Cavell’s analyst team recently returned from Enterprise Connect. It was a three-day event with dozens of keynotes, scores of meetings, and non-stop conversations!

Cavell has already covered most of the news releases from Enterprise Connect, so I wanted to take some time to bring you some of the broader CCaaS/CX trends we discussed across our briefings and saw on stage in keynotes. This year’s broad overall theme was ‘things that AI can do/is doing’, compared to previous years, which were more focused on potential AI features.

Customer Journey Analytics (CJA) was a topic at the forefront of discussions at the show, mainly linked to the new insights that can be gained by using LLM-powered intent and sentiment analysis and combining different databases and data sources.

However, like many of the latest technical developments, the art of the possible does not yet match the reality for most businesses, with many finding the prospect of adopting a unified CJA approach quite daunting. There is a significant need for data and knowledge management, creating a challenge to integrate different platforms that might silo data from each other, and you also have to build tools that understand specific industry and customer persona intents and sentiments.

This discussion mirrors what Cavell has been hearing from many of its channel contacts, who report that much of the industry’s adoption of these new methodologies is still hampered by knowledge silos, and many companies are still at the ‘data management transformation stage’. This is also reflected in Cavell’s latest survey of 2,000 telecoms decision-makers, where 34% of respondents had deployed or begun to trial AI for data analytics. This demonstrates they are starting to deploy these systems to help manage and transform their approach to data and analytics.

Looking at the long term, the huge potential upside for companies that can harness detailed CJA is that this is an area that will see widespread adoption in the short to mid-term.

Multiple discussions emphasised that not every prevailing wind in CX is pushing businesses towards the cloud.

Companies reported concerns over regulations, compliance, and losing control of specific key applications as reasons to keep an entire Contact Centre or some aspects of the Contact Centre in an on-prem deployment model.

While the overall agreement was that cloud Contact Centre would continue growing, many people point out how the cloud might be the future, but hybrid is still the reality.

This matches Cavell’s latest end-user research, where 37% of contact managers surveyed said they still had their Contact Centre deployed on-prem, and 35% had a hybrid model. However, the same companies reported that they expected to migrate further to the cloud within the next two years.

So, we still stand on the precipice of cloud migration. Although, there are some solid arguments for continuing with on-prem CX, and maybe even some additional arguments emerging, such as the need to perform specific AI processes on sensitive data in edge computing models instead of a public cloud deployment.

As I mentioned in the opening of this article, this show was a ‘stand-and-deliver’ event on AI for many attendees.

At the show, we saw refinements of many of the agent assistance tools launched last year, such as increased accuracy in summaries and transcriptions and more usability in those transcriptions, such as the ability to take actions directly from the transcription into a task manager.

The show also saw an increase in the number of front-line AI-powered tools, including virtual receptions and AI-powered IVAs to handle frontline queries. Two recent examples can be seen from Cisco and RingCentral. The RingCentral AI Reception was announced at the recent RingCentral Analyst Summit  which you can read more about here.

A large number of tools are also still being rolled out on the back end, such as intent mining, Gen-AI powered bot workflow generation, and many highly impactful low-risk tools that will make the management and operation of Contact Centres much easier.

Alongside the changing capabilities for analytics, we also discussed how we determine the success of the Contact Centre as the nature of interactions and the data that can be gained from them change.

Many of the conversations that were raised about metrics focused on average handle time and time-to-resolution, which have been strong legacy measurements of success. These metrics are becoming less popular when you enable more automation and self-service for simple queries. Your remaining queries will take longer to resolve on average, but they will also likely vary more as levels of complexity differ.

When we surveyed 800 Contact Centre decision makers in 2025, 24% said that first contact resolution would still be their top metric for measuring the success of AI, but 24% also said that their top metric would be the resolution rate of complex cases.

However, two newer metrics — customer effort score and agent confidence — were also discussed at Enterprise Connect. They used to be much harder to calculate and track, but they have become much easier with new approaches to data and analytics.

Customer Effort Score measures how much work the customer has had to do to engage with your CX ecosystem. This requires understanding the full customer wording journey, e.g., what they have done on social media, what they have done on the phone, and how many times they have called. However, once you have this data, you have a very helpful understanding of potential customer frustration and the effectiveness of your CX ecosystem.

Agent Confidence measures your agents’ confidence when dealing with a customer query. This is useful because it measures everything from how good the data and assistance tools you’re providing to whether your intelligent routing solutions are picking the right agent. One of the reasons this is also seeing a resurgence is because, using sentiment analysis and AI-powered data analytics techniques, you no longer need to rely on agents to self-report confidence, as you can analyse it automatically without reporting bias.

These two metrics are holistic; they pass a broad judgement on the quality and health of the Contact Centre interaction from both the agent and the customer side.

It’s hard to distil an entire week’s worth of learning into one short blog post, but I hope this post has given you an idea of some of the CX market trends seen at Enterprise Connect. Of course, if you have your trends or any more profound questions about the trends I’ve shared, please do get in touch!