Optimization and optimization software can be confusing if you can't distinguish the different types.
The word ‘optimization’ gets used in transport a lot, whether in the field of ground transport, air transport, or something else. It’s a nice word - everyone wants to ‘optimize’, particularly when there is more competition than ever, and companies are fighting for margins.
But it’s also ambiguous - everyone has a slightly different perception of what it really means, and this ambiguity results in a lot of miscommunication.
We’re going to try and make it a bit easier, and talk about two distinctly different types of optimization - process optimization and mathematical optimization - and then in future posts we will get into details about each.
Process optimization is typically defined something like:
“the discipline of adjusting a process so as to optimize some specified set of parameters without violating some constraint”.
Process optimization is what most people have in mind when talking about ‘optimization’, in software or otherwise.
That’s likely because it’s used frequently by creators of operations software, who say things like “optimize your scheduling” or “optimize your fleet”. It’s sometimes hard to distinguish the optimization types, as some operations software will incorporate small mathematical optimization components, but in general, when you hear software providers talking about ‘optimization’, they’re talking about improving your processes.
They usually do this by reducing the number of steps required to complete a job; automating some tasks that were previously done manually, or coordinating tasks between different departments. Moving software to the cloud, providing mobile apps, and other additions to a software product are sometimes used to do this.
Integrating quoting software with booking software, or scheduling tools with the dispatching software are also pretty common.
The end result is usually the same, if they deliver on their claim - they optimize your process by reducing the number of steps to complete a given task, and/or reduce the time required to complete a given task.
Mathematical optimization has some commonalities with process optimization, but is generally quite different.
It’s often defined generally as follows:
“optimization includes finding "best available" values of some objective function given a defined domain (or input) and constraints”
What this means in practice is turning the scheduling process into a mathematical function (objective function) to actually solve the scheduling or routing problem for a given objective (profitability, coverage, etc.).
In the ideal scenario, this means that instead of moving around routes, schedules, destinations and pickups, often in someone’s head, on a Gantt chart, or in a piece of software (it used to be a whiteboard), a single mathematical function solves your scheduling for you at the press of a button, and all the normal constraints (operating rules) you would consider in the scheduling process are taken into account.
Given that it’s a mathematical function, you can ‘optimize’ the solution for a given constraint - whether it be cost, revenue, profit, etc. Ideally you can even optimize against multiple constraints, and perhaps add in optimization criteria like customer satisfaction, or driver/crew work-life balance.
There are a wide range of applications for mathematical optimization - scheduling crew, scheduling vehicles, combining both, developing your pricing, and more.
The results of mathematical optimization can be the same as process optimization - if you’ve successfully implemented mathematical optimization, the process of solving whatever problem you’re solving should be much easier - you should be given a solution. Automation can be a large component of both.
But there are many more results of mathematical optimization that aren’t possible or expected results in process optimization.
As an example, let’s say you’re trying to increase the profit you generate at a flight charter company; it may actually make more sense to reduce the amount of business you do, outsourcing the least-profitable trips and routes. Process improvement would suggest you improve how you schedule that extra business; mathematical optimization would show you which routes to outsource to become more profitable.
There are also many different flavours of mathematical optimization, and the results of each can vary, yielding varied experiences for those who have tried to implement mathematical optimization. The techniques that are used matter for the types of problems that can be solved, and so you should consult experts when thinking about exploring some sort of optimization.
That said, almost any method of mathematical optimization that can be successfully implemented (very important) is a good idea, because of the quantitative insight it can give you into your business, and the control it gives you over tweaking your model. Implementing mathematical optimization can let you move from being reactive to testing what-if scenarios, and proactively improving your operations.