What’s a “Good” Schedule?

What’s your objective when building a schedule?

• Do you want to get as much done with the fewest resources?
• Do you want to distribute the work load as evenly as possible?
• Do you want to minimize cost?
• Do you want to satisfy the preferences of your workers?
• Do you want to minimize the failed promise dates?
• Do you want to satisfy your organization’s unique constraints and operational peculiarities?
• —

Most of us would answer “yes” to more than one of these questions.  It turns out that what makes a “good” schedule depends on each organization’s culture and business practices.

After asking hundreds of schedulers “What makes a good schedule?” we found out that “good” schedules are those made by using a series of “good” decisions while building it.  If we regard the scheduling process as a series of decisions made step-by-step, then if each step represents a good decision, the result is generally a good schedule.

Well this doesn’t help much in defining a “good” schedule, but it does suggest an algorithmic way to build an effective scheduling tool.  What if we could score each decision we made as the schedule was being generated?  Maximizing the score of each decision is easier than scoring the final schedule after all decisions are made.

The theorist in me is beginning to squirm!  Local decisions are myopic; they can’t necessarily see the impact on the final result.  But the pragmatist in me recognizes that the way human schedulers produce “good” schedules is by making “good” local decisions.  So yielding to the pragmatic approach suggests that we find a way to score each local decision and pick the highest scoring option whenever there are alternatives to consider.  After many years of using it, this pragmatic approach has proven to be successful in practice.  We will leave optimization theory to another blog.

Now, back to fact that most of us would answer “yes” to more than one of the questions that we started with.  Imagine that you are in the process of building a schedule.   You are confronted with questions like “Which of several resource combinations should I choose?”    Here are some metrics you might want to consider:

• Which resources did I choose for similar activities? (Consistency)
• Should I use the same resource for this activity that I did for its predecessor? (Continuity)
• Which choice leaves the remaining resources most available? (Density)

If your decision is about choosing a start time, your considerations might be:

• Which start time leads to earliest completion? (Shortest)
• Am I trying for a Just-In-Time schedule? (Latest)
• Will a delay allow something else of higher priority to schedule? (Priority)

There are, of course, many different metrics that could be considered for scoring each decision.  But since most of us answered “yes” to several of the initial questions, we might want some combination or blend of these metrics to achieve a compromise that fits our objective.

So here is the recommendation.  Use several metrics and create a normalized blend that meets your organization’s style and preferences.  Keep adjusting the blend until the results are “good” most of the time.  You can even adjust the metrics and the blend for seasonal variations, changes in priorities or management philosophy, or even changes in the marketplace that you work in.

Applications that give you that much control are good scheduling tools.  How does yours stack up?