A positive constraint exists only to force the question of how it can be improved. A budget can be a great example of this. By taking stock of what we have and planning how to spend it, we can better see efficiencies, opportunities, and waste. Other examples include product development processes and engineering practices.

When people meet a positive constraint that limits them, they should feel compelled and empowered to improve it—not circumvent it.

A negative constraint exists only to limit variation. The policies HR creates to limit litigation exposure are a classic example; improvement is out of the question. Other examples are IT security policies, database development procedures, and bedtime rules. None of these are inherently negative—they could easily be positive—but are usually enforced with no thought to improvement.

A lack of constraints limits decision making to whim and guesswork. Without a budget, we cannot decide what to buy today because we can’t say whether we’ll run out of money tomorrow or next year. Without standardized development process, we can’t see wasteful action and remove it.

People avoid constraints or attempt to perfect them before implementation because they think constraints must be rigid to be effective. But, a good constraint is only a starting point. We should build what we can today, knowing that we can, we must, improve it tomorrow.

First, peer pressure from successes like Toyota, SAS, Nucor, and Microsoft. Secondly, our egos believe, while everyone else does it wrong, we are smarter. Finally, it is quick and has positive results in the short-term. Developing strong leadership throughout an organisation is much harder than telling HR to adjust a few salaries. What then can we do about it?

How other companies pay, regardless of viability, is important because good people are scarce resources that we compete fiercely for. If top performers get top pay at one company, but the same as every other knuckle dragger at another, they are likely to choose the former. We have three choices: pay everyone top wages, eliminate knuckle draggers, or create an environment that entices top people without top pay, like Genentech or SAS.

Recognise that successful companies measure and reward people in limited and thoughtful ways. Nucor simply shares profits evenly. Toyota rewards success with increased responsibility. Microsoft does have a classic performance pay system. But, according to WashTech and the New York Times, they are having problems with it.

The question of ego is easy to address at most companies. Simply asking what other systemic change a company has done well recently is enough to put the unsaid thought that “we’ll succeed where everyone else is failing” in its proper place.

The responses to the last two, ease and quick results, depend on strategy. If we are focused on short-term goals, planning to sell and run, we should use performance linked compensation.

If we are focused on long-term goals, the answer is simple. Remove the link between the two and see what happens. Most organisations have not tried this, having started with performance pay and struggling with it since. Given the complexity of performance pay, and the commonality of failure, we should approach it as a second step (if ever), not as the first.

At the very least, removing it will allow people to stop worrying about pay and get on with what matters: delighting our customers with value and ourselves with our work.

Recommended reading: Measuring and Managing Performance in Organizations by Robert Austin, Why Incentive Plans Cannot Work by Alfie Kohn, Peopleware: Productive Projects and Teams, 2nd Ed. by Tom DeMarco, and Unjust Deserts by Mary Poppendieck.Also see David Anderson’s HR Myths: Merit Based Pay, Divide and Conquer, and Performance Buckets