Modeling sourcing policies
Sourcing policies define the path used for demand and product requests within the network model. For example, when a customer seeks a specific product, the sourcing policy sends the request to a site that can fulfill this demand, such as a distribution center. Sourcing policies are defined in two tables:
- Site Sourcing Policies – These are policies between sites, such as between a manufacturer and a distribution center.
- Customer Sourcing Policies – These are policies between a site and a customer.
Sourcing policy elements
A site's or customer's sourcing policies include the following elements:
- Site or Customer. The site or customer that sends a request for additional product.
- Product. The products that are served by the sourcing policy.
- Sourcing Policy. Method by which a source is selected, including multiple source or single source.
- Source. The potential single source or group of sources.
- Lead Times. Amount of time required before the product request can be met.
The following are sample Site Sourcing Policies records:
| Site Name | Product Name | Sourcing Policy | Policy Parameter | Source Name |
| MFG | Raw Material A | Multiple Sources(Most Inventory) | 1 | SUP1 |
| MFG | Raw Material A | Multiple Sources(Most Inventory) | 1 | SUP2 |
| DC1 | Product A | Single Source | 1 | MFG |
| DC2 | Product A | Single Source | 1 | MFG |
Types of sourcing policies
Sourcing policies can be implemented in a variety of ways, with slightly different requirements for implementation based on the type selected. The following sourcing policy types are available:
| Sourcing Policy | Description |
| Single Source | In a Single Source sourcing policy, the site can have its replenishment orders filled by only one source site. |
| Single Source (Select Closest) |
In Simulation, this policy selects the geographically closest site as the source. Decisions are made based on the distance to the sources, as specified in the Transportation Policies table. In Network Optimization, this policy functions in the same way as Single Source. The site is selected based on the overall minimum cost or maximum profit, based on the optimization objective. NOTE: In Simulation, the distance used is the distance calculated between the latitude/longitude of the source and destination sites, rather than the Distance populated in the Transportation Policies table. |
| Multiple Sources(Most Inventory) |
Simulation: This policy reviews the inventory for the specific product at all possible sources the moment the order is processed at the site. It then selects the site with the most inventory as the source. Optimization: This policy allows sourcing for the specific product from any of the possible sources. Network Optimization uses feasibility and cost considerations to determine which source(s) to use. |
| Multiple Sources(Order of Preference) | Simulation Only: In this policy, the site selects the source based on input defined in the Policy Parameter field. When the order comes in, source sites are checked by the order of preference for adequate inventory to fill the order. The first site that meets the necessary inventory is selected as the source site. If none of the sources have enough inventory to fill the order, the site of highest preference is selected. |
| Multiple Sources(Probability) |
Simulation: The source for the site’s order is selected randomly based on the probabilities specified in the Policy Parameter column. For example, if two sources have Policy Parameters of 25 and 75, a random value between 1 and 100 for each demand or replication. A value of 1 to 25 would favor the first source, from 26 to 100 would favor the second. Optimization: The order is divided based on the percentages specified in the Policy Parameter column, and then sourced out to the respective sources: If the value is non-zero, use the exact value. For example, if you enter 50, then 50% is sourced using this record. The Policy Parameter values should add up to 100, which represents 100%, however if the inputs do not add up to 100, the values will be scaled to 100 automatically.
Note: The Policy Parameter values for a site’s Multiple Sources (Probability) sourcing policies should add up to 100, which represents 100%. If not, the results are affected by the Enforce Fractional Sourcing Percentage selection. A value of 0 is not supported in Optimization; instead, a value of 1 will be used. You can express the Policy Parameter as 0 to 100, or 0 to 1. For example, values of 25 and 75 are evaluated the same as values of .25 and .75. |
| Multiple Sources(Split By Ratio) |
Simulation: The order is split according to the value you set in the Policy Parameter field. If the product Type is Continuous, shipment items can be split into fractional orders to meet the ratios. If the product Type is Discrete, the policy rounds the splits to the closest whole number based on the ratio, and ensures that the sum of the splits equals the original total. Optimization: The order is divided according to the percentages specified in the Policy Parameter field, and then sourced out to the respective sources:
Note: The Policy Parameter values for a site’s Multiple Sources (Split By Ratio) sourcing policies should add up to 100, which represents 100%. If not, the results are affected by the Enforce Fractional Sourcing Percentage selection. A value of 0 is not supported in Optimization; instead, a value of 1 will be used. You can express the Policy Parameter as 0 to 100, or 0 to 1. For example, values of 25 and 75 are evaluated the same as values of .25 and .75. |
| Multiple Sources(Fastest Path) | Simulation Only: All potential sourcing paths are considered. The delivery time for each path is calculated as sourcing time plus expected load, unload, and transport time or load, unload, and transport delay time. Sourcing then selects the fastest of the calculated paths. |
| Multiple Sources(Close to Due Date) | All potential sourcing paths are considered, and an option is chosen that delivers the shipment as close to the due date as possible without passing that date. If no path can deliver by the due date, the path that delivers the shipment the fastest is chosen. The starting point of a path is a site with sufficient inventory to fulfill the demand or a production site. The delivery time for each path is calculated as sourcing time plus expected load, unload, and transport time or load, unload, and transport delay time. |
| Multiple Sources(By Order Number) |
Simulation: allows you to create orders containing multiple demand lines, based on the Tracking ID value in the Customer Demand/Orders table. Simulation groups all the demand lines of that order into a single order. In this way, an order can have multiple items or products. The policy first attempts to fulfill the demand without splitting up any of the products from the order. The selection is made by order of preference. Thus, the site selects the source based on input defined in the Policy Parameter column. When the order comes in, source sites are checked by the order of preference for adequate inventory to fill the order. The first site that meets the necessary inventory is selected as the source site. If none of the sources have enough inventory to fill the order, the order is then split and each product is checked against the order of preference. |
| Source by Transfer |
Primarily used in Simulation: Never places replenishment orders regardless of the inventory levels. Used primarily in push models shipments in the Shipments table are designated to deliver products to the various sites. Also used in conjunction with scripting logic implemented in the scripts table. |
Specifying policies and processes with probability and ratio
The various policies and processes tables have definitions that support splitting between multiple records using probability or using a ratio between the records. For example, sourcing policies include these two policies:
- Multiple Sources(Probability) – Policies have a Policy Parameter value to determine the percentage to be sourced. For Network Optimization, the order is divided based on the probabilities specified in the Policy Parameter column, and then sourced out to the respective sources based on the probability split.
- If the value is non-zero, use the exact value. For example, if you enter 50, then 50% is sourced using this record. The Policy Parameter values should add up to 100, which represents 100%, however if the inputs do not add up to 100, the values will be scaled to 100 automatically.
- If the value is 0, this is treated as 0% and this record is not considered (similar to setting Status = “Exclude”).
- If the value is null, this record is considered “free”. In the case that all records with non-zero parameter values add up to less than 100%, the solver can determine the percentage for any free records. In the case that the non-zero parameter records add up to 100%, the free records are not considered (similar to Status = “Exclude”).
- Multiple Sources(Split by Ratio) – Policies have a Policy Parameter value that is used to establish a ratio between the policies to be used. For Network Optimization, the order is divided based on the probabilities specified in the Policy Parameter column, and then sourced out to the respective sources based on the probability split.
- If the value is non-zero, use the exact value to establish the ratio. For example, if you enter 6 for one record and 2 for the other, then 75% is sourced using the first record and 25% sourced using the second.
- If the value is 0 or null, this record is not considered (similar to setting Status = “Exclude”).
The Probability and Split by Ratio policies applies to the Policy Parameter column in the following tables:
- Site Sourcing Policies
- Customer Sourcing Policies
- Transportation Policies
- Return Policies
- BOM Assignments
- Production Process Assignments
- Sourcing Process Assignments
- Inventory Process Assignments
- Transportation Process Assignments
- Return Process Assignments
- Production Process Steps
- Sourcing Process Steps
- Inventory Process Steps
- Transportation Process Steps
- Return Process Steps
Examples of policies using probability
The following cases show Production Process Assignments records with the “Split by Probability” Assignment Policy.
Case 1: All Probability values defined:
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Probability | 10 |
| Manufacturer | Product 1 | Process 2 | Split by Probability | 20 |
| Manufacturer | Product 1 | Process 3 | Split by Probability | 30 |
| Manufacturer | Product 1 | Process 4 | Split by Probability | 40 |
Process 1 is used for 10% of the production
Process 2 is used for 20% of the production
Process 3 is used for 30% of the production
Process 4 is used for 40% of the production
Case 2: Four options each with a non-zero parameter, with the sum < 100
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Probability | 10 |
| Manufacturer | Product 1 | Process 2 | Split by Probability | 20 |
| Manufacturer | Product 1 | Process 3 | Split by Probability | 10 |
| Manufacturer | Product 1 | Process 4 | Split by Probability | 10 |
The percentage is scaled based on the sum of the Policy Parameter values:
Process 1 is used for 100 * 10/50 = 20% of the production
Process 2 is used for 100 * 20/50 = 40% of the production
Process 3 is used for 100 * 10/50 = 20% of the production
Process 4 is used for 100 * 10/50 = 20% of the production
Case 3: Four options each with a non-zero parameter, with the sum > 100
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Probability | 20 |
| Manufacturer | Product 1 | Process 2 | Split by Probability | 40 |
| Manufacturer | Product 1 | Process 3 | Split by Probability | 60 |
| Manufacturer | Product 1 | Process 4 | Split by Probability | 80 |
The percentage is scaled based on the sum of the Policy Parameter values:
Process 1 is used for 100 * 20/200 = 10% of the production
Process 2 is used for 100 * 40/200 = 20% of the production
Process 3 is used for 100 * 60/200 = 30% of the production
Process 4 is used for 100 * 80/200 = 40% of the production
Case 4: Four options, some having non-null parameter, with the sum = 100
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Probability | 60 |
| Manufacturer | Product 1 | Process 2 | Split by Probability | 40 |
| Manufacturer | Product 1 | Process 3 | Split by Probability | null |
| Manufacturer | Product 1 | Process 4 | Split by Probability | null |
While all records are considered valid to be included, the “free” records (Process 3 and Process 4) are not needed in the solution.
Process 1 is used for 60% of the production
Process 2 is used for 40% of the production
Process 3 is not used for production
Process 4 is not used for production
Case 5: Four options, some having non-null parameter, with the sum < 100
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Probability | 20 |
| Manufacturer | Product 1 | Process 2 | Split by Probability | 40 |
| Manufacturer | Product 1 | Process 3 | Split by Probability | null |
| Manufacturer | Product 1 | Process 4 | Split by Probability | null |
All records are considered valid to be included. The solver determines the percentage assigned to each of the “free” records (Process 3 and Process 4).
Process 1 is used for 20% of the production
Process 2 is used for 40% of the production
Process 3 is used for x% of the production (free)
Process 4 is used for (100 - 20 - 40) - x% of the production
Case 6: Four options, some having non-null parameter, with the sum > 100
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Probability | 120 |
| Manufacturer | Product 1 | Process 2 | Split by Probability | 80 |
| Manufacturer | Product 1 | Process 3 | Split by Probability | null |
| Manufacturer | Product 1 | Process 4 | Split by Probability | null |
While all records are considered valid to be included, the “free” records (Process 3 and Process 4) are not needed in the solution. The percentage for Process 1 and Process 2 are based on the sum of the Policy Parameter values:
Process 1 is used for 100 * 120/200 = 60% of the production
Process 2 is used for 100 * 80/200 = 40% of the production
Process 3 is not used for production
Process 4 is not used for production
Examples of policies using split by ratio
The following cases show Production Process Assignments records with the “Split by Ratio” Assignment Policy.
Case 1: All Probability values defined:
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Ratio | 5 |
| Manufacturer | Product 1 | Process 2 | Split by Ratio | 2 |
| Manufacturer | Product 1 | Process 3 | Split by Ratio | 7 |
| Manufacturer | Product 1 | Process 4 | Split by Ratio | 46 |
Process 1 is used for 100 * 5/20 = 25% of the production
Process 2 is used for 100 * 2/20 = 10% of the production
Process 3 is used for 100 * 7/20 = 35% of the production
Process 4 is used for 100 * 6/20 = 30% of the production
Case 2: Four options, some having null or zero parameter:
| Site Name | Product Name | Process Name | Assignment Policy | Policy Parameter |
| Manufacturer | Product 1 | Process 1 | Split by Ratio | 3 |
| Manufacturer | Product 1 | Process 2 | Split by Ratio | null |
| Manufacturer | Product 1 | Process 3 | Split by Ratio | null |
| Manufacturer | Product 1 | Process 4 | Split by Ratio | 1 |
Process 1 is used for 100 * 3/4 = 75% of the production
Process 2 is not used for production
Process 3 is not used for production
Process 4 is used for 100 * 1/4 = 25% of the production
Custom sourcing policies through scripting
Custom Sourcing Policies can be created for Simulation using scripts. Contact Coupa Support for more information regarding scripting.
Last modified: Friday May 12, 2023