Case Study: Improving a Scheduling Process

How Taxi Hochelaga Transformed Their Scheduling Process to Become an Industry Leader


Note: while Taxi Hochelaga is a ground-transport company, the same process detailed here is applicable to a wide variety of applications, including aviation, freight transport, and more.

Taxi Hochelaga knew they wanted to improve their scheduling process, but weren’t sure where to start.  They were scheduling an average of 700 trips per day (~200 vehicles) with a scheduling team of 5 people, plus input from dispatchers, and they were required to generate a new schedule each night between 8pm and 6am, when the first trips were dispatched.  

They were growing quickly, and having a hard time balancing their growth with important considerations like driver satisfaction, mostly due to the increasing pressure on their driver’s schedules (note: ‘crew’ and ‘drivers’ are used interchangeably throughout, as in this context both are applicable).  

They wanted to build a process that ensured they could handle their growth, while also providing all drivers with the same quality schedules.

Their scheduling team was made up of a group of experts who had been scheduling for years, and the scheduling process wouldn’t have been possible day after day without them.

Their initial plan was to increase their fleet size, giving them more flexibility with individual driver schedules, but they knew there were likely other options, and they were keen to make sure they investigated them before implementing changes.

Initial Assessment

After initially meeting Taxi Hochelaga, one thing was clear - they were motivated to make sure they had the best scheduling process available, and were willing to invest some time in making sure that happened.

Their data maturity was another priority, and we wanted to make sure that wasn’t a factor holding them back from improving.

The first step was to learn their current scheduling process, and how that schedule was operated.  A large part of the time required for this step is interpreting the intent and origins of particular details of the scheduling process.  

It’s particularly important to have a third party expert present for this part of the process, because often the origins of a step in the scheduling process are difficult to discover, and no longer make sense in the current context, but it’s difficult to identify this if you’re actually part of that process.

A third party can also help manage conflicting motivations.  It can be painful within any company to try and satisfy all objectives.  As a scheduler or operations manager, you have to deal with pressure from management to be resource-efficient and provide high-quality service, and at the same time must deal with concerns from your crew/drivers on the quality-of-life provided by their schedules.  These issues are present in almost every company, and things were no different at Taxi Hochelaga.

Bringing in an expert, and beginning to make the scheduling process more transparent, and programmatic, made it possible for schedulers to respond to both parties with concrete data, and also allows them to make measurable adjustments to improve throughout the process.

A good example of one of the discoveries in this process was that the scheduling team was first placing all recurring rides on the schedule first, and then solving the rest of the schedule.  It was done for good reason, as it was part of what made the process possible during the limited time window, but in practice it meant a decrease in efficiency because these constraints were introduced.

Making the Scheduling Process Data-Driven

The second step in the assessment was looking at historical scheduling data, and combining that with the knowledge extracted in the first interviews and the understanding of the scheduling process.

What this really means is taking a large amount of historical data, and comparing it to some initial results from an automated solver, which has incorporated the scheduling rules that were extracted in the first step.

Besides starting the detect patterns in the historical scheduling data, this process usually creates further conversations with the scheduling team.  Lots of comments like “this looks wrong on the automated schedule” and “this isn’t how we would do things” get pushed to “well why does it look wrong” or “why don’t you do it this way” (a classic case of asking “why” 3 times).

These conversations are extremely important, as they probe even deeper into the origins and decisions in the current scheduling process, which are sometimes uncomfortable (did you realize you were subconsciously favouring this set of crew/drivers?  That you’re less efficient with these types of routes?).  The key point, however, is that these conversations are the result of data-driven observations, which makes it much easier to have a meaningful discussion.

A good example of one of the discoveries in this part of the process was that a certain subset of the drivers was being favoured over others, for an understandable reason: because their operating rules were more flexible, they were easier to schedule.  But the tendency towards these types of drivers wasn’t a conscious decision, and this type of scheduling had a detrimental effect on the other crew.  This discovery was a result of an examination of the data, and wouldn’t have surfaced without the data-driven assessment.

Counter-Intuitive Discoveries

The primary goals of Taxi Hochelaga, in this case, were to make sure all the trips were fulfilled, and then to distribute them as evenly as possible among their contracted drivers, so that minimum revenue targets were reached.   

Making sure all trips were fulfilled was the primary objective, and prior to the assessment process it was assumed that to do this, they would have to increase their fleet size.

Instead, as a result of the analysis, and the introduction of an advanced solver to aid their scheduling team, they were able to make significant, planned changes in how they scheduled their trips.  By recognizing and adjusting things like removing the preference for certain types of drivers (by using a programmatic approach and new software), they were actually able to decrease their fleet size by over 10%, while taking on an additional 15-20% of work, completely the opposite of what they were planning.

By bringing in experts, making the process data-driven, and being open to sometimes painful discoveries about their scheduling process, Taxi Hochelaga was able to bring their scheduling process to the cutting-edge, and find the optimal balance between crew satisfaction, customer service reliability and company business objectives.