How can AI improve your average handle time?

When you break down a call (greeting, verify identity, understand the problem, research the problem, communicate the answer, closing, after call work), you can see opportunities during each of these phases to improve AHT. One of my favorites during the verify identity phase is voice biometrics, resulting in 30-45 seconds improvement. Going beyond relevance search to find a solution can be another one (i.e. semantic search). In short, when you break down a call, you can start to see places where AI can be applied to reduce seconds. I did a quick calculation and determined that 1 second of AHT is worth $2M per year in a 15,000 person center. The total savings potential is enormous.

I agree, by using NLP it can also help to understand the call drivers of a contact centre and understand the needs of the customer better and help the L&D department to give more development sessions to the agents. On the other hand it also helps the WFM team to forecast call volumes based on the call drivers as in a case of shared services campaigns and minimise cost as a whole.

Reducing AHT in customer interactions is at the top of nearly every customer service operation, BUT…if efforts to reduce AHT are detrimental to CSAT / NPS metrics, additional thought must be applied. Blindly applying AI doesn’t guarantee greater efficiency; rather, taking a discovery-led use-case approach helps to maximize value while mitigating negative repercussions. For AI to be beneficial to agents, you’ll want the following: 1. Real-time sentiment and conversational guidance in the moment it’s needed. Even a lag of several seconds loses effectiveness. 2. Specific guidance by call type. Done correctly, and when triggered by the topic at hand, AI can provide useful guidance and even quick links to relevant resources to save AHT.

A long-time challenge in all contact centers has been locating the applicable knowledgebase content to support complex inquiries where the agent doesn't immediately know the answer. These searches often returned many documents that agents would have to sort through in hopes of finding the correct information. AI, especially through machine learning and natural language processing (NLP), enables semantic search engines to understand and interpret human language in a nuanced way. This goes beyond traditional keyword matching to comprehend the intent and contextual meaning of queries. Semantic search not only speeds up the resolution process, lowering AHT but also enhances the quality of the customer interaction.

Utilizing AI, particularly Large Language Models (LLMs), in contact center operations presents a transformative approach to reducing Average Handling Time (AHT). These advanced AI systems can analyze extensive contact center data to uncover deep insights, identifying root causes, patterns, and trends influencing AHT. This capability empowers those with limited technical expertise to bridge complex data sets and conduct robust analyses. When combined with AI and LLMs, tools like Python become more accessible, allowing laypersons to perform sophisticated data analysis.