How long does it take your planners to create a good plan for a single day. Four hours? 12? 24?
A couple of weeks ago, I unveiled a customer’s new Quintiq optimizer. The customer’s puzzle was planning staff and equipment at a major airport. It was an exciting day because our team had spent weeks configuring the optimizer and now finally the customer was going to see the results of our labor.
These airport planners schedule staff by assigning them to specific time slots every week. Each slot requires employees with particular skills to be at precise locations, often with equipment. When taking all staff and all slots into account there are over 100,000 possible combinations. In addition, the planners need to take into account numerous rules and constraints which make the plan even more complex.
The complexity reminds me of a challenge that we use in Quintiq’s internal recruitment process. New applicants plan eleven employees for a two week period. A schedule is considered of reasonable quality if scores at least 90% of the optimal solution. Only one in ten applicants are able to create a 90% or better schedule. Even more telling, every one of these top performing applicants needs between four and eight attempts to produce even a reasonable solution – one that still leaves up to 10% room for improvement.
So let’s say that your planners are as good or better than these top applicants. Now consider that even the world’s best planner could not possibly check 100,000 possible combinations in a year, let alone every week.
With so much involved in creating a schedule, planners need to cut corners somewhere. Some of the most common shortcuts are to simplify the puzzle, use standard patterns, or reduce secondary tasks. As we’ve seen in previous blog posts – and countless real-life examples – these shortcuts have a detrimental effects on the quality of the plan, and therefore on the business results.
The best way to conquer and even take advantage the full complexity of the puzzle (and thus achieve the best result for your business) is to optimize the way in which planning is done. That means supporting your planners with the right planning and optimization technology. A good optimization algorithm is capable of taking into account all relevant information (which we’ve seen is considerable), and use it to solve the most difficult puzzles. Just as importantly, to do it very quickly.
The customer I am visiting is also excited to see the new optimizer in action for the first time. They don’t have to wait long. Within minutes the optimizer has come up with a result. The planners take a look. The new result has improved on each of the KPIs; a great result! While the planners are impressed, something bothers them. They point at two tasks and say: “We never plan these two tasks on the same shift.”
“Why?” I ask.
“Because it creates unnecessary travel time.”
So, I encourage the planners to try and improve on the plan. They take up the challenge. However, after an hour, they can’t come up with something better. In spite of going against their long-held assumptions, the best plan is still the one that took technology only minutes to produce. I guess some patterns are meant to be broken.