Skip to main content

How AI and speech analytics transformed Bell’s contact centre operations

By John Rocca, Director, Technical Product Management 

A Bell expert assesses how to best incorporate AI and Voice analytics into a company’s contact centre

By John Rocca, Director, Technical Product Management Contact centres are complex environments. With numerous calls coming in through multiple channels and a workforce of agents to manage, it can be challenging to deliver consistently high-quality customer experiences. As one of the key drivers influencing the evolution of the contact centres,  artificial intelligence (AI) can help organizations handle this complexity – and enhance both the customer and agent experience. For example, AI-powered assistants can give human agents the information they need, when they need it. Additionally, smart analytics can generate actionable insights based on complex call transcripts. These are just a few of the exciting capabilities available with AI for contact centers. The question is, how do you get there?

Like so many large enterprise companies, Bell faced this question with our own call centre operations. To meet the growing demands of our business, we needed to transform to take full advantage of AI-powered capabilities. Needless to say, this was not an easy task; we have one of the largest contact centres in Canada  with more than 12,000 agents and 200 queues to manage.

To gain more insight into our approach, I recently sat down with Harrison Lepofsky, Bell’s Director of National Sales, and Jeff Kurys, Director of Artificial Intelligence, to reflect on our transformation process and the results we have seen so far. 

What were the main drivers for Bell’s contact centre transformation? 

Harrison: Bell is a big company with thousands of agents. That scale continues to be our biggest challenge, especially when it comes to being consistent in the way agents introduce customers to other Bell products, accessories and services.

Having to manually monitor calls for insight into agent performance and coaching opportunities adds to the challenge. Additionally, the monitoring team was small and the process fully manual, so it was impossible to capture a good sampling of an agent’s true performance.  

How did you solve for the challenges you faced?  

Harrison: It was important to first sit down with Jeff and his team to articulate our challenges and goals. It was also important that we engage in proof-of-concept trials to demonstrate the potential of AI and get people comfortable with the new tools and processes.   

Jeff: Harrison is absolutely right. It all starts with open discussion. We worked together with his team to ensure that we had clarity on their business challenges and operational needs. With this insight, we were able to build  and deploy the most appropriate AI-powered tools.

Most importantly, we knew that the team would benefit from Speech AI. By using natural language processing and machine learning to analyze transcripts of every call, Speech AI generates actionable insights for improving agent performance. With speech AI in place, agents now get support from a virtual assistant while conversing with customers. AI follows conversations in real-time and can give agents fast access to knowledge repositories they can use to answer questions and meet requests. It can also  enhance the content within those repositories, making them more consistent and readable. 

We also deployed AI to improve training and agent onboarding by simulating different customer personas for agents to practice on. We have a new process for knowledge reinforcement that tests agents on material they’ve recently been trained on to check for retention, which is similar to popular language learning apps. AI also makes it possible to personalize coaching plans by identifying specific areas of opportunity based on an analysis of a single agent’s entire conversation history. 

The initial deployment of these AI initiatives took approximately three months, starting with one AI model in April 2023. In less than a year, we now have 14 in place and continue to grow. 

With AI implemented in Bell’s contact centres, what’s possible now? 

Harrison: Thanks to the tools built by Jeff’s team, we can now monitor every single call rather than only the small sample that was possible before.  This gives us the full picture of agent performance, which means we can fully track cross-sell conversations and generate accurate cross-sell pitch rates for individual agents and our entire roster. 

With the insights available to us now, we’ve also been able to improve pitch quality. As an example, we’ve been able to leverage Speech AI to learn that agents using the word “I” when discussing offers close at higher rates than those who use “we.” Now we can coach agents to phrase their offers using more personalized language.  

In the same way, we can analyze the words and phrases being used by the agents who are more successful at cross-selling a specific product, then train other agents so they can take the same approach. This is especially useful for complex offerings. 

What have been the impacts so far? 

Harrison: Now that we can identify coaching opportunities across our agent base, cross-sell pitch rates have risen by 40% in just six months – which has helped us improve revenue. We’ve also enhanced our customer experience, indicated by increases in other key metrics, like Net Promoter and Customer Satisfaction scores. 

There are so many opportunities in the market with AI. Speech AI, in particular, is a great example of how Bell is leading the way in operationalizing AI solutions. This is just the beginning.

Bring AI to your contact centre  

It’s clear that AI can improve contact centre performance. With our own adoption of AI, Bell’s frontline management team is now more productive, consistent and unbiased in the way we assess and manage agent performance.  Our agents receive better coaching and training – adding a level of confidence to each pitch and interaction.  

Transforming a contact centre is never easy. Our own experience proves that it takes planning, collaboration and a deep understanding of how contact centres operate and how artificial intelligence can help.  For me, these learnings are invaluable and play a critical role in how we design, implement and support AI-powered contact centre solutions for business customers across Canada.   

Whether you’re moving to a cloud-based contact centre or piloting AI use cases, our team of contact centre professionals can help you find the Contact Centre with AI solutions that best fit your needs.