Territory Planning (Transportation Optimization)

Introduction

Territory Planning is a new capability in the Transportation Optimization (Hopper) engine that automatically clusters customers into geographic regions - territories - and restricts routes and drivers to operate within those territory boundaries. This reduces operational complexity, improves route consistency, and enables delivery-promise logic for end consumers.

Territory Planning is available today in Cosmic Frog for all users and is powered by an enhanced high-precision Genetic Algorithm. Please note that all Hopper models, whether using the new Territory Planning features or not, can now be run using this high-precision mode.

Why Territory Planning Matters

Traditional routing optimization focuses on building the most cost-efficient set of routes. In many real-world operations, however, drivers do not cross territories. A driver consistently serving the same neighborhoods knows the roads, customer patterns, and service requirements.

Key benefits include:

• Operational Efficiency: Drivers become familiar with their assigned territories, knowing the streets, traffic patterns, and customer locations, which leads to faster deliveries and improved customer service.
• Predictable Service Windows: By understanding which territories are served on which days, organizations can provide accurate delivery window estimates to customers at the point of purchase. For example, customers can enter their zip code online and immediately see when the next available delivery opportunity is for their territory.
• Easier Network Management: Territories allow for clearer operational structure, with local teams managing specific regions rather than coordinating across the entire network daily.
• Simplified Planning: When entering new markets or experiencing demand changes, territories can be redesigned to optimize for current conditions rather than relying on outdated manual territory designs.

How Territory Planning Works

With the Territory Planning feature one new Hopper input table and two new Hopper output tables are introduced.

Input Table

Territory Planning requires one new input table, Territory Planning Settings, while supporting all existing Hopper tables. This table defines the characteristics and constraints of the territories to be created during the Hopper solve, and can be found in the Functional Tables section: 1. Territory Type: A descriptive name for the territory configuration (e.g., "Large Territories", "Balanced Small Territories").

2. Number of Territories: Specify how many territories to create. This field is not required: territories will be created based only on the constraints which are specified if the number of territories is not specified.
3. Constraint Fields: Define limits for each territory. These fields are also not required: territories will be created based on the constraints which are specified, in combination with the number of territories (if set).
   • Min/Max Delivery Locations
   • Min/Max Pickup Locations
   • Min/Max Delivered Quantity‍
4. Status: Control whether the territory type is active in the model; often changed through scenario items when running scenarios to try out different territory settings. Options are Include and Exclude.

The universal compatibility with all other Hopper tables ensures you can add territory planning to any existing Hopper model without needing to restructure your data.

Output Tables

Territory Planning generates two new output tables, the Transportation Territory Planning Summary and the Transportation Territory Assignment Summary, in addition to all standard Hopper outputs. The Transportation Territory Planning Summary table provides one record per territory with aggregate KPIs:

1. Territory Name: Unique identifier for each territory (e.g., "1", "2", "3").
2. Territory Type: The type of territory, as named in the Territory Type field on the Territory Planning Settings input table.
3. Scenario: The scenario the territory is part of.
4. Performance Metrics:
   • Number of delivery locations
   • Number of pickup locations
   • Total delivered quantity
   • Total delivered volume
   • Total delivered weight
   • Number of delivered shipments
   • Number of routes
   • Total distance
   • Total time

This table is useful for:

• Comparing territory sizes and workload distribution
• Identifying imbalanced territories that may need adjustment
• Exporting summary statistics for reporting and presentations

The Transportation Territory Assignment Summary table shows the detailed assignment of each customer to a territory:

1. Location Name: Each delivery location in your network.
2. Territory Name and Territory Type: The territory the location is assigned to, and the type of this territory.
3. Scenario: Links to the scenario that generated the assignment.
4. Geographic data: Latitude and longitude of the location, for mapping purposes.

This table enables:

• Creating detailed territory maps showing location assignments
• Analyzing geographic cohesion of territories
• Exporting assignments for operational implementation
• Comparing territory assignments across scenarios

Besides the 2 new output tables, the following existing transportation output tables have new fields Territory Name and Territory Type added to them: Routes Map Layer, Transportation Asset Summary, Transportation Segment Summary, and Transportation Stop Summary. This facilitates filtering on territories in these tables to for example quickly see which assets are used in which territories.

Algorithmic Hopper Enhancements

Territory Planning uses Hopper's advanced genetic algorithm to simultaneously optimize multiple objectives:

• Minimizing transportation costs (distance, time, number of assets)
• Creating well-defined geographic territories without overlap
• Respecting constraints on territory size (customers, volume, weight)
• Balancing workload across territories (coming soon)

The genetic algorithm is available in Hopper's High Precision solver mode and powers all Hopper optimizations, not just territory planning. To turn on high precision mode, expand the Transportation (Hopper) section on the Run Settings modal which comes up after clicking on the green Run button at the right top in Cosmic Frog. Then select "High Precision mode" from the Solver Focus drop-down list:

Template Model

A model showcasing the Territory Planning capabilities can be found here on Optilogic's Resource Library. We will cover its features, scenario configuration, and outputs here. You can copy this model to your own Optilogic account by selecting the resource and then using the Copy to Account blue button on the right hand-side (see this "How to use the Resource Library" help center article for more details).

Inputs

First, we will look at the input tables of this model:

1. The breadcrumb trail at the top of the Optilogic platform indicates we are in the Cosmic Frog application, and a model named Territory Planning Template Model is open.
2. Use the Module menu to open the Data module where a model's input, output, and custom tables can be found.
3. Here, we are looking at the input tables:
   1. Use these icons to swap between showing input tables (left-most icon), output tables (middle icon), and custom tables (right-most icon).
   2. There are several filter options for the tables here, we have clicked on the second icon to only show non-empty tables.
4. We see that a subset of the standard Hopper tables has been populated; this model contains:
   
   • 100 customer locations - all in and around Atlanta, Georgia, in the US
   • 1 facility - a distribution center in Atlanta
   • 1 product
   • 1 asset with capacity for 24 units, using 1 rate with a fixed cost of $100 per route and a distance-based cost of $1 per mile
   • The 100 records in the shipments table are for each customer receiving 1 shipment of the included product; the quantities of these vary from 1 unit to 20 units
5. The new Territory Planning Settings table in the Functional Tables section is shown in this next screenshot:

1. We have opened the Territory Planning Settings table
2. Two Territory Types are specified in the table, the first one is called "Small Territories". The specification for this requires 5 territories to be created, each delivering to a minimum of 18 locations. Note that the other input fields (see list above in the Input Table section further above) of this table are not used in this example, we just want the territories to adhere to these 2 conditions.
3. The second Territory Type is called "Large Territories" and this one requires 3 territories which each deliver to at least 30 locations.
4. Both Territory Types have their Status field set to Exclude, each will be included in a separate scenario where a scenario item changes its value to Include.

Let us also have a look at the customer and distribution center locations on the map before delving into the scenario details:

Scenarios

In this model, we will explore the following 4 scenarios:

• No Territories: the cost-optimized solution where customers are not required to be clustered into territories
• Predefined Territories: manually designed territories are enforced
• 3 Optimized Territories: this scenario will return the lowest cost solution that contains 3 large territories which each deliver to at least 30 locations
• 5 Optimized Territories: this scenario will return the lowest cost solution that contains 5 smaller territories which each deliver to at least 18 locations

To set up the predefined territories scenario, the Notes and Decomposition Name field on the Shipments table are used:

1. We are on the Shipments input table
2. In the Notes field, the territory the shipment should be assigned to is entered. There are 5 territories that were manually designed: Central, North, East, South, and West.
3. Through a scenario item in the Predefined Territories scenario, we will set the Decomposition Name field to the value of the Notes field. The Decomposition Name field indicates which shipments are grouped together to be solved as a sub-problem by the Hopper engine which will create 5 territories in this example case (Central, North, East, South, and West).

The configuration of the scenarios then looks as follows:

1. Use the Module menu to switch to the Scenarios module
2. The 4 scenarios as described above are set up:
   1. No Territories - this scenario does not contain any scenario items; all input tables will be used as is.
   2. Predefined Territories - this scenario contains 1 scenario item (shown on the right) where the Decomposition Name field on the Shipments table is set to the value of the Notes field. See the previous screenshot of the Shipments table for an explanation of these fields.
   3. 5 Optimized Territories - this scenario contains 1 scenario item name "Include Small Territories". The configuration of the scenario item is not shown in the screenshot, but it changes the Status of the record specifying the Small Territories territory type in the Territory Planning Settings table (see screenshot further above) from Exclude to Include.
   4. 5 Optimized Territories - this scenario contains 1 scenario item name "Include Large Territories". The configuration of the scenario item is not shown in the screenshot, but it changes the Status of the record specifying the Large Territories territory type in the Territory Planning Settings table (see screenshot further above) from Exclude to Include.

These scenarios are run with the Solver Focus Hopper model run option set to High Precision Mode as explained above.

Outputs

Now, we will have a look at the outputs of these 4 scenarios and compare them. First, we will look at several output tables, including the 2 new ones. We will start with the Transportation Summary output table, which summarizes costs and other KPIs at the scenario level:

1. For the Predefined Territories scenario, we see it has the highest total transportation cost of the 4 scenarios, due to the total route fixed costs being the highest. More routes are needed when using manually designed territories as compared to the 3 & 5 Optimized Territories and the No Territories scenarios.
2. As we expected, the No Territories scenario has the lowest overall cost as this scenario is only driven by finding the lowest cost solution and does not need to adhere to constraints around clustering customers together in territories. Both the total route fixed costs, indicating fewest routes needed out of all the scenarios, and lowest total distance cost as compared to the other 3 scenarios. This scenario may have reduced operational efficiency however as drivers are not assigned to a defined territory.
3. The 3 Optimized Territories scenario has a lower total cost as compared to the Predefined Territories scenario due to the reduction in route fixed costs being higher than the increase in distance costs. This shows that designing territories using an optimization algorithm can improve cost efficiency as compared to manually designed territories.
4. Like the 3 Optimized Territories scenario, the 5 Optimized Territories scenario also has a lower total cost as compared to the Predefined Territories scenario. The cost is very comparable to that of the 3 Optimized Territories scenario. Choosing between the 3 or 5 Optimized Territories scenario for implementation will come down to deciding if larger or smaller territories will fit better with how the routes will be run, e.g. how many drivers will need to be assigned to a small vs a large territory, will each territory be run from a central or local entity, etc.

The next 2 screenshots are of the new Transportation Territory Planning Summary output table:

The top 3 records show the summarized outputs for each territory in the 3 Optimized Territories scenario, including the territory name and type, the number of delivery locations, the number of pickup locations, and the total delivered quantity. The bottom 5 records show the same outputs for the 5 territories of the 5 Optimized Territories scenario.

Scrolling right in this table shows additional outputs for each territory, including the number of delivered shipments, number of routes, total distance, and total time:

The other new output table, the Transportation Territory Assignment Summary table, contains the details of the assignments of customer locations to territories. The following screenshot shows the assignments of 6 customers in both the 3 Optimized Territories and 5 Optimized Territories scenarios:

Please note that this table also includes the latitudes and longitudes of all customer locations (not shown in the screenshot), to allow easy visualization on a map.

Next, we will have a look at the locations and routes on maps, these are also preconfigured in the template model that can be copied from the Resource Library, so you can have a look at these in Cosmic Frog yourself as well. The next screenshot shows the routes and locations of the Predefined Territories scenario on a map named Transportation Routes. The customers have been colored based on the predefined territory they belong to:

We can compare these routes to those of the No Territories scenario shown in the next screenshot. The customers are still colored based on the predefined territories, and if you zoom in a bit and pan through some routes, you will find several examples where customers from different predefined territories are now on a route together.

The following 2 screenshots compare the 3 and 5 Optimized Territories scenarios in a map called Territories Transportation Routes. The first shows the customers colored by territory. This is done by adding a layer to the map for each territory. The Transportation Stop Summary output table is used as the table to draw each of these "CZs Territory N" layers. Each layer's Condition Builder input uses "territoryname = 'N' and stoptype = 'Delivery'" as the filter; this is located on the layer's Condition Builder panel:

We see the customers clustered into their territories quite clearly in these visualizations. Some overlap between territories may happen, this can for example be due to non-uniform shipment sizes, pickup / delivery time windows, using actual road distances, etc.

This last screenshot shows the routes of the territories too, they are color coded based on the territory they belong too. This is done by again adding 1 map layer for each territory, this time drawing from the Transportation Routes Map layer table, and filtering in the Condition Builder for "territoryname = 'N'":

Best Practices

• Use High Precision Mode for the solver focus setting
• Start with an unconstrained scenario
• Add territory constraints
• Compare multiple territory configurations and their costs