Question: What’s the difference between real-time guidance and next-best-action recommendations in interaction analytics solutions?


Real-time guidance (RTG) applications leverage artificial intelligence (AI) technologies to understand live conversations as they occur and to use this information to deliver actionable alerts or recommendations to agents and/or supervisors. In order to provide in-the-moment guidance, these solutions rely on an application’s ability to find and pull real-time information from disparate systems and databases (including knowledge bases) and combine it with contextual information about the interaction from the agent’s desktop and/or conversation. Data about the agent’s profile (tenure, skill proficiency, key performance indicators (KPIs), etc.) can be factored in through integrations with quality management, workforce management, performance management, and human resource systems and other applications. Data about the customer and their preferences can be retrieved and incorporated through integrations with a CRM solution or other servicing systems. The real-time information collected while the customer is on the line can be aggregated and compared with business rules or AI-based predictive analytics models/algorithms. Based on the real-time data analysis and/or client-defined rules, the application presents agents with the most appropriate information or action to take. 

Next best action (NBA), which also leverages AI, is slightly different from RTG. It provides targeted prompts for new sales, up-sell, cross-sell, retention, collections, or other purposes, derived from real-time insights from the interaction and contextual information from other applications. Based on real-time data analysis, agents are presented with call-outs, pop-up boxes, scripts or step-by-step instructions that provide the most appropriate action to take/optimal offer to make, etc. In other cases, next-best-action recommendations are used to make sure agents say the correct things at the right times, as is the case with the mini-Miranda in collections environments. The recommendations may also pertain to the need to properly verify a caller’s identity or to read a non-disclosure statement and/or capture consent for a sales transaction.