Workload balancing uses all agents in a balanced way, but still considers the best possible agent-interaction matches. When agent idle time falls outside acceptable limits, workload balancing overrides predictive routing recommendations.

Predictive routing pushes calls toward the agents most likely to achieve a positive KPI outcome. When compared with the standard approach of contact centers that target the longest waiting suitable agent, predictive routing can create an ‘imbalanced distribution’ of calls . Higher-scored agents can receive more calls than other agents or more calls than they would have received without predictive routing. Lower-scored agents may receive fewer interactions, which can reduce learning opportunities. 

When you enable workload balancing, Genesys Cloud rescores the agents in one of the following two ways:

• Uses agent idle time: Predictive routing evaluates agents based on KPI outcomes and then rescores agents based on their idle time. Genesys Cloud calculates idle time based on the queues assigned to an agent. These queues could include interaction types other than the inbound voice interactions that predictive routing supports. Scores for agents with low idle time adjust downward to increase the time between interactions, and scores for agents with high idle time adjust upward to increase the frequency at which they receive interactions. After rescoring, the system routes interactions based on the new scores. Occupancy values for your busier agents go down, while those values for under-occupied agents go up. 
• Uses the current occupancy rate: Predictive routing compares the existing occupancy rate with the occupancy threshold values that you specify. If agent occupancy for the last hour is below or above the threshold, Genesys Cloud then rescores the agents up or down, respectively. Occupancy is the percentage of time agents spend handling interactions against available or idle time. Occupancy percentages appear in the Agents Status Summary view. If you do not specify occupancy threshold values, Genesys Cloud defaults to the agent idle time method.

Because enabling workload balancing reduces the frequency at which the most suitable agent is chosen, the trade-off is a decrease in the positive KPI impact of predictive routing. The best way to check this impact is to run comparison tests. Manually run the comparison test for two weeks without workload balancing, then enable workload balancing and test again for two weeks. Compare the impact on your target KPI for the two tests. For more information, see Test predictive routing for your queues.

To set up workload balancing, see Create and configure queues.