CommBank's analytics chief on how its AI-powered 'Customer Engagement Engine' is changing everything
Commonwealth Bank Chief Analytics Officer Andrew McMullan says the bank's AI-driven customer engagement engine has transformed the way the bank is set up, how it interacts with customers and hopes it will help people keep their homes as Australia battles Covid's impact. He's hardly effusive about the usual big CRM and customer experience technology providers but credits "visionary" chief marketer Monique Macleod as pushing the analytics team further, faster and deeper. Meanwhile, its new Bill Sense should have service providers across the economy - energy companies, insurers, gyms, telcos and maybe even SVOD providers - on their toes.
You need to know this:
- Andrew McMullan is a former lecturer and research associate in statistics
- Now driving Commbank’s AI-powered analytics ‘customer engagement engine’ to deliver personalisation at scale
- Its ‘Benefits Finder’ has unlocked more than $150m of benefits targeted to those that qualify for them
- Has also provided “ten times more leads to home loans” team that were “300% better leads than they have ever had before”
- Now launching ‘Bill Sense’ to help customers manage payments – and tell them if providers are charging over the odds.
“Monique Macleod [Commbank's CMO] has been driving this from the very get go, and now her team is pushing us even further in terms of personalisation. They want to go further, faster.”
Twenty years ago, Commbank’s chief analytics officer Andrew McMullan was a lecturer at Glasgow University and completing a PhD in machine learning and statistics.
Back then, he would “press run on code and then leave for a couple of days”, finding out only then that his models “hadn’t worked out quite the way I had hoped”.
Today, the bank’s customer engagement engine is “running hundreds of billions of data points in seconds”. Its data strategy is literally paying dividends, with the engine helping to highlight benefits that customers can claim ($150m-plus claimed so far) and a new Bill Sense set to make tailored recommendations when Commbank’s AI thinks customers are paying too much for services. Telcos, gyms, energy suppliers and the like have been warned.
Post-Covid, the bank’s “number one priority is keeping people in their homes”, says McMullan. Putting money back into customer pockets using machine learning and big data might just make the difference between success and failure.
“We have six million regular app users, so it's easy to get scale and learning very quickly. We may turn a message on and select a group of customers; as they engage with that, we would learn in minutes, hours from that model what works, what doesn't.”
Buy-in, not sell-in
Technology vendors and consultants do a great job in selling their wares and ability to deliver step change in customer experience.
But asked to rate the go-to big CRM and CX engines that plug into Commbank’s systems, McMullan instead talks about the quality, compute power and intelligence provided by Pega and H2o.ai that underpin its bespoke system.
Either way, people enable and deliver change, says McMullan; without buy-in at the highest level, it won’t work.
As a result of that support, the programme’s success is such that “it has fundamentally changed the way we think as an organisation and also how we have set ourselves up,” says McMullan.
“From the very first day I arrived in Australia and went to work with Commbank, it was super-clear to me that the senior leadership of the bank were truly committed to building relationships with customers, improving the experiences that customers had when they interacted with us.”
He cites CMO Monique Macleod as a major influence in driving the alignment of marketing and technology functions so that customers get better outcomes.
McMullan describes Macleod as “one of the visionary leaders who wanted the customer relationship banking programme to work and who pioneered the customer engagement engine”.
“Monique has been driving this from the very get go, and now her team is pushing us even further in terms of personalisation [of comms]. They want to go further, faster,” he says.
Which is exactly what McMullan intends to enable. He says the bank now has the tools and structure to quickly scale new variants, products and services.
The Commbank app is a key plank.
“We have six million regular app users, so it is easy to get scale and learning very quickly,” says McMullan. “We may turn a message on and select a group of customers; as they engage with that, we would learn in minutes, hours from that model what works, what doesn't. Then, when we're happy with its performance, we can switch it easily into the production environment and allow that conversation to be prioritised for customers who look like those [first batch] customers who have really enjoyed engaging with it.”
Meanwhile, McMullan says the engine has benefitted from “more than a thousand pieces of feedback” from frontline staff in branches and call centres. That affords a double benefit: improving the system for customers, while creating “incredible trust in the engagement engine from the people who serve our customers”.
In the beginning...
In a precursor to its current test and scale app approach, the customer engagement engine was born of humble origin some four and half years ago via a single laptop in a Penrith branch.
“Their team wanted to really understand the conversations that our frontline team members wanted to have with customers, the conversations that they felt would really make a difference,” McMullan explains.
“The vision was: How do we use our systems, our data and understanding of customers, take what the frontline staff know are brilliant conversations and build those out in a way that all of our frontline can have a great conversation with every single customer when they come in to bank with us?,” he continues.
“So we spent a little bit of time in that branch, in that region. Then, when we were happy and when our front line colleagues told us that the conversation set that we had were genuinely great conversations for customers that helped them get more from their banking with the Commonwealth Bank of Australia, we rolled that out across the nation.”
Since then, all customer channels have been wired into the engagement engine, which now makes some 30 million decisions a day to present “the best next conversation” between the bank and the individual customer, “returning answers in milliseconds”.
How it works
“Like most organisations, we have a big data layer, so we've organised all of our data that allows us to understand everything about our customers,” McMullan explains.
“Sitting on top of that, we use Pega software to do our customer decisioning. So I have a product called ‘Customer Decisioning Hub’, which we use to run all of the different models and scoring to understand what's the best conversation to have next.”
That process runs across the app, the website, call centres and in branch, selecting the most appropriate digital message, or recommending the most appropriate conversation to staff.
“We realised across Australia there are hundreds of benefits or rebates that Australians are eligible for, but many people just didn't know were there. A few hundred thousand customers have used Benefit Finder in the last few months alone. Maybe they are on reduced hours. Maybe they have lost their job. All of that money really helps right now.”
As people brace for the impact of Covid-19 on their lives and livelihoods, banks have seen massive increases in engagement – and McMullan says Commbank’s investment in systems and data have come into their own.
“We've had 250 million personalised customer interactions in the mobile app alone related to Covid,” he says, using home loan customers as an example of how well that has worked.
“They would get a message in the app that if they clicked on, would take them through to the landing page, which would show them what options were available on their home loan, such as a six month deferral, or whatever other option was available to them.”
The bank took a similar approach across its services, using the engine to guide customers to their available options – and has been updating comms as the Covid situation has evolved.
One of the biggest ways Commbank showed up for customers during Covid was via a programme it commenced six months prior - aptly called ‘Benefit Finder’.
“We realised across Australia there are hundreds of benefits or rebates that Australians are eligible for, but many people just didn't know were there,” says McMullan.
So it set about helping to match customers with rebates based on their data. Today there are close to 300 rebates and benefits on the platform. Those with kids will see prominent messages for ‘Active Kids’ vouchers, allowing them to claim $100 if their children play sports, says McMullan. “Whatever the right benefits are [for the individual], they become available and they can click through and claim those benefits.”
“As Covid kicked off, we very rapidly made sure that any benefits, rebates, money that was available to people that were going to face hardship were put into that tool,” he continues. ‘We added another nineteen benefits very quickly.”
Over the last 12 months, more than half a million customers have used CBA’s Benefit Finder to start the process of claiming credit they are entitled to, says McMullan, with “a few hundred thousand in the last few months alone”. Customers, he says, are increasingly using the feature in a time of need.
“Maybe they are on reduced hours. Maybe they have lost their job. All of that money really helps right now.”
The bank’s data capability also came into its own during last year’s bushfires.
“When the bushfires started to take hold, straightaway we got the team together and had an emergency assistance package that we could let customers know ‘if you are in this area and you're impacted by this, here's all of the things that you can get help with’,” says McMullan.
That system was updated daily by affected postcode, he adds. But then the bank started to take a more proactive approach based on where fires might be heading, “hooking ourselves up to updates from the fire service, which allowed us to understand where the fires were going”, says McMullan.
That led it to identify 124 postcodes that were most at risk at the fires’ peak.
“As soon as we had that data from the fire service, our systems would include those customers and we would start to contact them either via messages in the app or if they called,” says McMullan. “Also, if they went into branch, or they logged on, we knew they were in an area that was impacted. So we let them know ‘here's all of the things that we can do for you right now’.”
“Within the first 12 months of the program, we were sending something like 10 times more leads to our home loan lenders - and those leads were 300% better than any of the leads they'd ever had before.”
Net promoter scores and satisfaction when Commbank is conversing with customers using its engagement engine are “absolutely higher than when we don’t,” says McMullan.
Were proof ever needed that better customer experience walks hand-in-hand with balance sheets, Commbank’s home loan sales provide one example.
“When we started investing in the customer engagement engine, one of the things we really wanted to do was help customers who were looking to buy a house,” says McMullan. “Within the first 12 months of the program, we were sending something like 10 times more leads to our lenders to help customers.”
Meanwhile, those leads were “300% better than any of the leads they'd ever had before in terms of identifying customers that were asking for our help, getting to them at the right time and helping them with their aspirations of buying a home”.
“With Bill Sense, we really want to help customers understand what bills they have, what they're paying, and the opportunities for them to maybe get a better deal somewhere if they're paying more than we think they should be.”
While new lending is a fundamental banking tenet, Commbank’s top priority for the foreseeable post-Covid future is “to keep every one of our customers in their home”, says McMullan.
On that front, its data brain has a new target: service providers that are charging its customers over the odds.
“Bill Sense is a new feature that we’re really excited about,” says McMullan. “We’re helping customers understand all of their bills and to keep on top of when they are coming in. That can be quite hard for customers. The new feature makes it easy.”
As well as providing alerts if bills change, Commbank will harness the engine to drill down into what customers are paying for services - and benchmark them.
“We really want to help customers understand what bills they have, what they're paying, and the opportunities for them to maybe get a better deal somewhere if they're paying more than we think they should be,” says McMullan.
If one of Australia’s biggest banks is rolling out a bespoke price comparison service based on high grade machine learning and hundreds of billions of data points, providers across the Australian services economy should probably sit up and take note.
The business and marketing industries are bracing for the impending storm brought about by Apple’s changing IDFAs (Identifier for Advertisers) and Google’s departing cookies, but location advertising is far from over Emma-Jayne Owens writes. Consent popups will be the gatekeepers of permission and brands and publishers will need to prove their value to land an ‘accept ’. It’s a good thing, but there is a big education job to do around this to help consumers understand their options. Here’s the why and how.
It may not feature in a marketer’s conventional media planning ‘matrix’ but the evidence is compelling around real ‘fans’ of TV formats showing major increases versus ‘viewers’ for ad attention and recall. Fans deliberately interact with content outside of viewing hours and – well, simply – as fans, they’re in the right mood. 10 ViacomCBS’s Michael Stanford lays it all out.