The AI Agent Metric Most Contact Centres Miss
Anubhav Mohanty explains why tracking how a call ends, alongside 100% automated QA and real-time dashboards, is the metric that matters for AI agents.
Up nextThe AI conversation in customer experience is moving past the obvious question.Overview
About this session
In this short from a contact centre summit panel, Anubhav Mohanty of Orvera AI points to the metric many contact centres overlook when they deploy AI agents: how the call ends. A conversation can start with an irritated, frustrated caller and still finish with a satisfied one, and from a customer experience standpoint the ending is what counts.
He explains how Orvera measures it. Every call goes through automated QA, the results feed a real-time dashboard of outcomes and opportunities, and those findings go back to developers to keep improving the agent, because no AI agent is perfect on day one. The clip is designed for contact centre and QA leaders choosing what to measure.
Key takeaways
What comes out of it
The ending matters most. How a call ends defines the customer experience.
Sentiment in real time. Track how the caller feels from start to finish.
100% auto-QA. Every call is scored, not a sample.
A feedback loop. QA findings go back to developers to tune the agent.
TranscriptRead the full session
Anubhav Mohanty 00:00:00.00 We do 100% auto-QA and then do a dashboard out of that QA so that you are also on a daily basis, real-time monitoring what the outcomes are, what your top areas of opportunities are. And that goes back to a feedback mechanism to the developers to kind of tweak the AI agent. So because it cannot be perfect on day one. And then on the sentiment side of it, we also, it should be a real-time sentiment tracking. The start of the call can be different and the end of the call is different. But from a customer experience standpoint, what matters is how did the call end?
Anubhav Mohanty 00:00:31.40 So if an AI is able to get the sentiment from irritated and frustrated to satisfied, I think that's a huge win.
Keep watching
More clips
Like what you
just watched?
The same agentic voice, agent assist and quality workflows, running on your own call flows in 3 to 6 weeks.