TerriTool A.I. Route Optimisation

About

TerriTool is a route optimisation tool for territory sales teams that was initially built for EDI express in an effort to automate the 20 – 30 hour excruciating route creation task each salesmen had to do every month for their monthly accounts. This solution quickly evolved in to a product called ‘TerriTool’ that is currently being sold to territory sales teams as a solution to reduce the amount of time spent creating territory based routes and also manage their day-day sales activities within the product.

Problem

Once TerriTool had become a well-established route generation system and CRM for territory salesmen. Spencer noticed the need for a different type of route that would be more tailored to territory salesmen. Similar to the manual method he used to use to produce routes, he wanted to have routes generated but with importance to clusters. As territory salesmen focus on a certain area of a territory at a time, he found that the TerriTool TSP route generation system was overly optimised and provided a route which was, in reality, the quickest but would a lot of the times take him outside of a certain area and then bring him back to that area a couple of accounts (waypoints) later.

We had to create a route generation system that was able to detect groups/clusters. Based on the detection of the groups/clusters, the route generation system was required to create an optimised route that would navigate the user between all the accounts(waypoints) of one area before moving to the next. The goal was to find a balance between the time taken to complete the route and keeping the salesmen within a certain area before moving on to the next. A good example would be, if the territory salesmen had accounts in Los Angeles, Fountain Valley, and Sant Ana, then the system should generate the route in the most optimised form while keeping the salesmen within Los Angeles until all the accounts of Los Angeles are completed before moving on to the next most logical (in terms of route optimisation) area.

Process

From our solutions assessment we decided that the best way to create a route like this was to replicate the manual steps that Spencer would do but with a machine, since in his manual process there was a ‘decision making’ component to it, we had to utilise machine learning. Our goal was to first replicate his level of decision making in terms of selecting clusters but then have the A.I. system do a better job at it by considering multiple parameters, more than Spencer would during his manual process.

Once we had determined how the in-house TSP route generation system would work (For NDA reasons this cannot be disclosed), we carried on to the prototyping stage of the project, where we prototyped, tested, refined, prototyped, tested, refined, prototyped, tested, refined… until we and Spencer felt that we nailed the flow of creating a route.


Once the solutions assessment was complete, we were then ready to start development of TerriTool and the TerriTool in-house TSP route generation system. Our Solutions assessment ensured that we were on the right step before we started development and had no hiccups during the development stage of TerriTool and the TerriTool in-house TSP route generation system.

During the solutions assessment, we decided that we would name this A.I. route optimisation system “Smart Cluster Generation”. The key concerns we had to answer in our solution assessment were:

  • How do we determine the clusters/groups.
  • How do we determine the order in which the clusters/groups should be.
  • How do we determine the best ‘Start’ and ‘End’ point for each cluster/group within the route.
  • How do we ensure that the system is self-learning to enhance it’s ability to determine clusters/groups based on multiple parameters.
  • How do we ensure that the system is self-learning to enhance it’s ability to determine the best order in which clusters/groups should be.
  • How do we ensure that the system is self-learning to enhance it’s ability to determine the best ‘Start’ and ‘End’ point for each cluster/group within the route.
  • What parameters can we work with?
  • What parameters are actually useful to further optimise?
  • Which other systems will we be using and how will they work together to achieve the Smart Cluster Generation system

Once we had answered the above (For NDA reasons this cannot be disclosed), we carried on to the prototyping stage of the project, where we prototyped, tested, refined, prototyped, tested, refined, prototyped, tested, refined… until we and Spencer felt that we nailed the flow and method of creating a the smart cluster route.


Once the solutions assessment was complete, we were then ready to start development of TerriTool Smart Cluster generation system. Our Solutions assessment ensured that we were on the right step before we started development and had no hiccups during the development stage of TerriTool Smart Cluster generation system.

Solution

With the smart cluster generation system, we were able to achieve routes of 500-5,000 waypoints with bias towards clusters, keeping salesmen within a certain area before moving on to the next. Our solution for the TerriTool system turned a 20-30 hour task in to a 5-10 minute task, while producing a much more optimised clustered route (due to the consideration of more parameters) when compared to the manual process Spencer would follow. This meant that Spencer and his team could spend more time and make the most of their time on the field making sales with a route more tailored to their workflow and spend less time on non-revenue generating tasks.

Testimonial

Recent Projects