore whether arrivals can be spread across the day without making customers wait elsewhere. If the collection schedule is fixed, temporary intake support during the peak may be more appropriate.

For the two-hour transfer rule, examine why it exists. If it is only an administrative habit, more frequent releases may help. If each release creates a substantial setup burden, compare the saved waiting time with the extra work.

For a review bottleneck, test reviewer availability and the proportion of jobs needing specialist attention. Adding general intake staff will not resolve a decision that only a qualified reviewer can make.

Evaluate the whole journey after each change. Extra intake output can move the queue to review. Match the operating method to demand variation, product mix and the cost of holding work between stages.

Even when average demand is below capacity, unpredictable arrivals and variable job lengths can create queues. A team scheduled to be fully occupied on average has little ability to absorb a long job or a brief interruption.

Do not turn that observation into a universal “ideal utilisation” percentage. The appropriate buffer depends on variability, service promises, the cost of delay and the cost of capacity. A predictable production run and an emergency-response service need very different arrangements.

Use the case facts to assess the trade-off. If the last 5% of capacity utilisation produces a large increase in missed deadlines, it may be expensive rather than efficient.

For this service, a sensible initial recommendation is to address peak intake and the release rule before purchasing faster processing equipment. State what you would measure during the trial: time from arrival to delivered result, performance during the morning peak, errors or repeat work, and additional operating cost.

Record stage-level waits as well. They help explain whether an improvement removed the constraint or moved it downstream. Keep urgent and routine work visible separately so a better average does not conceal deterioration for one group.

If the trial meets the service target at acceptable cost, expand it. If the queue moves to review, update the diagnosis and compare the next intervention. Operations improvement is a sequence of evidence-based decisions, not a one-time declaration of the bottleneck.

Try QSR Drive-Thru Slowdown to practise this approach in a different setting. Trace the customer from joining the queue to receiving the order, and explain which stage you would measure first. For a deeper look at the capacity denominator, continue with the capacity-utilisation framework.