CallMiner’s LISTEN Global Tour has wrapped up for another year, with four events bringing together CX leaders from around the world to discuss the future of AI, automation and conversation intelligence. But what were the biggest lessons for contact centres?
Call Centre Helper was on the ground for the final two stops of the tour, attending LISTEN Prague in September and LISTEN West in Dallas in October.
Across both events, we caught up with industry experts, customers and the CallMiner team to discuss some of the biggest challenges facing contact centres, and how organisations are approaching the next stage of AI adoption.
While there was plenty of discussion around new technology and increasingly sophisticated automation, much of the conversation came back to a familiar challenge: how do we make sure these investments are actually making a difference?
From measuring automation more accurately to improving AI governance and making better use of customer intelligence, here are five of our biggest takeaways from LISTEN 2026.
One of the more interesting discussions at LISTEN West focused on whether organisations are accurately measuring the performance of their automated customer interactions.
After all, it’s one thing to report that a chatbot is containing 70% of enquiries. But how many of those customers have actually had their problems resolved? And how many have simply abandoned the interaction or contacted the organisation again?
During his presentation, Jeff Gallino, Co-Founder and CEO of CallMiner, discussed examples where independently measured containment rates were considerably lower than the figures organisations expected.
Some automation providers were reporting containment rates of 50%, 60% or even 70%, while CallMiner’s assessments had uncovered substantially lower results in certain deployments.
This was also the thinking behind CallMiner’s Automated Quality and Experience Assessment (AQX), a service designed to assess how effectively automated interactions are performing, including bots developed by other providers.
For contact centre leaders, there’s an important distinction to make here.
Just because an interaction has been contained doesn’t necessarily mean the customer’s problem has been resolved.
Organisations need to look beyond containment rates and consider repeat contacts, customer satisfaction and the wider cost of delivering customer service.
Otherwise, they could find themselves investing in automation that looks successful on paper but isn’t delivering the improvements they expected.
AI is making it easier than ever to develop new applications, automated workflows and virtual agents. But as organisations become capable of building these solutions at speed, how do they keep everything under control?
This was a central theme of discussions across both Prague and Dallas, with CX leaders highlighting the industry’s shift from rapid AI experimentation towards a more structured approach to implementation.
At LISTEN Prague, Bruce McMahon, Chief Product Officer at CallMiner, described this as the industrialisation of AI.
With organisations now able to generate enormous amounts of code and develop new applications in a matter of minutes, he raised an important question:
“How do we pick the right idea?”
By LISTEN West, much of the conversation had moved towards the importance of validating, governing and controlling the AI solutions organisations are putting in front of customers.
There were also discussions around automated testing, governance capabilities and ways for AI systems to identify opportunities for improvement.
And this is something contact centres will need to think carefully about as they introduce more AI agents into their operations.
It’s no longer enough to know that an AI agent can complete a customer interaction. Organisations also need to understand whether it followed the correct processes, how its decisions were reached and what happens when something goes wrong.
With regulatory expectations around AI continuing to develop, having appropriate oversight in place will become increasingly important.
Despite the attention being given to automation, another recurring theme across both events was the changing relationship between human agents and AI.
Speakers discussed how organisations are moving towards a hybrid workforce, where AI manages more routine enquiries while human agents focus on increasingly complex interactions.
However, AI also has an important role to play in supporting those agents.
Throughout LISTEN, there were demonstrations of new real-time guidance and coaching capabilities, including tools designed to provide agents with relevant information during customer conversations.
These included customer journey summaries, knowledge-based recommendations and new ways for supervisors to monitor and support interactions as they happen.
There were also discussions around using AI-powered simulations to help agents practise different customer scenarios before dealing with them in a live environment.
For contact centres, this raises some interesting questions about how agent roles and training requirements might change over the next few years.
If AI takes care of more straightforward enquiries, agents are likely to find themselves dealing with a greater proportion of complex or sensitive interactions.
Having enough agents available won’t necessarily be enough. Organisations will also need to make sure those agents have the right skills, experience and support to handle the enquiries coming their way.
And that could mean rethinking everything from onboarding and coaching to knowledge management and workforce planning.
While much of the conversation around AI focuses on what the technology could eventually achieve, LISTEN also provided an opportunity to hear from organisations already putting it into practice.
At LISTEN Prague, we spoke with Bernard Questiaux, CTO of N-ALLO, about the organisation’s work with energy provider ENGIE.
Bernard explained that the organisation had been using automation technology originally developed by VOCALLS, now part of CallMiner, to automate customer interactions.
“We are reaching now 15, 20, 25 percent of automation of these interactions,” he explained.
It is worth distinguishing this from the 50–70% containment rates sometimes claimed by chatbot vendors. Those figures relate to interactions handled by individual bots, whereas ENGIE’s estimate covers the proportion of all customer interactions that are fully automated across four languages. The two measures are therefore not directly comparable.
However, getting to that point hasn’t been without its challenges.
Operating in Belgium means supporting customers across four languages, creating additional work around testing, configuration and improving automated interactions.
Bernard also highlighted the difficulties of being among the first organisations to introduce new approaches to AI-powered customer service.
“There is no clear methodology, so we have to invent and to create the methodology,” he explained.
And that’s an important point for organisations that are still in the early stages of their own AI journeys.
Successful automation requires more than choosing the right technology. Teams need to understand their customer journeys, identify suitable opportunities for automation and establish ways of testing whether their solutions are working.
There’s also a lot to be said for starting with a manageable use case, learning what works and building on those results.
Particularly when the technology itself is still developing.
Contact centres have access to enormous amounts of customer information, from call recordings and chat transcripts to complaints, customer feedback and interaction outcomes.
But how much of that information is actually being used to improve things across the wider business?
At LISTEN West, we spoke with Miriam from Intrefy about the importance of breaking down departmental silos and making better use of contact centre intelligence.
One of the points she highlighted was the need for closer collaboration between customer service, sales, marketing, product and IT teams.
After all, contact centre data can tell organisations far more than how well their agents are performing.
It can highlight recurring product problems, identify gaps in customer communications and reveal weaknesses in business processes that other departments may not even be aware of.
Miriam also discussed the value of having a dedicated contact centre data analyst who can translate customer interactions into insights that the wider business can understand and act upon.
It’s an interesting suggestion, particularly for organisations that have invested heavily in conversation analytics but are still struggling to get those insights in front of the right people.
And it connects with another of the themes running throughout LISTEN.
Speakers highlighted the importance of creating an ongoing cycle of improvement, where customer interactions generate intelligence that can be used to improve agent performance, develop automation and inform future decisions.
Those resulting interactions can then be analysed again to identify further opportunities for improvement.
There were also discussions at LISTEN West around Model Context Protocol (MCP) connectivity, which could allow external AI tools to access intelligence generated through conversation analytics platforms.
This could help make customer intelligence more accessible to teams that don’t regularly use specialist analytics tools.
However, having access to more information is only useful if someone is doing something with it.
For CX leaders, the opportunity is to make customer intelligence a resource for the wider organisation, helping different departments understand what’s going wrong and where improvements could be made.
Looking back at LISTEN Prague and LISTEN West, there’s certainly no shortage of opportunities for AI to change how contact centres operate.
We’re seeing more sophisticated automation, improved real-time support, new approaches to coaching and increasingly advanced ways of understanding customer interactions.
Of course, organisations are starting from very different places. Some are still getting to grips with their customer data, while others are already introducing more advanced automation. Knowing where you are now, and what you need to put in place before moving further, matters just as much as the technology you choose.
But perhaps the biggest takeaway from both events was the importance of getting the fundamentals right.
As Bernard’s experience at ENGIE demonstrates, implementing AI successfully can involve developing new processes and learning as you go.
And as Miriam highlighted, even when organisations have access to valuable customer intelligence, they still need to make sure that information reaches the people who can put it to good use.
There are also important questions to answer around how automation is measured, how AI systems are governed and whether the technology is genuinely improving the customer experience.
With so many organisations now investing in AI, perhaps the question for 2027 shouldn’t be how much more can we automate, but how much value are we actually getting from what we’ve already introduced?
With so much changing across the contact centre industry, it’s always interesting to hear how other organisations are approaching challenges like AI adoption, automation and making better use of customer data.
If you’ve got experiences or ideas you’d like to share, why not join the Call Centre Helper Reader Panel?
It’s a chance to have your say on the issues affecting contact centres, contribute to our industry research and help us explore the topics that matter most to people working in customer contact.
Reviewed by: Jo Robinson