Optional sources of data for predictive routing

While some of the required data is available for Genesys predictive routing to readily consume, you can make other additional data available for predictive routing to use. 

If available, predictive routing uses data from the following sources:

  • If your organization has web survey enabled and you have enough survey data available for agents assigned to GPR-enabled queues, predictive routing uses the survey data for model training and routing prediction purposes.
  • Agent profile data
  • Participant data

While not mandatory, the availability of the following optional data can help your organization achieve better routing results.

Note: Before you save the information, ensure that the text is spelled correctly.

For the agent profile data to be considered for routing decisions, make sure you have the following information available for a maximum number of agents who are working on queues that have Genesys predictive routing enabled.

  1. Add agent’s employment and certification data. See Standard profile fields
    1. Click Admin.
    2. Under Directory, click My Organization.
    3. In the Search field, to filter the type the first few letters of the person’s first or last name.
    4. Enter the following data:
      Section  Field
      HR
      • Hire Date
      • Employee Type
      Skills and Certifications Certifications
      General Department
  2. Add the agent’s ACD skill and language data. See Add an ACD skill or language to a user.

For customer data to be considered for routing decisions, customers must have attributes, or participant data, assigned to them. The participant data must meet the following conditions:

Participant data Conditions
Name of attribute

Can contain:

  • Alphanumeric characters
  • Underscore

For predictive routing purposes, participant data attributes are considered case-insensitive. Genesys highly recommends to not have two participant data attributes with the same name.

Note: If you use special characters such as spaces, dots, and commas, predictive routing does not consider the attribute for routing decisions.
Value of attribute

Can contain:

  • Numeric values (integer or decimal)
  • Boolean (true/false). Ensure that every attribute contains either a true or false value; Genesys Cloud does not use any default value if the value is not populated.
  • String values
    • String values are case-sensitive.
    • Long strings are ignored.
Availability Genesys recommends that you attach attributes to all or most of the conversations. If an attribute is attached only to a small volume of interactions, predictive routing does not use the data for routing decisions. 
Cardinality

If attribute values are of string type, ensure that the cardinality of those values is low. If cardinality is high (for example every conversation has different value for certain attribute), predictive routing does not use the data for routing decisions. This cardinality condition does not apply to attributes with numeric values.

PII data

Genesys recommends that you do not add any PII as participant data attributes as either keys or values. 

Note: Any participant data, which satisfies these criteria may be used by predictive routing models. Therefore, Genesys recommends that you review whether any attributes might introduce bias such as race or gender and exclude them.

Following are examples of participant data attributes that enhance the accuracy of predictive routing decisions:

Attribute name Description Allowed values
CUSTOMER_IS_VIP Is the customer a VIP true/false
CONTRACT_TYPE Type of contract the customer has silver/gold/platinum
INTENT Intent of customer 1,2,3,4,5,6
PROPENSITY_TO_CHURN Propensity of customer to churn Any number (may be between 0-100)