What you need to know:

• NAB’s ‘Customer Brain’ transformation, now closing in on three years of development, is revolutionising engagement. Chief Data Officer Christian Nelissen and his team have spearheaded a centralised, enterprise-wide decisioning engine to unify customer interactions across channels in real-time.
• Politics, not just tech, is an important challenge. Nelissen says the real work lies in overcoming organisational turf wars, conflicting KPIs and fragmented customer ownership,  it's not just building the tech stack.
• From outbound blasts to real-time personalisation, NAB has shifted from mass outbound campaigns to in-app, in-session activation, leveraging over 100 million monthly app logins to engage customers contextually as they interact.
• Propensity modelling is out, triggers are in with NAB preferring real-time behavioural triggers over traditional propensity models, allowing for highly defensible, explainable decisioning, crucial in tightly regulated banking.
• Hyper-standardisation keeps hyper-personalisation scalable. Strict rules of engagement prevent campaign chaos, enabling 50-plus business-led engagements while avoiding customer fatigue and duplication.
• Banker conversations have been rewired, with Customer Brain feeding contextually relevant leads that boost product consolidation and deepen relationships.
• The program has expanded across NAB’s brands, scaled business adoption from 10 per cent to 70 per cent, and is central to CEO Andrew Irvine’s customer-centric vision.
• NAB’s focus remains on controlled GenAI adoption for creative variation, while maintaining explainability to satisfy regulators in a tightly governed sector.

NAB is into its third year of building what it calls its 'Customer Brain', which in turn has helped it to overhaul its customer engagement strategy. That's seen it move from batch campaigns to real-time activation, from static segments to adaptive triggers, and from product selling to milestone-driven relationship building.

The result is a personalisation model that feels intimate at scale and one that’s built to survive the political machinery of a large bank. And indeed it is navigating the politics of decision-making that has proven to be one of the most important challenges.

“People don't fundamentally understand that a lot of the stuff, even though it's very, very tech heavy, is still just a people/ political problem,” says Christian Nelissen, NAB’s Chief Data Officer. 

Getting to the point where the Customer Brain powers real-time engagement across millions of customer interactions, and spans the full range of products, channels and business unit requires more than just enterprise architecture. It demands organisational discipline, standardisation, and a willingness to fight functional silos according to Nelissen, who was speaking at a media roundtable in the recent Pegaworld conference in the US.

"You’re really forcing the issue of who owns the customer," Nelissen says. “In a complex environment like ours, it's not really obvious who owns the customer. Maybe the marketing people do, but they don't have the P&L, so they don’t have the final decision-making rights. You’re constantly in a situation where you have to hold the line on what’s right for the customer to ensure you get the right outcomes."

Nelissen has taken it back to square one – literally. 

"When I started out working in branches, we saw customers every two weeks, when they came in to cash their paycheques. The branch manager knew your family, your needs, and gave advice that wasn’t just about pushing product. That’s the model we’re trying to recreate. We call it taking banking back to the ‘70s.

"That’s where taking banking back to the ’70s becomes so important, because you’re trying to set the scene for people about how you want to talk to a customer."

"Like in the ’70s, we started with what was important to the customer, not what was important to the bank. Creating that mindset and bringing people together around it gives you a great starting point for a conversation that asks, what do we really want to do here, and what’s the right answer?

“I had to really push and use my political capital to make sure intelligence didn’t get built into the digital platform,” he says. “Because the moment you allow different channels to start embedding their own intelligence, you create walls you can’t take down later.”

Speaking of lessons he learned from an earlier stint at Royal Bank of Scotland,Nelissen uses a construction metaphor to explain his architectural discipline: “You never want to build a wall where you need a door later on.”

"You’ve got to figure out what’s going to be important in the future, what you need to fight for, and what you can afford to let go and sort out later."

Reshaping engagement

NAB’s answer to the political and operational challenges of customer engagement is a centralised enterprise-wide decisioning engine, primarily powered by Pega.

At its core, the platform listens across every customer interaction, from mobile app visits to branch conversations to call centre inquiries, constantly updating its understanding of customer intent. It then determines, in near real-time, what action, offer or service message should be presented next.

“This isn’t just about marketing,” Nelissen says. “We refuse to call them campaigns. We’re shifting the organisation away from that mindset entirely.”

NAB’s decisioning platform constantly evaluates smaller cohorts of customers in real time. Where legacy outbound campaigns may have mailed 100,000 credit card offers monthly, NAB’s system can now target 20,000 highly relevant customers at 15 per cent response rates — far superior to the sub-1 per cent rates common in traditional marketing.

As Nelissen puts it the economics change so significantly that you can go to much smaller cohorts of the population and get the same result as if you went to much larger groups

Beyond sales conversion, Nelissen insists the true value is in reclaiming customer attention for service and engagement, the elements that ultimately build trust and lifetime value.

The result: the ‘one brain’ model not only optimises sales but also orchestrates service journeys such as Know Your Customer (KYC) renewals, travel notifications, payment dispute updates, and account maintenance — delivered in real time, through automated decisioning and orchestration.

"We turned on Pega at [NAB] and drove up engagement by 40 per cent just by doing a few simple things. While I’m seeing that, I’m just going to keep doing the simple stuff and doing it really well. At some point, I’ll reach a stage where a 2 or 3 per cent uplift will make a big difference, and even 1 per cent will be critical. I’ll get to that but at the moment, I can get 10, 20, 30 per cent just by doing the simple stuff."

That 70's bank

According to Nelissen, “When I started out working in branches, we saw customers every two weeks, when they came in to cash their paycheques. The branch manager knew your family, your needs, and gave advice that wasn’t just about pushing product. That’s the model we’re trying to recreate. We call it taking banking back to the ‘70s.”

By anchoring his data strategy to this metaphor, Nelissen has created a cultural narrative that cuts through the technical complexity.

“Data and marketing and all that is a really complicated area, and for people who don’t engage easily with it, giving them a handle on it is a great way to start."

In practice, this means hyper-personalised, contextually relevant interactions that feel intimate, even at scale. A simple birthday greeting delivered via the teller screen may sound trivial, but Nelissen says moments like this galvanised NAB’s internal stakeholders.

He recounts one pivotal moment at NatWest when one of the senior execs witnessed first hand the power of personal connection. "They were in this branch in Scotland, and an 89-year-old guy walked in. The teller said 'Happy Birthday,' and he started crying because he had no family, no friends. He happened to walk into the branch on his birthday, and he started crying. The staff started tearing up. It was just one of those moments."

"Whether it’s by luck or by design, you have these moments. You have to get the organisation to remember why people connect with you as a bank."

Out with propensity modelling, in with real-time triggers

For NAB now, beneath the goal of customer intimacy lies a technical architecture that is quietly overturning years of marketing orthodoxy, and particularly the reliance on propensity modelling.

“I’m not a huge fan of propensity models,” Nelissen admits. “By and large, they’re very old school. I’m a much bigger fan of triggers.”

Rather than attempting to predict customer intent based on lookalike cohorts or statistical likelihoods, NAB’s model instead observes live customer behaviour and reacts accordingly.

“In Australia, one of the signals that you’re looking for credit elsewhere is when you start downloading all your statements, because the credit provider wants to see them. We can see that, and then we can react to it."

This shift is particularly powerful in highly regulated markets like banking, where explainability and auditability are paramount. Trigger-based models offer more defensible decisioning than opaque statistical propensities.

And while AI plays a critical role in powering NAB’s platform, Nelissen draws a sharp distinction between the statistical machine learning embedded in Pega and the hype surrounding generative AI.

“In the end, we’re a regulated organisation. and at some point, we need to be able to explain why we did something in a given situation. Do I think we’re going to end up in a world where generative AI has a role in figuring out a more agentic, driven approach—not just in how it’s presented, but in what gets decided?

NAB has already deployed Gen AI for content creation in marketing and is exploring more agentic AI models but only where risk and governance can be properly managed.

"Maybe it’s going to take us a little while to get there. The regulator hasn’t asked us about it, but we know that if we screw it up, the regulator will absolutely ask, how did you decide to sell back to or push that to these customers? And there are plenty of examples of others who’ve gotten into trouble over that."

For now, Nelissen is focused on the executional discipline that keeps the engine running and evolving.

The Customer Brain

As Mi3 reported last year NAB embarked on an ambitious marketing transformation program to move more aggressively into the era of real time decision-making. It's a strategy that is central to the approach of three of Australia's Big Four banks – Commbank and ANZ has also invested heavily in the same Pega-based solution NAB is using. Westpac chose a different path, attempting to build its own real time decisioning platform, a move which ended in failure and cost the bank at least $70 million.  (It has since lent into the Adobe stack for new capabilities.)

Today, that project which NAB refers to as its Customer Brain has morphed into a  always-on customer engagement engine that is fundamentally reshaping how the bank interacts with its customers.

Speaking at the recent Pegaworld event in Las Vegas, NAB’s Jessica Cuthbertson, Executive, Customer Decisioning and Data Science, and Lisa Marchant, Head of Customer Decisioning, provided inside look at how NAB is dismantling the traditional campaign model, navigating the perils of hyper-personalisation, and scaling AI-powered engagement across millions of customers as the projects heads towards its three year anniversary

From campaigns to conversations

In its pre-‘Customer Brain’ state, NAB’s customer engagement model was archetypal legacy bank marketing: scheduled email campaigns, occasional banker calls, and periodic sales pushes. As Marchant described it: "Pre-brain, we did a lot of campaigns, usually very email heavy. You might get a banker call once every six months, and we communicated to you at very regular and set intervals."

That model is being upended. Instead of episodic campaigns, NAB is engineering a continuous, fluid system of customer interactions driven by real-time signals, machine learning models and adaptive customer journeys.

“We’re reacting to what customers are doing,” Marchant said. “There’s a lot more connectivity between sales, services and engagement messages.”

The project has also seen NAB rethink the customer lifecycle. Rather than focusing only on acquisition or sales events, the bank now balances service, education, and nudges throughout the relationship. According to Cuthbertson, about 60 per cent of Customer Brain’s actions are service-based, while 40 per cent are directly sales-oriented – a deliberate hedge against over-monetising the relationship at the expense of long-term loyalty.

Hyper-standardisation not hyper-personalisation

In an era where martech vendors relentlessly tout the virtues of hyper-personalisation, NAB is leaning into hyper-standardisation instead of hyper personalisation.

“Hyper-personalisation does not mean choose-your-own-adventure,” Cuthbertson said bluntly. “It does not mean you allow chaos into your system. So instead of hyper-personalised, I advocate for hyper-=standardised.”

That has seen NAB codified frameworks and playbooks for how actions are built and managed across its domains. Marchant elaborated: “We've written and developed certain rules of engagement. This has been really important as we extend across domains.”

The rationale is simple but often overlooked: unchecked flexibility quickly metastasises into operational entropy. Without discipline, you could have an explosion of contact policies, Marchant warned. That's a scenario that could grind decisioning systems to a halt.

By driving standardisation, NAB believes it achieves scale, speed, and governance, while still delivering differentiated content at the action level. This model enables NAB to optimise both who it targets and who it deliberately excludes—avoiding the trap of endlessly debating campaign inclusion lists.

400 actions; 2000 adaptive models

The Customer Brain is also at the heart of an expanding technical ecosystem. Initially launched on NAB’s core Red Star business, it has since expanded to support its white label brands through a multi-application architecture. Marchant outlined the scale: “By the end of the year, we’re expecting around 400 actions in market, and we've got a significant number of data signals that are available now available to us to build.”

Driving those actions are approximately 2,000 adaptive models that continually assess customer eligibility, intent and engagement propensities. The models feed real-time decisioning engines that determine not just who should be contacted, but whether an action should be shown at all.

Cuthbertson highlighted NAB’s use of “adaptive cut-offs”—thresholds that prevent actions from being shown if predicted engagement drops below acceptable levels. “It’s worth showing nothing rather than contributing to banner blindness,” she said.

Creative fatigue remains a constant threat in an always-on system. NAB combats this by regularly injecting creative variants into its actions. Generative AI is increasingly used to assist marketers in generating multiple variants, which can be rotated and tested in market.

This is where GenAI is helping Cuthbertson noted. “[We're] working with our marketers to get all the different creative variants in. Let’s test them and see who they resonate with. If they connect with a small subset of customers for a period of time throughout the year, that’s great.”

Milestones vs sales messages

One of thme ore effective insights from the Customer Brian is also one of its most human: people like to celebrate themselves.

The team discovered that milestone-based messaging, basically celebrating customer progress, consistently outperforms traditional sales messaging. Marchant explained: “What we’ve also been doing is celebrating milestones for customers on their home loan. For example: “Hey, you’ve paid down 25 per cent, 50 per cent, 75 per cent.” Both actions work really well; they’re among our top-performing actions. However, what’s really interesting is that the action focused on “Well done, you’ve paid down 25 per cent, 50 per cent, 75 per cent,” which is about something the customer has done, outperforms the one about the relationship with NAB by 50 per cent. It’s kind of common sense."

Cuthbertson meanwhile shared her personal experience receiving one of these milestone messages: “It showed me how much I had saved over the year. And I’m a sucker for confetti - it played an animation. How do you measure that? It’s a moment of fun and recognition.”

This behavioural insight extends across NAB’s portfolio: celebrate customer achievement rather than simply pushing product offers. The approach generates both engagement and NPS uplifts – building long-term stickiness with the brand.

Bankers, rewired

The shift in approach extends well beyond digital channels. Insights generated by Customer Brain are now directly feeding NAB’s banker networks, enriching customer conversations with relevant, contextual prompts.

Marchant described how leads are dynamically generated for bankers based on observed customer behaviour: “Our banker ecosystem had been very home loan retention and sales focused. What we’ve started to do is take successful actions from the digital space and incorporate them into banker conversations."

She said, "Our banker conversations are becoming more rounded and customer-relationship focused. We’re sending different types of leads to our bankers, around product usage and different products. One thing we’re really proud of is that we’ve been able to scale the leads we deliver to our outbound bankers through the customer brain. And at the same time, we’ve also been able to increase engagement."

One example involved offset accounts linked to mortgages. As Cuthbertson explained, bankers can proactively identify savings sitting in non-offset accounts and have simple, value-adding conversations with customers: “How can I transfer that for you? It’s going to help save you money.”

Crucially, these conversations often catalyse broader product consolidation, with customers moving accounts from rival institutions to NAB. "What we found is a lot of customers say, 'Oh, wait a minute, I have transaction accounts in three other banks. I’m going to move those over to you so I can get the full value from my relationship.' It’s a really nice conversation for the bankers to have, and that’s what we try to do in that channel – make it simple and easy for them, and really bring that through."

What went wrong and how NAB fixed It

Even with its progress, NAB’s transformation hasn’t been without hiccups. As the system scaled, unexpected challenges emerged – particularly around overexposure and engagement decay.

Marchant recounted a period where engagement metrics dipped, prompting a forensic review: “We found an action that customers had been overly exposed to within our mobile ecosystem. Customers had just become bored of it.”

The bank responded by tightening contact policies and adjusting delivery cadences. The result? Engagement levels rebounded.

Beyond frequency management, NAB also learned the importance of continually refreshing content—whether by introducing entirely new actions or injecting creative variants into existing ones.

Another key insight: old-school A/B testing paradigms don’t easily map to adaptive systems. Marchant illustrated this with a deposit campaign that initially favoured generic messaging over holiday savings prompts. But seasonality shifted consumer sentiment.

From January to March, the team saw the holiday savings message "Absolutely go through the roof” she said. In a legacy A/B model, that variant would have been prematurely turned off. Instead, NAB has embraced creative diversity, allowing multiple treatments to coexist and capitalise on contextual triggers as they naturally arise in the market.

Modern banking martech

For NAB, Customer Brain is no longer a project with a point to prove. Instead, business adoption has surged from 10 to 70 per cent and the team is hopeful of further growth in the year ahead. It is now a core operational capability as it has evolved into a system of engagement that balances automation with human relevance, and data science with behavioural insight.

There is also a noticeable change in the way the executives talk about the impact of the Customer Brain. Last year, the story was all about the benefits to the bank. This year the customer benefit has taken centre stage, perhaps reflecting how the NAB team has enhanced its understanding of the potential of the system, or perhaps reflecting the customer obsession of CEO Andrew Irvine who joined took the role in April last year. As Cuthberton noted, "Our strategy is really focused on becoming the most customer-centric company... not just the most customer-centric bank."

The progress is clear. Where once the bank was beholden to rigid campaign cycles, it now operates in a state of dynamic equilibrium balancing service, sales and engagement based on evolving customer needs.

Marchant summed up the shift succinctly: “We’re seeing a much more fluid communication approach. That's a really a big shift for us."