The changing role of the AI-enabled contact centre

The changing role of the AI-enabled contact centre 1260 720 Kane Simms

I was recently at an event organised by Kore.ai in The Shard. A lovely spot where I had the pleasure of facilitating a workshop on large language models vs small language models alongside Cobus Greyling.

One of the other speakers was Stuart McCann from Boston Consulting Group, who shared some research and insight into the future potential operating model for call centres that are enabled by AI. There was some great data and perspectives there that I think will help you adjust your mental model to the future of service delivery.

Removing Tier 1 Support

The first point Stuart made was that the first line of response in contact centres will be taken away. This is the typical, repetitive, simple stuff that customers should be able to handle themselves. Tier 1 support.

We’ve spoken numerous times about this here at VUX World which is the Transactional phase in our use case maturity framework.

Even today, most businesses could automate their tier 1 support without AI, quite honestly. AI just bridges the gap to help plug conversational channels into those existing automated processes. Either that, or it gives you the impetus to automate those processes in the first place.

What’s left is Experts to resolve more complex issues, Supervisors to oversee proceedings and operational management to run the show (with AI ops folded into this).

Running the service vs continuous improvement

The second thing Stuart highlighted is that, today, 90% of contact centre work time is spent on running the service. That’s keeping the lights on and getting through work.

With the introduction of more AI tooling and the removal of Tier 1 support, AI starts to handle more of this day to day work. This means that the Experts, Supervisors and Operational Management folk can spend most of their time continually improving service delivery and working on more long-term, strategic goals.

We’re already seeing this shift within organisations that were earlier out the gate with first generation chatbots. People who used to work on the phones are now operations managers for the chatbot and spend their time designing conversations that’ll impact thousands of people, continually improving those conversations to increase performance.

From reactive to pro active

Most of what the contact centre handles today is reactive support work. Something goes wrong, people call. Many organisations have been working for some time to reduce this support work by introducing self-help and self-service solutions.

Working backwards from self-service, Stuart shared the concept of self-heal. This is where systems and processes, using technology and, where relevant, AI, can fix problems before they hit customers and generate the need for self-service or support.
In order to do that, you need to be able to pre-empt the problem by understanding when it’s about to happen.

This then, is the ideal service model that contact centres should seek to adopt in future:

  1. First, pre-empt the issue before it happens
  2. Then, automatically fix the problem before it arises, or before the customer is aware/effected.
  3. Failing that, have self-service solutions available.
  4. When that doesn’t work, bring in your people.

Things that need to change

In this model, you’ll need to rethink a few things.

  1. The people that now answer those calls need to be experts. If your technology has failed and people can’t service themselves, it’s because the issue is sufficiently nuanced and complex. In future, more advanced reasoning agents may be able to take some of this off your plate, but in the medium term, you’ll need expert people.
  2. The metrics you use to judge service quality need to be adjusted. Average handle time becomes pointless when every one of your contacts is complex. Your AHT will go through the roof as soon as AI takes on Tier 1 support. You’ll need to focus your efforts more on reducing customer effort and increasing resolution rates.
  3. Your operations people need different skills. Managing an AI implementation is similar to managing a team of people. They need monitoring, analysing, training and to develop new skills. You’ll need people that understand generative AI, prompt design, prompt engineering, NLU design, conversational experience design, data analytics and more.
  4. Needless to say, you’ll need a whole bunch of new technology to accomplish this. Yes, some of that will be AI and generative AI-based, but you’ll also need data storage, knowledge management, system connectivity, workflow creation and journey orchestration, data visualisation and analytics, a whole bunch of software to monitor, flag and resolve issues as they arise, and much more.
  5. This all makes for a pretty exciting time in the contact centre over the next few years. And while generative AI has a little way to go, it’s starting to get to a place where businesses are churning out production-grade applications. So this all might happen faster than you think.

    Share via
    Copy link
    Powered by Social Snap