Jump to: navigation, search


Section: agg-feature
Default Value: No default value
Valid Values: none. This option takes no values—its presence alone within the [agg-feature] section issues the described instructions to RAA.
Changes Take Effect: After restart of the aggregation process.

Enables the aggregate table ID_FCR (AG2_ID_FCR), which provides aggregate information to support reporting on Genesys Predictive Routing (GPR).

To have RAA exclude Predictive Reporting data, remove this option from this section. The table is not presented if feature is not enabled.



Section: agg-feature
Default Value: No default value
Valid Values: none. This option takes no values—its presence alone within the [agg-feature] section issues the described instructions to RAA.
Changes Take Effect: After restart of the aggregation process.

Enables Genesys Predictive Routing (GPR) - related columns in four aggregate tables (AG2_ID, AG2_I_AGENT, AG2_QUEUE, and AG2_QUEUE_GRP).

To have RAA exclude Predictive Reporting data from these tables, remove this option from this section. The columns are not presented if feature is not enabled.


Key-Value Pair

Also known as a KVP. A data structure that is used to communicate or store a piece of information. A KVP consists of a key, whose value is an S.100 symbol, and a value, which may be any of a variety of data types, including a key-value set (thus, making the structure recursive). The key identifies the meaning of the data that is contained in the value.



Deploying: Integrating with Genesys Reporting

Predictive Routing (GPR) uses the rich trove of data gathered in Genesys Info Mart to analyze past histories of agent performance and interaction handling, which provides a basis to start matching based on predicted performance. In turn, GPR generates a variety of information about its functioning in the form of key-value pairs (KVPs) attached to Events. This KVP data is stored in the Info Mart database for use by downstream reporting applications, such as Genesys Interactive Insights (GI2 or GCXI), and to be fed back into GPR to refine predictors. In addition, Stat Server sends the KVP data to real-time reporting applications, such as Pulse.

Reports are available through the following reporting applications:

  • GI2 / GCXI as a part of the Genesys historical reporting offering. The following reports are available: Predictive Routing AB Testing Report, Predictive Routing Agent Occupancy Report, Predictive Routing Detail Report, Predictive Routing Operational Report, and Predictive Routing Queue Statistics Report. For details, see the Supplement page in the Genesys Interactive Insights Universe Guide.
  • Pulse, where the Agent Group KPIs by Predictive Model and Queue KPIs by Predictive Model templates for real-time reporting are available from the Genesys Dashboard Community Center .

In addition to the powerful Genesys Reporting solution, Predictive Routing offers in-interface analysis reports, including the following:

  • Feature Analysis
  • Agent Variance
  • Lift Estimation

See Creating and Interpreting Reports in the Predictive Routing Help for details.

Importing Info Mart Data into Predictive Routing

There is no automated link or pipeline that sends Info Mart data to GPR. To transfer data, set up a job (using Python scripting, for example) to extract the data and create a CSV file, which you can then upload using either the API or the GPR application.

For information about the structure and content of Info Mart data, see the Genesys Info Mart Physical Data Model for your RDBMS.

For information on how to load CSV data into Predictive Routing, refer to the following:

Sending Predictive Routing Data to Genesys Info Mart

Genesys Info Mart receives data from Predictive Routing in EventUserEvent TEvents generated by T-Server or SIP Server.

When you use GPR to match agents and interactions, set up your routing solution to attach KVP data about the routing process to an EventUserEvent ("UserEvent"). The UserEvent contains AttributeThisDN with a value of Route Point, which identifies where the strategy is executed, and AttributeUserData, which holds a list of KVPs containing data about the interaction. This KVP data is stored in the Interaction Concentrator Interaction Database (IDB), from which Info Mart gathers it and prepares it for use in GI2/GCXI historical reporting on GPR activity and performance. Genesys Info Mart has specific requirements for the KVPs that must be attached (see GPR KVPs for Genesys Reporting, below).

For general information about how Genesys Info Mart uses attached user data, see the Genesys Info Mart and Attached User Data section of the Genesys Info Mart Deployment Guide.

Genesys Info Mart release 8.5.009 and later provides support for GPM reporting out-of-box, with no additional configuration required on the Genesys Info Mart side. However, to send GPR data to Genesys Info Mart, as well as to see GPR data in GI2/GCXI reports, you need to modify the configuration of GPR, Interaction Concentrator, and Reporting & Analytics Aggregates (RAA, the aggregation engine hosted by Genesys Info Mart).

Enabling GPR Reporting in Your Deployment

  1. In IRD or Composer, set up your routing solution to attach the required KVPs in UserEvents.
    To have Predictive Routing attach this KVP data, configure the following options configured on the Predictive_Route_DataCfg Transaction List object:
    • send-user-event - Enables attaching the Predictive Routing-specific key-value pairs.
    • vq-for-reporting - Indicates the virtual queue where the strategy is executed.

    For the list of KVPs attached, see GPM KVPs for Genesys Reporting, below.

  2. Enable Interaction Concentrator (ICON) to store the UserEvent data.
    Ensure that ICON and IDB have been deployed as Genesys Info Mart requires, and that ICON is connected to the T-Server(s) or SIP Server(s) handling the GPR interactions. For more information, see Preparing Interaction Concentrator in the Genesys Info Mart Deployment Guide.

    When you configure the ICON application, configure the options in the [custom-states] section so that ICON will store the required KVPs in the G_CUSTOM_DATA_S table in IDB. For more information, see Important custom-states ICON Configuration Options and Configuration Considerations in the Genesys Info Mart Deployment Guide.

  3. Enable aggregation of GPR data. (Required for GI2/GCXI reporting or other applications that use RAA aggregation.)
    In the [agg-feature] section on the Genesys Info Mart application object, specify the enable-gpr and enable-gpr-fcr options.

  4. Verify the reporting data chain.
    After a few interactions have been routed with GPR in an operational mode, verify that the required KVPs are being sent, stored, and used as expected:
    • Check the T-Server/SIP Server logs to verify that UserEvents are being sent with the required KVPs.
    • Check the ICON logs and the G_CUSTOM_DATA_S table in IDB to verify that ICON is recording the required KVPs.
    • Check the GPM_* tables in the Info Mart database to verify that Genesys Info Mart is correctly transforming the data.

GPR KVPs for Genesys Reporting

The following table describes the KVPs that Genesys Info Mart uses to enable GPR reporting.

  • Use the drop-down selector at the top of the table to increase the number of entries to show, to see all the Predictive Routing KVPs without having to scroll through the table. You can also use the Search box to quickly locate a specific KVP.
  • Although the Predictive Routing short name is GPR, the GPM_* prefix shown in the table below is correct. It reflects an earlier name for the product.

KVP Description KVP Type Info Mart Database Target
KVP Description KVP Type Info Mart Database Target
(Optional) The Employee ID of the agent to whom the interaction was routed. INT GPM_FACT.RESOURCE_CFG_DBID, RESOURCE_.RESOURCE_CFG_DBID
The rank of the agent in the target group, based on agent scores sorted in descending order. SHORT GPM_FACT.AGENT_RANK
The score of the agent to whom the interaction was routed. INT GPM_FACT.AGENT_SCORE
Indicates whether features from the customer record were successfully retrieved from the customer relationship management (CRM) database and used to calculate agent scores.
Valid values: 0 (= No), 1 (= Yes), unknown
The average score calculated for a sub-group of agents in the target group for whom the global model was utilized in score computation. INT GPM_FACT.GLOBAL_SCORE
The score of the best-matching agent in the target group. INT GPM_FACT.MAX_SCORE
The median score for the target group of agents to which the agent belongs. INT GPM_FACT.MEDIAN_SCORE
The message that displays when the Predictive Routing result reported in the gpmResult KVP is an error. CHAR(255) GPM_FACT.MESSAGE
The score of the worst-matching agent in the target group. INT GPM_FACT.MIN_SCORE
The mode in which Predictive Routing is operating, as specified by the prr-mode configuration option.
Valid values:
  • prod
  • off
  • dry-run
  • ab-test-time-sliced
  • unknown
The name of the model used to calculate agent scores for the interaction. CHAR(255) GPM_MODEL.MODEL (referenced through GPM_FACT.GPM_MODEL_KEY)
The UUID of the model. CHAR(24) GPM_MODEL.MODEL_ID (referenced through GPM_FACT.GPM_MODEL_KEY)
The name of the predictor in the Journey Optimization Platform (JOP). If an error is encountered, the section name in the Predictive_Route_DataCfg Transaction List object is used as the predictor name. CHAR(255) GPM_PREDICTOR.PREDICTOR (referenced through GPM_FACT.GPM_PREDICTOR_KEY)
The UUID of the predictor used for scoring. CHAR(24) GPM_PREDICTOR.PREDICTOR_ID (referenced through GPM_FACT.GPM_PREDICTOR_KEY)
The result of Predictive Routing processing. In the case of errors, the gpmMessage KVP contains the error message.
Valid values:
  • 1 - Ok
  • 2 - Authentication to scoring engine failed
  • 3 - Scoring request failed
  • 4 - Agent list is empty
  • 5 - URS overload, interaction skipped
  • 6 - Predictor not found
  • 7 - Failed to build scoring request
  • 8 - SetIdealAgent or SetReadyCondition execution error
  • 9 - Interaction log not found in global map
  • 10 - Unknown error
The sequence number of the attempt to route an interaction using Predictive Routing. INT GPM_FACT.ROUTE_ATTEMPT_ID
Indicates the scenario under which the interaction was processed. For more information about the scenarios, see the planning information about Interaction Flows.
Valid values:
  • agent-surplus
  • call-surplus
  • unknown
The size of the scored target group (in other words, the length of the list of agents received from the scoring engine). This information is more useful for real-time than for historical reporting. SHORT GPM_FACT.TARGET_SIZE
The value depends on the mode in which Predictive Routing is operating (see gpmMode).
Valid values:
  • 1 - When the mode is ab-test-time-sliced, indicates that the interaction was selected for Predictive Routing. When the mode is prod, indicates the normal case, when Predictive Routing occurred without error.
  • 0 - When the mode is ab-test-time-sliced, indicates the interaction was processed with skill-based routing. This value is also reported for all interactions completed without errors when the mode is dry-run.
  • unknown - When an error occurred in one of the Predictive Routing subroutines and the solution defaulted to skill-based routing.
The amount of time, in seconds, the interaction spent in the queue used for Predictive Routing decision-making. INT GPM_FACT.WAIT_TIME

Comment on this article:

blog comments powered by Disqus
This page was last modified on 5 March 2018, at 16:18.