Predictive metrics can forecast the future.
Some do this using forecasting models, which trawl through reams of historical data to predict likely outcomes.
In contact center workforce management (WFM), these predictive metrics are already commonplace, as planners forecast contact volumes, handling times, and occupancy rates.
Nevertheless, until recently, few service operations had embraced the broader application of predictive metrics beyond forecasting. GenAI is helping to change that.
Using its reasoning capability, GenAI can inhabit a role and then predict the most likely outcome based on an interaction.
For example, consider the expected net promoter score (xNPS) available on evaluagentCX, a widely utilized contact center quality assurance (QA) platform.
xNPS harnesses GenAI to inhabit the shoes of a customer and, based on the service interaction, share the outcome that customer would have likely left on an NPS survey. So now, they don’t have to fill in a survey at all.
Diving deeper, Ben Cave, Product Director at evaluagent, said:
“In more than 85 percent of cases, the solution predicts the same NPS outcome that the customer actually gave the conversation.”
According to Cave, up to 97 percent of customers ignore the post-call NPS survey. As such, CX teams may miss 97 percent of customers open to an upsell or likely to churn.
xNPS may change that, exemplifying how predictive metrics can boost the future contact center.
During his session at the recent Contact Center Performance Summit, Cave explored this further, sharing four ways predictive metrics will shape the contact center of tomorrow.
1. Predictive Metrics Will Support More Personalized & Unique Experiences
Nowadays, many contact centers put automated services in front of the live agent, offering self-service, gathering upfront information, or simply clarifying the customer’s contact reason.
Yet, with “containment” a central objective, some businesses do everything possible to keep customers within the conversational AI interface.
That irritates customers who just want to speak to a live agent. Many contact centers don’t take enough account of this.
One reason why is that it’s tricky to predict which customers have a strong preference for human-to-human interaction. Noting this, Cave said:
“We are working on ways in which you can better forecast how a customer would like to interact with the contact center, so it can route them optimally.”
Cave suggests that evaluagent is getting closer to predicting which resolution will best suit an individual customer and leveraging that information to tailor a human or AI agent’s responses.
“All of these things are possible by predicting more about what a customer wants from what they say, their previous interactions, and broader customer journey,” he added.
“By using more of these predictive metrics in the future, no two people’s experience with a contact center will be quite the same.”
2. Predictive Metrics Will Enable Fairer Evaluations of Agent Performance
Across the industry, live agents must handle more complex customer questions as AI snaps up all the simple, transactional queries they previously used to take a breather.
A focus on hyper-specialized agents, improved routing, and upgraded supervisor support should come with that transition. Yet, contact centers must also refocus their agent performance metrics.
After all, as contacts become more com flecting negatively on agents who may have faced a prolonged sequence of challenging conversations.
That’s not necessarily fair, and evaluagent is fighting back with predictive metrics. Cave noted:
“We have the ability to difficulty-adjust the scores given to agents in the QA platform, so you can begin to take account of conversation complexity.”
In doing so, evaluagent monitors signals such as how much abuse the customer used and how easily they accepted resolutions. Then, it adjusts scores accordingly and builds a fairer view of agent performance.
Additionally, the vendor offers aptitude forecasting, taking the last three months of an agent’s performance data to model forward, determine a longer-term trend, and predict their path.
