From Cost-Cutting to Customer Growth: A Smarter AI Strategy - with McKinsey partner Tey Bannerman

For 20 years, we've been creating innovation in the CX industry, and now we're seeking out brilliant new perspectives on CX you just won't find anywhere else. I'm Richard Owen. Welcome to the CX Iconoclast. In this episode of the CX Iconoclast podcast, I'm welcoming Tay Bannerman to the show.

Tay is a McKinsey partner focused on deploying AI in large enterprises to improve customer retention and growth problems. so pretty much right at the center of the customer AI universe. Tate likes to emphasize that AI could solve real human problems rather than just serve as a technological novelty, and he's pretty critical of the reflexive use of AI for cost-cutting and job elimination, instead urging companies to prioritize long-term value creation and enhanced customer experience.

Hear, hear. In this interview, he shares a compelling case study of a B2B payments company that successfully shifted from cost-cutting orientation to a customer-centric strategy using predictive analytics to reduce churn. He's also highlighting the importance of engaging frontline employees in AI design and adoption, a critical lesson that I think we all need to learn. Our conversation moves on to address the dual challenges of employee resistance and leadership skepticism.

Tay notes that fear of job losses and poorly designed tools can truly undermine adoption unless employees are engaged as co-creators. At the executive level, the challenge is often a cultural reliance on instinct over data, especially when it comes to dealing with predictions. He advocates for a balanced approach where leadership combines intuition with data to scale better decision-making. And he does give us one really great quote.

Predictive data gives leaders scale. Intuition alone simply can't reach 100,000 customers. Hear, hear. Enjoy the conversation.

T. Bannerman, thank you very much for joining us here today. T. Bannerman, Thank you for having me.

T. Bannerman, Thank you for having me. T. Bannerman, So let's dive straight in because I think we have a lot to talk about in terms of what your perspective is and also by implication some of the perspectives McKinsey holds on the whole topic of artificial intelligence.

One of the things that you talked a lot about, which I think will resonate with a lot of people here, is that the objective ultimately of artificial intelligence is going to be to serve humanity by solving real business needs right address human concerns rather than simply you know be interesting for its own sake um could you elaborate a little bit more on that how you see that as a topic yeah i mean it's a it's a good opening question i i think i if i just think about technology in general and the path that technology has taken i think historically you've had technology that's been released or that's been developed that's interesting but doesn't serve a real purpose in terms of humans having a problem that needs to solve an opportunity that we've latents and and then if you just carry that forward ai is the same it's another technology albeit a very powerful one but the companies the products that succeed are going to be the ones where ultimately consumers say i need this i'm going to use this and the ones that fail are the ones that you know that when nobody uses them and i think that goes for you know whether it's an internal tool or platform that a company is bringing in or something that's a consumer product and so i don't think that stands in revolutionary i think it's just us refreshing that in the age of ai and in the current hype cycle that we're in don't you worry that the opposite turns out to be true in terms of tech i mean 20 years ago did people say hey we need a better ad serving engine because i really want to be served more advertisements or i'd really like some platform to curate my news so that i live in an echo chamber and i'm not exposed to other ones right so we've seen plenty of evidence that successful technologies that make people a lot of money arguably don't advance humanity yeah yeah no that's that's a a great point i i mean i think what what's kind of troubling is those companies essentially feeds upon the fact that our um how do i have not put this dramatically they see upon the fact that human attention is fallible and and there's a lot that you can do in terms of making people addicted to things and addicted to consuming content and then if you can monetize that then you essentially kind of see the kind of flaw in human biology and so i think there is definitely a concern around companies continuing to discover and see upon those flaws to say hey how can we make people more entrenched in kind of you know being addicted to content being in those echo chambers i think on the other hand you will see companies that go hey we have evidence that this is not good in terms of people's outcomes in terms of young people's outcomes and so on and we're going to create technology that kind of is built on different values and i do think that we are already seeing a shift with consumers where they go we are not going to use those platforms that take us into those echo chambers that cause addiction in kind of young people and i think that will continue where you will then have these ai companies that are developing models and so on kind of have to move because consumer kind of demand and sentiment is is changing yeah that's an optimistic view and i i mean i obviously hope you're right i mean i think that there's if there's money to be made, right, there's going to be companies that want to fill the gap.

And as you said, playing on the fact that we're, in some ways, we're so unfamiliar with technology still as a species. We've got all this evolution behind us, and now we're being bombarded by this. Just take a simple example. So we're used to taking our cues from body language, from visual cues, from voice.

If artificial intelligence can create highly accurate and convincing simulations of sound and video, which it most certainly can. Yeah, yeah. Right? Aren't we fighting our basic human nature to sort of say, wait a second, this is fake.

And we're going to take, it's going to be very hard for us to overcome that. I mean, look, I'm not here to make a case for regulation, but I'm here to make a case that we need more white hats than black hats out there right now. I fully agree. I fully agree.

And I do think you're absolutely right. technology is moving at a pace faster than humans can discern. And I think that is definitely one of the things that keeps me up at night, so to speak. You already have AI technology that can very easily be used to manipulate people in ways that they can't even discern.

And to your point, having more people in the space that are good and that kind of, you know, and that kind of build in, you know, the ethics, the morals that kind of, you know, humanity has thrived upon, I think is going to be critical. And I think, you know, I think there probably needs to be a movement to get people who aren't just technologists closer to building and evolving these models. Now, maybe there's a finally room for the person who did that philosophy degree. Philosophy.

That finally pays off. You know, your mother said it was a waste of money. Like, you know what? Look at me now, mom.

Yeah, you should have been a programmer. It turns out, no good. No no we don need those anymore When you think about the corporate side of this I mean the sort of equivalent is this drive towards cost elimination as the first instinctive reaction right now look let not be naive if you a business and you're trying to achieve highly more efficiency and there's all this pressure to accomplish it you'd be a fool for ignoring the potential to essentially create efficiency within your organization yeah but i wonder whether it's just so reflective reflexive i should say that you You go off and you try and say new technology equals people elimination.

And that sort of deprives you of the opportunity to start thinking in terms of how can you actually create growth and productivity from technology? Yeah. And I do think that I think the pattern that I typically see is you have executives in a position where they've been given a mandate or where they're under time pressure to deliver results quickly. And I think when you see that pattern, the natural answer is cost reduction or chasing that headline or chasing the next quarter.

I think when you do have executives and leaders who either aren't in that vicious cycle and have an opportunity to step away from it or are in a corporate culture where there isn't that pressure, I think then the most successful ones go, okay, where do we want to be? And what is the gap between where our organization is now and where we want to be? And then what is the cost of complacency? And I think that's where you see companies really go, we need to focus on customer experience and you need to focus on value creation because that is ultimately what is going to kind of, you know, make us relevant in three, five years.

Whereas the short term is not going to make us relevant in three, five years. do you see good examples now emerging of companies that are starting to think broader about this and start to see this potential to deploy this for upside I mean you don't necessarily have to name names but is there an emerging cadre of companies thinking a little beyond the whole notion of cost reduction of course that's a big theme we want to explore in the conference yeah yeah yeah and I think I mean I'm not going to name it, but there's a company that I've been working with over the last couple of years.

And they essentially emerged from kind of a cycle of transformation and cost reduction where they've done a lot to cut costs. They had new leadership in and that new leadership was transformative in terms of one, laying out a strategy that was focused on value creation. but then two saying we're going to thoughtfully implement technology and thoughtfully implement ai where it makes sense to with not just a focus on okay efficiency and cost reduction but on fundamentally going how do we create new value for our customers and for the organization and so what they did was and so this is a payment company b3 payments so they spent three months kind of with their, and this was top leadership CEO, with customer service representatives, having weekly sessions, understanding the problems that their customer service representatives faced and understanding how those customer service representatives drive loyalty and drive value in their existing customer base.

Because from a strategic perspective, one of the main levers that have been identified is trend reduction. How do we reduce trend? and so what they ended up implementing was twofold. On the one hand, it was a predictive analytic system that predicted churn, that predicted which customers would lapse, but also predicted why, whether it was likely to go to a competitor or so on, and then gave the customer service representatives predictive signals as to, okay, here's when you should reach out because reaching out ahead of time is better than reaching out when the contract is already over and what is the information that you should give them.

But then the other side was treating your customer service representatives as a customer as well. How do you create an experience that gives them confidence when they're having these conversations? And so that meant servicing the right information to customer service representatives as well and that fundamentally shifted how the company thought about who are our existing customers, how do we serve them, but then also how do we make sure our customer service reps, who are that front line, are using the data and insights that we can discern from AI and these models in a way that ultimately builds something that is not as measurable, confidence.

I thought that was a really fascinating journey that the company went on from pure focus on cost reduction to kind of, you know, what I've described. And I think the question that drove them was essentially in five years time, are our customers going to be happy with the level of personalization and the level of service that we give them today? The answer was no. And I think the answer for most companies is no.

And then they went, what do we need to do now if we don't want to be in a situation in five years time where we're playing catch up with everyone else? I mean, I think that's pretty far-sighted in part to be, I love your point at the end there saying they thought five years ahead, right? Which is a really interesting question to pose. I think it also reframes the conversation because if you give yourself a bit of runway, right?

If you say to yourself, all right, five years is a long enough period where we can change anything in the company. Yeah, yeah. Pretty much everything's in play over a five-year period. You sort of stop the debates as to whether or not, oh, can we do it?

We have the resources, right? Over five years, in a big enough company, anything's possible. But it's this mental switch towards then thinking about a future state which delivers much better outcomes for customers, but also importantly this idea of engaging the employee base. This is going to be a challenge, isn't it?

Because if we just go to people and say, hey, well, first of all, you've got a large part of your audience within the company thinking AI equals it's going to take your job. And by the way, they are bombarded with that message every single day. Right. So.

AI is going to take your job. Hey, we've got a new AI project. Would you like to be part of it? We'd really like to explain how to make the machine more knowledgeable about how you do your work.

We'd like to transfer some of your expertise to it. Who wants to sign up first? Right. And we're going to be surprised that employees' reaction to that are going to be, you know, hell no.

Yeah, yeah. So we've already got this negative sort of environment that we're projecting around this. Absolutely, yeah. We have to fight to overcome that, don't we?

I mean, the tone in which companies approach this and how they educate their employees is going to determine success for them. Because if they meet resistance, this is not going to work. Yeah, absolutely. But I don't think this is new either.

I think for years, companies have been trying to implement AI tools and kind of AI models and receiving resistance from employees. I work with an insurance company that kind of implemented kind of models to help their insurance service reps basically kind of give better recommendations to customers and literally almost 0% adoption. They completely ignored the AI recommendation and went no I gonna use the tools and the intuition that I kind of yeah I have And that was a platform designed by engineers and data scientists There was no work done to sit with the customer service reps to really understand what are the frustrations day to day?

What do they really actually need? Because if they had done that and we had come in to essentially redesign and fix that, they would have understood that again you know customers the customer service reps it wasn't the the accuracy of the recommendation that they had problems with they had problems with things like being able to easily access kind of you know a customer's data when they you know when a customer called and it was currently taking 10 minutes and they had sticky notes everywhere and so ultimately if you're bringing a technology that that isn't solving those problems and you haven't designed this with your employees, you will get that rejection.

And to your point, there's also a lot out there about AI taking jobs and so on. And kind of the organizations that I've worked with where, you know, they failed drastically in terms of trying to bring in AI tools and have employees adopt them is when they had not done anything at all to kind of one, understand employee concerns, and then two, kind of, you know, start to not just educate employees, but show them where these tools could help them in their day-to-day, help them be better, but then also reassure them that this isn't about replacing you.

This is about helping you do your job better. You know, and there's a little difference in the case of an AI technology. I appreciate your point that in some ways this is nothing new. We've had multiple waves of technology, and each wave has been accompanied by some job displacement.

I mean, it's been inevitable. AI is a little different though, because, you know, if you think about the very, at the heart of the concept of machine learning is learning. And if, and if, if you want to build effective solutions, you have to engage the organization in an act of essentially teaching, right? In some ways, we are all transferring our knowledge to the machine.

And it's not like you're putting in place, you know, a payroll system. Yeah. You switch it on. It works.

Hopefully it does what it's expected to do. You switch on these systems. They are like a sort of five-year-old child that they got great potential. It's very good, right?

They need coaching. And if you want to be successful in bringing them up to their full potential, the organizational needs to see the potential in that. They need to see, wow, if we invest in this five-year-old, we're going to end up with a super smart 20-year-old any day soon. And we're not even going to have teenage kid problems.

You know, that's worth an investment. But also, you know, we want them to feel ownership for that investment. And this notion that you gave where frontline team members were involved in this seems like an oversight we've made in previous systems iterations. And we run the risk of doing the same this time.

So I'm fully on board with that. It just seems like it's a different mental state. What the example you gave was the long-term view and also the employee engagement just seemed like two standout success factors. Yeah, but you're right in that I think what makes this time around different is that in some cases you are training the system with the knowledge from the humans that it will, in some cases, replace some of the tasks that they were doing.

But I think that, I mean, we're already seeing instances of companies kind of, you know, reversing some of the decisions they've made around kind of going all in on AI, replacing huge chunks of their workforce. Klaner is kind of, you know, the most famous example right now, where, you know, that was a combination of kind of, you know, employee resistance, but also realizing that the technology that they kind of introduced wasn't as good as the humans it was replacing. and you know exactly to your point because the inherent knowledge that kind of you know a lot of the humans possessed wasn't just around the the kind of you know the emails they sent and the the kind of the actual outputs but a lot of the the kind of the intangible kind of you know relationships they had and so on and the ability to kind of understand um someone beyond just the you know the email you've had them to them so i i do think again the companies that approach this from a perspective of kind of how do we see this as almost an ecosystem and how do we make sure that kind of this is pushing towards outcomes that help employees that help the organization and not purely profit and efficiency i think when you take that as your kind of you know your foundation i think you tend to kind of you think about okay what are the adoption programs it can bring in how do we thoughtfully kind of have people kind of, you know, willingly contribute the knowledge that helps them as opposed to kind of, you know, just fully resistant and be apathetic towards the change.

Let's go to the other end of the spectrum because whilst you might think, okay, there's resistance from frontline employees because they're fearful for their jobs. I could make an argument. There's more resistance from leadership in companies. because they're inherently used to making decisions based on instinct, right, and pattern matching from previous careers.

You hire somebody at the top of a company because they have experience, and many people interpret experience as simply pattern matching. Well, I've done this job before. My field says do it this way. And now we're sort of introducing data sets.

And first of all, a lot of companies have struggled with data-driven decision-making, period, right? They're just not that culture. Now we're introducing data sets that are not even showing people actuals, but now we're starting to show people prediction. Prediction, yeah.

And the word prediction has these connotations in people's mind that means unsafe or unreliable, right? And so we're going to executives and saying, you want to make your choices, maybe choices that are counterintuitive for you based on algorithms that assign probabilities. I mean, that's going to have a mixed reaction, right? Absolutely, yeah.

Resist that or go with it. I mean, it's going to really divide the next generation of leaderships, you could argue. Absolutely, absolutely. And I mean, I do think that, I mean, intuition absolutely has a space.

I don't know of any company that makes decisions purely on data and not without kind of, you know, leadership. using their experience and expertise, the skills that they've developed over time. But I do think that where I've seen leaders turn is where you explain to them that this is not about losing your intuition, but this is about scale. So your intuition around strategic direction, around the needs of your organization, that's why you're there as a leader.

but there's only kind of so much that you can do. You can have knowledge about 100 of your best clients but you have 100,000 clients. The data, the predictive data gives you that scale as a leader to combine with your intuition to go, okay, I am going to amplify myself 10, 100 fold. And I think the instances I've seen that light bulb turned on is when I see leaders start to kind of, I guess, be less skeptical about the predictive modeling, about the predictive data, because they go, like, it's either I do nothing with these kind of other customers, or I rely on the data and the insights there combined with my intuition to do something I otherwise wouldn have at all does that make sense yeah it does make sense to me i mean i think that the the we know that people have a lot of biases right we're yeah we're basically human and so we have all our favorite biases you pick the one you like the best i mean you know i think that you've got a lot of confirmation bias is probably the biggest one i see in executive teams right and In that environment where we have these very natural human biases, probabilities are a bit of an alien concept for people to get their heads around.

And at the end of the day, if we're talking about AI, we're talking about probabilities. A prediction is a probabilistic outcome. And we know that people have trouble understanding even probabilities. And that's not exclusive to less well-educated people, in fact, at arguments.

It doesn't correlate to degree of education. This idea that if you say something's 60% likely to happen or 90% likely to happen, how do people interpret those numbers? How do they process this idea of lack of certainty? And how do they incorporate that into their decision making?

And this is a whole generation of leadership. Remember, I mean, the guy's running companies, probably my age. you know and and if they they grew up in business in an era where they made their successes as you said they got there because of you know hard work dedication people skills intuition and ability to read the room right all of these skills and that was saying all right we've got this all of these probabilistic outcomes that need to be injected into it and without becoming a data science expert you have to incorporate that into your thinking Yeah.

Let's say for the sake of argument, I don't believe this is true for an instant, the business schools start teaching this effectively today. We're still 20, 30 years away. Yeah, yeah, yeah. Yeah.

No, you're right. You're right. And I do think there's going to be a gap for a good while. And I don't see kind of a quick way of closing that gap.

I do think that part of it is cultural. I think there are companies that have a culture of data informed decision making across the organization. And those companies tend to use kind of, you know, the data inside more, more so than others. But I remember reading a stat too that like in most organizations, a minimum of 55% of the data that they collect goes completely unused.

and I do think that if you can build a bridge between here's the data that we have here's the decision that we're trying to make and here are tools that essentially fill in the blind spots that you might have in your intuition or your experience to help you make better decisions. I do think that a lot of it is about essentially how you design that experience because I think a lot of the data we have is kind of very raw. You have dashboards, you have tables and so on which are hard to engage with and how to like make real decisions using so i do think if this all comes again down to that experience that you're creating yeah look i actually think the point you're making on how ai can generate language and communication in much better ways synthesize information and right i mean this is partly why we've seen such an aha moment with generative technologies, right?

You've got behind it a valuable engine in a large language model, but you have this natural language processing interface, which has made it seemingly more accessible. But I also wonder, the flip side of that is, if the presentation layer becomes much better, right? If the machine literally talks to you and says, okay, here's the answer, are we going to, in some ways, stop asking the question as to where did it get that answer from, right? Are we are we going to be seduced by the attractiveness of the information medium, right?

And so all of a sudden, it's a race to present bad ideas in beautiful and elegant ways. And so if the machine is extremely adept, and again, this is the danger, extremely adept at presenting ideas in a very natural kind of sounding way, then you do switch off the part of your brain that says, why is that true? or the curiosity. When you see charts and data, I don't know about you, but what I immediately start to look at is, well, how's the axis designed, right?

Okay, so I'm probably not typical of humans in that regard. But how am I being informed here? Is the calculation correct? But if that's taught to me in a very natural language kind of way, it's easier to understand, but it's also in some ways, Maybe it kills our skepticism that we need to have if we're going to be good at processing data and information sources.

This is new territory, isn't it? Yeah, absolutely. Absolutely. And I do think kind of it is a new paradigm around making information more accessible and more digestible by humans who ultimately communicate kind of with words.

And so I do think that's kind of going to be a hugely powerful kind of like need in terms of creating interfaces for decision makers to translate, navigate the complexity of data and disparate data that they may have. and to connect it with what should I do or what could I do and then help me understand the rationale, help me kind of, you know, really, really kind of be able to use this insight in a meaningful way that aligns to my context. I do think that is kind of going to be a paradigm shift in terms of how we can understand and use data.

And to your earlier points, people of all walks struggle to understand numbers, probability data, And if you can make it more accessible, I think you can really change a lot around data and insight literacy. Well, that's a great place for us to wrap up, I think. It's an optimistic story. But one of the things I take away from this whole conversation is, like every wave of innovation, it's going to create winners and losers, right?

I mean, it's going to amplify that. And maybe this is the greatest amplification. Maybe this is the greatest sorting we've ever seen since Harry Potter. I mean, this could be the big sorting here, when companies over the next five years are really going to be separated out by the choices they make.

And that's got to be an exciting idea, especially if you're on the right side of this, right? And you're an advocate. So one final question before we wrap. next year we're talking about predictions where's everything gonna finish in the league because I think now we're gonna see whether predictions really are useful or not what are we saying give me a number this fifth you heard it your first no algorithms involved in creating that whatsoever yeah human with all of the activity today it's been absolutely fantastic really enjoyed talking to you.

Appreciate it. Thanks. Thank you so much. Cheers.

Thanks for listening to the CX Iconoclast from OCX Cognition. Subscribe wherever you get your podcasts so you won't miss any of our thought-provoking conversations. And please get in touch if you want to learn more about what OCX Cognition's predictive CX analytics platform can do for your business by providing complete insights into every account continuously updated and connected to operations. You'll find contact info in the show notes.