Billions of Personal Interactions

Many hark back the ‘good ol’ days’ of branch-banking when the Bank Manager knew each of his customers and provided highly personalised service and advice. Whether or not these halcyon days ever existed, it is difficult to see how these analogue relationships could scale to the digital world. Recent research found that at some banks over 90% of all customer contact now takes place over the phone, internet or smartphone. With more and more customers choosing digital-first interactions, creating that personal touch in digital relationships at scale is the new challenge.  Expectations also change in the digital world. Fast response and instant decisions are the norm, whichever channel is preferred. To meet evolving customer demands, whilst not only remaining profitable, but improving cost-income ratios, banks must find ways to manage billions of hyper-personalised interactions at low cost. This demands the right technology, but also the right operating model to leverage data analytics to save money and grow revenue.

Marketers have long espoused the concept of marketing to a segment of one. Treating every customer as an individual, and matching products and services to their exact requirements in the moment is now possible in ways that never were in face-to-face branch interactions. With 3.6 billion digital banking customers by 2024 and the ability to collect data at every touchpoint capturing data is no longer the problem. But, whilst many banks now have effective digital interfaces with their customers, many back-office processes still require heavy manual interaction. These kill any chance of intelligent, real-time conversations with customers over digital channels.  More fundamentally they drive cost.

Operationalising automation

To deliver personalised service and reduce costs banks need to operationalise automation from end-to-end removing human interaction from the vast majority of tasks. Creating models that leverage data to make fast, accurate and auditable decisions 99 per cent of the time freeing humans to deal with more complex tasks and exceptions.  

Modelling and automated decisioning systems are becoming the pillars of low-cost, highly personalised banking. It is where fintech innovators and ‘big tech’ have created an advantage over traditional banks – not by having more or better data but orchestrating the data they do have to support predictive analytics at scale. Every click on the PayPal website, every interaction with the app triggers thousands of models that determine in milliseconds the best next action. That could be a risk mitigation action to prevent fraud, or a new marketing interaction to cross or upsell a service. Either way, an automated decision has saved or made some money.  No humans are needed to make these decisions which is why Neo banks and other data-driven financial service providers have cost income rations of around 40% compared to the 60-70% of traditional banks. 

Not only have these new entrants started with ‘green-field’ technology and avoided the data silos that plague traditional banks, but they have created operating models that put predictive analytics at the heart of operations. For them it is not replacing the branch and the bank manager with an algorithm but building an entirely new business model based on thousands of algorithms. They link multiple digital channels into consistent customer journeys supported by thousands of automated decisions driven by predictive analytic models. And, to a large degree, these run themselves.

Order of magnitude shift

By contrast, traditional banks struggle to create and deploy handfuls of predictive models across their business. The demands of finding, joining, and preparing data to provide useful insights is hard and time consuming. Even when built the challenges of operationalising models means that almost 80% fail to deliver the intended value.

This has clearly got to change. To meet the evolving demands of customers and drive down costs, banks need to be deploying thousands of models to automate interactions with tens of millions of customers. Leading banks are beginning to orchestrate their data and create predictive model factories to do just this. It’s a conceptual leap as much as a technological one – the Teradata platforms that many already operate provide the perfect foundations. Using Teradata as a single data platform provides the framework for data science teams to industrialise their approach. Teams can collaborate to create models that cross product and organisational boundaries to focus on customer journeys. Models, features and data sets can be shared and reused to deliver maximum value – the efficiency and ROI on data science can be dramatically improved as they shift from creating 10’s of models to 1,000’s each year.

This is the scale of the challenge. The industrialisation of data analytics within banks to support hyper-personalised digital decision-making at speed for tens of millions of customers. Creating change of this magnitude can only be done from the top. Creating a few models to help a line of business, or specific products is not enough. Banks must prepare to build data factories that can build, test and deploy thousands of analytics models that deliver value along the whole customer lifecycle. CEOs and other senior leaders need to set the direction of travel and be ready to transform. Speed and agility will be key to win in the age of hyper-personalisation as we’ll see in the next blog.