The AI transformation gap: why better technology demands better leadership

AI is changing customer operations faster than many organisations are changing the way they lead them.

We're investing in technology that can automate routine interactions, coach people in real time and uncover insights that were previously difficult to access. But if we're still measuring performance against yesterday's targets, managing people in the same ways and struggling to act on the information we already have, how much transformation are we really achieving?

It's a question I explored during two panel discussions at Centrical's The Big Game in London, where we discussed the changing role of AI, frontline performance and operational leadership.

Having spent more than 25 years in customer operations, on both the client and outsourcing sides, I've seen technology change almost every aspect of how we work. What hasn't changed is our responsibility to deliver better outcomes for customers and create an environment where our people can succeed.

And that's why I believe the biggest challenge in AI transformation isn't necessarily the technology. It's whether we're prepared to rethink the operating models, performance measures and leadership behaviours surrounding it.

AI is changing customer operations. Leadership must change with it.

There are two areas where I see AI making a particularly meaningful difference.

The first is removing routine activity that doesn't necessarily require human involvement. Identity verification is a good example. Where the data is available, the systems are connected and the process is straightforward, technology can handle routine verification efficiently, freeing colleagues to focus on more complex customer needs.

The second, and perhaps more exciting, is enabling people to achieve things that weren't previously practical.

Take training. Traditionally, an advisor joining a contact centre might complete two or three role-play exercises during their initial training. With AI-powered coaching, they can practise scenarios repeatedly, learn from their responses and build confidence before handling live customer interactions.

At Firstsource, we've seen AI coaching reduce time to competency by 20–25%. That's a meaningful operational improvement, but there's a bigger opportunity here. We're not simply accelerating training. We're helping colleagues become better prepared and more confident in their roles.

The real question is what we do with that capability.

If AI takes on more routine work, the interactions reaching our people will increasingly require judgement, empathy and problem-solving. That changes what we need from our frontline colleagues and, importantly, from the leaders supporting them.

We cannot transform the work without reconsidering how we lead it.

Rethinking what good performance looks like

One of my longstanding frustrations in customer operations is our reliance on traditional performance measures, particularly average handle time (AHT).

AHT still has a place. It's useful for workforce planning and understanding capacity. But I've never believed it should be treated as a measure of whether we've delivered a successful customer outcome.

Consider two interactions. One colleague resolves a complicated customer issue in ten minutes. Another completes the interaction in five, but the customer has to contact us twice more before the problem is resolved.

Which represents the better performance?

If we're only looking at handling time, we're missing the point.

As AI increasingly manages straightforward enquiries, this becomes even more important. The work left for human colleagues is often more complex, and we need to recognise the value of getting those interactions right rather than simply getting through them quickly.

We should be asking better questions. Why did the customer need to contact us? Was their issue resolved? Did we deliver what we promised? How many times did they have to come back?

And we need the right information to answer them.

Many organisations still operate across fragmented systems, making it difficult to understand the full customer journey. The quality of AI output will only ever be as good as the information and context available to it.

At Firstsource, we're developing our Intelligent Context Framework (ICF) to help connect that information across platforms and give our people and technology a more complete understanding of the customer.

But better data is only part of the answer. Insight has little value unless someone can act on it.

That's where leadership, coaching and frontline autonomy become so important.

Frontline autonomy requires a different kind of leadership

Early in my career, I managed a team of 46 people processing paper registration cards for pay-as-you-go mobile phones.

The daily target was 260 cards, but one colleague regularly managed 360. It challenged me to think differently about what motivated the team. We introduced a simple incentive: reach a higher target, maintain quality, and you could finish early.

It worked because we understood what mattered to our people and gave them greater ownership of their performance.

Today, that kind of manual processing has largely disappeared. Technology has transformed the work beyond recognition. But the fundamental leadership challenge hasn't changed nearly as much.

We still need to understand what motivates people, give them the tools to succeed and create an environment where they feel trusted to make a difference.

And I think there's a particular responsibility on team leaders.

I don't believe you can lead a customer-facing team effectively if you've lost touch with what it means to serve the customer.

Team leaders should understand the products, policies and processes their people work with. They should be able to do the job themselves. And they should be close enough to their teams to understand the challenges they face every day.

I'd like to see every team leader spend at least one day a month handling customer interactions and being assessed against the same quality standards as their colleagues.

Not because we need another compliance exercise, but because that experience builds empathy, keeps leaders connected to the reality of the operation and helps them identify where things need to improve.

AI can help us here too.

We can analyse conversations, identify coaching opportunities and deliver feedback directly to an advisor's desktop, often far more quickly than traditional quality assurance processes allow. Colleagues can review that feedback, acknowledge it or challenge it, becoming active participants in their own development.

That's a very different proposition from simply using technology to monitor performance.

It's about giving people the information, confidence and autonomy to improve.

And there's an important distinction between measuring the quality of an interaction and knowing whether we delivered the outcome the customer expected. An automated quality assessment may tell us that a conversation was handled well. It doesn't necessarily tell us whether the promised action happened afterwards.

That requires curiosity, accountability and leaders who understand what happens beyond the conversation itself.

Making AI transformation work at scale

Of course, empowering people is only part of the challenge. Organisations also need to deliver consistent outcomes across teams, clients and geographies.

This is where a well-designed operating model and Centres of Excellence can make a real difference.

For me, the value of a Centre of Excellence isn't simply standardising processes. It's creating a foundation of expertise, consistent ways of working and proven practices that can be shared across an organisation.

In a global operation, that means establishing clear standards for areas such as quality and workforce management, while recognising that different clients and geographies will have their own requirements and cultural nuances.

It also means understanding our responsibility as an outsourcing partner.

Clients rightly expect us to perform within their systems and meet their operational targets. Delivering against those commitments is the starting point. The additional value comes from applying our experience to identify better ways of working, challenge existing processes and bring forward opportunities for improvement.

AI creates new possibilities here, but it doesn't remove the need for human judgement.

I recently saw an agentic AI solution developed by our team for a retailer handling order enquiries. What struck me was how naturally it responded to the customer, including recognising frustration and responding with empathy.

But there comes a point where a conversation moves beyond providing information. A customer might dispute a decision, request a financial adjustment or raise an issue requiring additional judgement or investigation.

That's where the handover matters.

Routine processes can be automated, but when an interaction requires human discretion or an exceptional decision, responsibility must be clear. Whoever takes ownership needs the context, authority and information to resolve the issue, not simply another task passed between systems.

The strongest operating models won't be those that automate the greatest number of interactions. They'll be those that understand where automation adds value, where human judgement is essential and how both can work together to deliver a better experience.

The real test of AI transformation

There's enormous opportunity ahead for customer operations. AI can remove unnecessary work, accelerate development and help people make better decisions. But achieving its full potential requires us to change more than the technology.

We need to rethink what we measure, give colleagues greater ownership of their performance and develop leaders who remain connected to the people and customers they serve.

Above all, we need to be prepared to challenge established ways of working rather than simply automate them.

Because the real test of AI transformation won't be how much technology we've deployed.

It will be what our people and customers are able to achieve as a result.

Turning insight into impact

Discover how Firstsource combines AI-enabled technology, operational expertise and human insight to transform customer experiences.

‍

‍

‍