As we move into a new decade, there is a lot of talk about process automation and what it can do.
Let’s look at process automation in last mile logistics solutions that use some type of an AI driven assignment process.
In a perfectly automated world, computers would be able to analyze logistics problems and come up with an immediate solution that meets all customer requirements and do it in a less expensive way than humans could do. But is this really possible given the day-to-day constraints of real-life logistics?
First, let’s agree that process automation is a good thing.
The automation worst case is no automation at all. For a product delivery or service organization this would be printing out work tickets and then assigning them to workers using nothing more than a map and a whiteboard.
A better case is having an application that does the work assignment where the user must enter the work tickets manually. It’s double data entry, but at least the assignment process is automated.
Even better is a case where the work could be exported into a spreadsheet and manually imported into the assignment application. Since almost every business application has a way to export data, this is usually the easiest way to achieve automated work assignments.
Way better is being able to press a button in the main business application and have the data sent to the assignment application with no extra work at all. This requires some level of process automation which may, or may not, be available in the business application.
But does it get any better that this?
If you are in the product delivery and/or pickup business, the answer is probably no.
Regardless of the ability of a logistics process optimization application, product still needs to be put on a truck with the final stop in the nose and the first stop in the tail. Reordering the stops to avoid traffic conditions usually means that the driver (and usually it is only the driver in the truck) must move boxes or pallets around in the truck to get to whatever needs to be delivered.
In most cases, the only reason to depart from the regular route is an interruption by a major event (like fire or flooding) or if some stops cannot be made because the delays throughout the day were long enough to cause missed customer time windows.
In the case of deliveries with simultaneous pickups (like furniture or appliance delivery with the pickup of the items being replaced), there is usually a helper in the truck along with the driver. But rerouting still means a lot of extra manual work moving things around.
Better real time automation starts to come into play where there are many deliveries, followed by many pickups, such as in an LTL application.
Effective automation can consider pickup requests that come in during the day and can plan the return route with pickups in such an order that considers real time conditions. It can also start the pickups before all the deliveries are complete if there is enough space to accommodate them.
The route planning process for service organizations depends on what the service technician must carry to the customer site. If the truck contains only a stock of commonly used parts, then technicians could be rerouted during the day. But if the technician needs to carry specific parts to specific customers, there can be some flexibility in the order that stops are made, but each technician must make all the stops that have been assigned.
The way to decide what level of automation is to start with a thorough business process review to document the current process and then use that as the starting point for a better (usually automated) process.
Besides the obvious questions of what this new process needs to do, there must also be a consideration of whether or not the data that affects logistics is maintained in the business system.
As an example, I once rode along on a truck making deliveries of cooking oil to fast food restaurants. The driver told me he had to skip the next stop on his list and come back later because the restaurant manager did not want the oil hose placed across the drive-up lane during the lunch time rush. The real issue is that this specific time window request was not maintained in the business system, and there was not even a way to record this. Obviously, quality route planning depends on knowing business constraints, and the constraints can be considered only if they are recorded somewhere.
Modern AI applications that use machine learning depend on learning from a very large number of observations. After observing millions of lane markings, cars can now tell if they are straying from their lane.
But if a stop is visited only a few times each month, and at different scheduled times, it might take years (or maybe decades) of observations to determine that the driver goes out of route when a certain stop is scheduled at a certain time of the day.
So, for almost all product based, and most service based, organizations, the way to achieve planning and operation efficiency is to use a route planning application that integrates well with an existing business application and can be used to create efficient routes.
Looking for an efficient and cost effective route planning solution that can be integrated into your existing business process? Take a look at Strategic Movements to see what we can offer.
Like the blog? Sign up and be the first to hear what we have to say.