Managing Utilization: Between Idle Time and Overload

Why companies swing between idle time and overload

Many small companies have two states. In one, the team works flat out, overtime is normal, quotes pile up, and customers ask about dates. In the other, work is lacking, and the owner accepts whatever comes, including orders with little contribution margin. A few weeks later, the team is flat out again. Both states cost money, just in different ways.

The article on managing demand instead of just serving it starts with demand: lead times, prices, and a target range for the order backlog. This guide starts with the work itself. It asks: How does a small company manage its utilization so that neither idle time nor overload occurs, and which utilization is the right goal in the first place? To answer it, we clarify the terms, introduce Kingman’s formula as a thinking model, review research findings, weigh two perspectives, and arrive at a utilization traffic light with fixed measures.

A team that is one hundred percent utilized has not a minute for the next urgent order, and that order always comes.

Terms: capacity, utilization, idle time, overload

Capacity is the working time a person or team has available in a period. Practical capacity is the part of it that is actually usable after breaks, coordination, and switching between tasks. The articles on what a process costs and capacity planning put it at about 80 percent of paid time. This assumption applies here too.

Utilization is the share of practical capacity that is occupied with work, with planned projects as well as the base load of day-to-day business. Idle time is practical capacity without meaningful work. Overload exists when more work arises than practical capacity can hold, so that work is left undone or shifts into overtime. It is important that utilization is not a measure of diligence but of the ratio of work to time.

The thinking model: Kingman’s formula

The guide to capacity planning describes the basic rule from queueing theory: at 80 percent utilization, a new task waits on average four times as long as its processing takes; at 95 percent, nineteen times as long. In 1961, the mathematician John Kingman showed that a second factor belongs to this rule, one often overlooked in companies: variability.

In simplified form, Kingman’s formula reads: waiting time ≈ variability × utilization factor × processing time. The utilization factor is utilization divided by the free remainder, so at 80 percent it is 0.8 ÷ 0.2 = 4. Variability describes how irregularly orders arrive and how differently long they take. With ordinary, random variability its value is one, and then exactly the figures from the capacity planning guide result. If orders arrive more evenly and are more similar, the value falls. If they come in bursts and differ greatly, it rises.

This leads to the most important idea of this guide. There is not one right utilization, but one that fits the variability. A team with even, similar orders can work at 90 percent and has the same waiting times as a team with ordinary variability at just over 80 percent. A team with rush orders, emergency calls, and very different tasks, on the other hand, gets into waiting times at 80 percent that another would only reach at almost 90 percent. An image from road traffic fits: a highway with evenly moving trucks tolerates dense traffic. The same highway with on-ramps, lane changes, and braking jams up with far fewer vehicles.

What studies show about people under load

The formula treats processing times as fixed. But people are not machines, and research shows that they respond to load. KC and Terwiesch (2009) analyzed operational data from two very different areas of a teaching hospital: patient transport and cardiothoracic surgery with about 3,000 patients. Under increasing load, staff worked faster. With 20 percent more load in transport, transport time fell by 30 seconds; with 10 percent more load in cardiothoracic surgery, length of stay fell by two days.

This acceleration did not last, however. Longer periods of high load, measured by the authors as overwork, slowed the work down again: one percent more overwork lengthened the stay in cardiothoracic surgery by six hours. Overwork also went along with lower quality. Ten percent more overwork was associated with a two percent higher risk of mortality, and under load, patients were discharged earlier.

That people react differently is shown by Ashkanani (2023) with 58,460 calls handled by 23 employees in the call center of a Kuwaiti hospital. On average, the team worked faster under load. Looked at individually, some sped up, others slowed down, and for some the pace only changed above a certain load. The average concealed these differences.

Run at high utilization or keep a buffer? Two perspectives

The first view says: every unused hour is paid for and earns nothing. High utilization spreads fixed costs over more orders and lowers the hourly rate the company must charge. On top of that comes the finding that people work faster under load. A buffer is therefore a convenience a small company cannot afford.

The second view says: the buffer is the condition for urgent work to find room, for deadlines to hold, and for errors to stay rare. Whoever plans close to full utilization pays with waiting times that rise steeply according to Kingman’s formula, and, according to the studies, with quality when the load persists. The guide to measuring on-time delivery shows the same logic from the other side with Little’s law: more orders open at once lengthen throughput time.

I consider the second view the more robust one, with an addition from the first. Idle time really is expensive, and short peaks are not a problem. The goal is therefore neither maximum utilization nor a generous buffer, but a target range whose upper limit depends on variability. Whoever reduces variability may plan higher. This shifts the question from the team’s discipline to the design of the processes.

The utilization traffic light: measure and decide in advance

Managing means deciding in advance what happens in which situation. The tool for this is called the utilization traffic light here. Every Monday it is read from the weekly plan for the current and the next two weeks: occupied hours from base load and projects, divided by practical capacity. Each color has fixed measures.

The limits depend on variability. A company with many rush orders sets the upper limit of green closer to 75 percent; one with even series orders closer to 90 percent. Besides the figure, two signals belong on the traffic light that show overload earlier than any calculation: overtime in recent weeks and the number of tasks that slipped from one week into the next. If both rise three weeks in a row, yellow applies even if the calculation shows green.

Which levers work against idle time and overload

According to Kingman’s formula, there are three adjusting screws: variability, utilization, and processing time. Four groups of levers follow, ordered by their lasting effect.

1. Reduce variability

   Split orders into more even packages, place recurring services in quiet weeks, bundle rush orders into fixed time slots, and limit the number of orders open at once. This allows higher utilization without longer waiting times.

2. Make capacity flexible

   A working-time account that balances peaks and troughs, employees who master several tasks, and partners for subcontracting agreed in advance. That way capacity follows the load instead of the other way around.

3. Steer demand

   State lead times openly, promise dates according to utilization, and actively approach existing customers during idle time. The rules for this are described in the article on managing demand.

4. Fill idle time sensibly

   A list of work that can wait: revising templates, maintenance, training, tidying the filing. Idle time thus becomes planned time for improvement.

In overload, it pays to look at the bottleneck first. Often not the whole team is overloaded, but one person or machine where work piles up. Additional hours only help there. How to find this point is described in the article on Goldratt’s theory of constraints. Recurring peaks over the year, for example before Christmas or in spring, are covered in the article on planning seasonal fluctuations.

Conclusion: how does a company manage its utilization?

We began by asking
a small company manages its utilization so that neither idle time nor overload occurs, and which utilization is the right goal. The right goal is a range, not a maximum. According to Kingman’s formula, its upper limit depends on variability: with ordinary variability, waiting times rise steeply from about 80 percent; with even work later, with erratic work earlier. The studies on work under load add that short peaks are absorbed, but sustained overload costs speed and quality.

In practice, this means: calculate utilization every Monday from base load and projects, keep a traffic light with measures set in advance, and reduce variability first, because that opens the range upward. The sober observation is: a company does not become capable because every hour is planned, but because the unplanned hour is available for what is urgent.

The next step

1. Calculate your team’s practical capacity for next week: paid hours times 0.8.
2. Enter base load and planned projects and calculate utilization.
3. Estimate whether your orders arrive more evenly or more erratically, and set the traffic light limits accordingly.
4. Write down three measures each for blue and red that apply without further discussion.
5. Create a can-wait list of tasks for quiet weeks.