Predictive analytics and machine learning are being used to personalize multi-channel customer service at Rabobank, the Netherlands second-largest bank.
The bank has taken a tech-driven approach to identify customer needs and deliver actions when, where and how they are needed, with analog and digital converging. According to Coon Wiers, Rabobank’s Senior Product Manager, Customer Relevance:
We started with an idea that we want to have the same experience on our digital channels that we provide with our local branches. That really bumped up the perspective on what we want to be - not just providing information, but really make it personal. And let's start it off with a wildly clear segmentation system, where we provide groups of customers with certain information.
That foundational idea soon expanded, he adds:
Around two and a half years ago, we saw that that wasn't sufficient anymore. We saw that we were lacking the flexibility to really make it personal. That's when we moved over to Pega. We did that with the idea that the self learning system and machine learning could help us get through the next phase. We've rolled out Pega over the last year. We have it in the app and on the web and in the call center. We are now really looking into making the journey where we personalize a continuous dialogue, showing the next best actions during or at the end of a sales process to engage into a next process or a cross-sell opportunity.
At the heart of the journey is analytics, he says, something that has had a transformational impact:
Like many companies, we were always positioned around product pilots and profit and loss for those products.So you’d get campaigns during certain time periods that are really focused on getting a lot of customers in during a time. But with Pega we were able to shift that towards customer lifetime value, where our measurements around revenue-per-customer and cost-to-serve per customer were being used as the main driver. In order to do this, you need to be able to really show where a customer is at a personal level in the journey.
So you can't really use big segmentation; you need to be one on one. Pega provides that with their self-learning models. What we do is we measure or let the model learn, based upon the engagement the customer has with the offer action. That way we can connect back to the action and outcome.
This has delivered quantifiable results. For example, in Q4 2019, Rabobank was able to lower its service costs by 1.5% per customer, just by serving better surface messaging.
Rabobank’s efforts have been helped by the way the organization is structured. Wiers explains:
We're organized with the agile way of working. We have ‘tribes’, and you've got ‘squads’ working for those ‘tribes’. As a product manager, I work in a ‘tribe’ and, as a ‘tribe’ we are responsible for setting up the Pega stack, making it available for the company within the channels. Next to us we've got a group that's working for a ‘chapter’, a group of people that actually share their knowledge towards all the other groups, and they are functional users.
They are all users within Pega and they get all the different questions on a day-to-day basis, and really make it work. So you've got our’ tribe’ on the one hand for the technical enablement and you’ve got the people that are working functionally within Pega. So that way we can bring Pega towards the company. That's done through marketing at the moment. Marketing is also a ‘chapter’, and they have people in all the different ‘tribes’, such as mortgages or investments. They will get new insights from the ‘tribes’, get new ideas, build campaigns and offers, provide them to the functional users and they will put them into use in Pega. But if for instance, there is a channel not connected or there is a system not connected for the data, then we will come into play and assure that that's that's built.
This ‘tribal’ approach had some initial challenges, he recalls:
At first that was difficult because when you were changing the organization, you [could] see that there was a lot to do and a lot of people have different priorities. What we really found out is that if you connect the day-to-day basics directly to a team, that works out quite OK. Then you get a value stream of new requests. All those requests we put into a list, and we look at them on a quarterly basis. Then we really see [what] we come to do for the next three to six months and plan that and prioritize that.
Senior management buy-in has been critical, says Wiers:
It is a strategic project for our management board. Success is a huge part of the deal….On a day to day basis and weekly we provide metrics analytics to our management board so that they are always aware of what's out there performing and what new stuff is coming up
Equally important is ensuring company-wide engagement, a need which has resulted in some imaginative engagement approaches:
What we do is we try to get as many people in as we can and what we do is we can bring them the ideas and what they can gain from Pega, but we also bring them along with celebrations. So we do a roadshow, and when something like a new technical solution goes live, we go to a certain group or department or team that's working with that solution. celebrate there with some drinks and some food, and then their senior manager will give an insight on what it will bring for them. That's really our rollout celebration time.
We've got one really special celebration and we call it our Decommissioning Pink Rabbit We're moving away from IBM and we're going towards a Pega stack. We have the thing that there is a Pink Rabbit suit in the company. If we decommission something from IBM (or another vendor), then the senior manager needs to wear the Pink Rabbit suit at the decommissioning event. That's always a good laugh and brings some humility to everyone.
Having a laugh is something that would forgivably be in short supply at the moment in light of the COVID-19 pandemic and its impact. This has been a complicating factor, admits Wiers:
Let's be honest - these have been unprecedented months. We've never seen this before and we have to adapt to a changing reality. You see that happening with our customers as well. When we first were hit was in the beginning of March. Everyone was in lockdown. People turn towards the bank for security, just to be sure that they can still use their bank products, get money and be able to at least spend money on groceries. That was our first goal, to really position all the communications towards that, that we built a trust.
It also gave us the time to really go to our portfolio and remove stuff that wasn't matching the new reality. For instance, insurances around travel - that's not really a good thing to show when you're not allowed to fly. Now we've been in the lockdown for two-and-a-half months, we're really getting towards the personalization again, focusing on what people are really looking for and how we can support them. That is a giant task because we still don't know what's ahead of us, but we try to do it one step at one day at a time.