Mr. Big, the general manager, strode into the engineering department. "Say, Sparky," he said, adopting his collegial I-need-something tone, "I'm doing a financial plan for next year and I need to know how many projects engineering can do." Sparky, the crusty old VP of engineering, replied, "Exactly what kind of projects are we talking about?"
"There you go again," snorted Mr. Big. "You engineers always answer a question with a question. Can't you just give me a straight answer?"
"Different projects require different amounts of effort," sputtered Sparky. Mr. Big replied, "If we did five projects last year and we improve each year, I'll plan on seven next year." Sparky choked and exclaimed, "That's loading us to 140% capacity!"
"Sparky, old boy, many years ago I learned that you must never give engineers any idle time. They'll either waste it playing rock-paper-scissors, or endlessly diddle with an already adequate design. An engineering department must be overloaded to function efficiently."
The belief that overloads make engineering more efficient is widespread almost everywhere except in engineering. The hundreds of companies that I've surveyed average 98.5% capacity utilization in their engineering departments. This virtually guarantees that most projects will be completed late.
Why? Because engineering is not a deterministic process. Since there is variation in both the start times and the durations of engineering tasks, we'll see very large work queues if we load to 100% utilization. Consider a factory. No competent manufacturing manager would load a factory over 90% utilization. That's because there's enough uncertainty in manufacturing to make good delivery performance impossible at this loading level. Yet, compared to engineering, manufacturing is a highly repetitive activity with little variation.
Logically, we should load engineering to an even lower level of utilization than manufacturing. High utilization rates simply guarantee that resources will be blocked when they are needed most. In fact, queues are actually 10 times longer at 99% utilization than they are at 90%.
Nevertheless, some managers assume that backlogs make engineers efficient. In practice, large work queues almost never improve the efficiency of the organization. Instead, they add overhead and create a large incentive to expedite work through the queue.
Not only does such expediting take time away from value-added work, but it is inherently a zero-sum game. As soon as one project manager bumps a job to the head of the queue, all the rest have an incentive to regain their original position. Expediting will never raise efficiency and usually lowers it.
How do we get out of this mess? Sadly, the misconception that excess capacity encourages waste is deeply ingrained in many managers. The best way to change their minds is to show them solid fact-based analysis. Measure the size of queues in your process over the last 12 months. Then have finance calculate the financial cost of these queues. Repeat this calculation using progressively higher levels of capacity.
The results are usually quite compelling. They'll clearly show that you need a certain amount of excess capacity to optimize profitability. The key is how you communicate to management. The language of management is dollars. If you don't speak this language, there is little chance you'll be heard.