How analytics matter
Building people’s financial wellbeing, using data
Creating original hypothesis for how to use data and analytics, by making the right bets on people’s behaviour
by martin gronemann and sebastian barfort
What's the problem?
For a moment, let’s forget the hype surrounding the potential of the digital transformation and take a look at the facts.
In a 2019 survey of c-level technology and business executives, more than 92 percent report that the pace of their big data and AI investments is accelerating. But an eye-opening 77 percent, up from 65 percent last year, report that business adoption of big data and AI is a major challenge.
The solution to this challenge is not more money. In fact, many companies already invest heavily in data and AI, and have been doing so for years. So what prevents them from creating real impact with data and analytics? To answer this question we must start somewhere else - with the customer.
What's the insight?
When we meet executives at global and regional financial institutions, we often find that the key reason the hype around data and analytics often falls short of expectations is that they are failing to understand what truly matters to their customers. It is hard, often impossible, to build the right solutions on top of the wrong hypothesis.
The consequence is often that financial institutions end up using digital services in a way that reinforces the "cold" transactional relationship they are trying to break out of. For example by pushing irrelevant, automated communication to their customers like the 42 year old private banking customer who received an invitation to a special theatre play for kids 5-8 years old. The only problem was that his kids were 13 and 15.
One top of actively aliening customers we also see how financial institutions miss out on a lot of sales opportunities because data and analytics are not used in a helpful way. Banks that have invested a lot of resources in coming up with new digital solutions but not making sure that their customers systemically get introduced to these. An insurance provider who created a GDPR insurance offering but then didn’t reach out to the relevant SMB customers until 5 months after GDPR became a reality. The list goes on.
These failed attempts at digital services are not the product of a shortage of talent. Indeed, many financial institutions have invested heavily in data scientists and machine learning engineers. Rather, it is the consequence of generating hypothesis solely from thin data - millions of data points about what people do, and when and where they do it.
To generate truly valuable customer insights, we combine thin and thick data - accounts of why people do what they do and what is truly meaningful to them. Thick data's strength comes from its ability to understand why your customers behave as they do. We regularly use thick data in our work with some of the most sophisticated technology companies in the world. This might come as a surprise to some, but it shouldn't. They know better than anyone that to unlock bigger value of their abundance of data, they need to understand what their customers are looking for in the first place.
When your digital solutions are built on the right hypothesis you will be able to increase product uptake by understanding what your customers need - and when they need it.
What's the potential?
We see that the financial institutions who succeed in their digital strategy are building on the right hypothesis about their customers. Getting your customers right is no easy task. But it is paramount. Only then will your data scientists be able to build the digital products that deepen your relationship to customers. Only then can they ensure that your customers remain with you longer. Only then can they start building the high-value digital experiences that increase their use of your products and services.
When your digital solutions are built on the right hypothesis you will be able to increase product uptake by understanding what your customers need - and when they need it. You will increase customer retention by providing "warm" rather than "cold" services. And you will open a new world, one in which data is no longer siloed, but used across the organization to generate new offerings.
In the following pages, we describe how you can use data and analytics to strengthen your relationship with the people that matter most: your customers.
How can you use data and analytics to build a stronger customer relationship?
Below, we describe some examples of what building on the right hypothesis allows you to do. In some cases, the financial institution stored the necessary data already, but needed new creative ways to combine and analyse them. In others, we worked directly with the client to collect the necessary data that enabled them to build the right new offerings.
But in all cases, to get it right, we started with the customer.
Drive engagement by systematically capturing customers' Passion Points
We were recently asked by a regional bank to develop a new perspective on their private banking offerings. By understanding how affluent people take care of their wealth , we discovered how they directed a disproportionate amount of attention to a particular part of their wealth - what we call their "passion money". Common passion points included start-ups, philanthropy, and green tech.
The bank has access to an immense amount of data about their customers -- likely more than businesses in any other industry. But they had no information about passion points, and for this reason, many of their digital products felt off-target and off-putting. One customer, a mid 40ies inner-city entrepreneur who had recently sold his company, received invitations to webinars about healthy ageing. Another, a young inheritor looking to build a network in the start-up world, was invited to go sailing.
Building on our foundational insights, we have recently started working with the bank to systematically capture information about passion points. Our process leverages the best of human and automated approaches. Initially, the data collection is primarily manual. But as the size of this training dataset grows, the bank's data scientist can start looking for patterns, ultimately scaling the process to every single customer.
Knowing their customers' passion points has opened new opportunities for the bank to drive engagement and unlock value pools by systematically targeting advisors, products and events to customers' passion points.
Build a warmer customer relationship by proactively reaching out in the right moment
A few years ago we were approached by a big insurance company. Despite having a leading position in the market, they wanted to explore how they could differentiate their brand and create a coherent digital experience for their customers.
Traditionally, the world of insurance is a reactive and transactional one. Most customers we met only interacted with their insurance company after something had gone wrong. By understanding people’s need for safeguarding we saw that the insurance company had a big opportunity to build a deeper relationship with their customers by moving to proactively reaching out in the many moment - big and small - when peoples worries and concerns rise.
The benefits were clear -- but so were the risks. When an insurance company proactively reaches out, they better get it right. There is one way to be helpful around big decisions that change your uncertainty for good, and another for self-initiated projects where customers look for ongoing dialogue and inspiration. Building on our insights, we identified a clear logic for when and how to engage right in different situations.
Knowing the when and how for engaging the customer, we started working with the company's data scientists to identify these moments in their customer data. We were pleased to discover that most of the information necessary was already available to the company. But without the right hypothesis, their data scientist had been unable to connect the dots.
The potential of data and analytics to transform entire industries is real. But to unlock its value, it’s necessary to move beyond the hype and develop a mature understanding of how data science and analytics fit into organizations.
How do you get started integrating data and analytics into your organization to make lasting impact? In our experience, the right way to get started often involves the following three steps.
STEP 1: Develop the right hypothesis
More digital services are being built today than ever before --and the rate of growth is exponential. To build the "warm" digital services that allow you to stand out in this increasingly crowded market, you need the right hypothesis.
A reliable way to get your customers right is to take an outside-in approach -- what we call Sensemaking. The key is rephrasing your business question into a human one. A human question is broad. It is not "How can we grow private banking within a year?". Rather, it is "How do people take care of wealth?"
Starting with a human question allows us to study your customers' lives beyond their interactions with a bank or other financial institution. It is through this process we generate truly original perspectives. This is what enabled us to discover passion points, and when and how an insurance provider can engage their customers.
Starting from our foundational insights, we help our clients build a data-centered strategy, with clear hypothesis about how new digital products and services will change behavior and generate value for your company.
STEP 2: Build a prioritized roadmap with clearly defined outcomes
In our experience, building a successful roadmap for how to get started involves three steps.
First, you need to prioritize among the opportunity areas. You do not want to spread your resources too thin, or have your data scientists pursue too many opportunities at the same time. We recommended focusing on few well-defined projects to ensure a 'red thread'.
Prioritizing opportunity areas should be no different than any other business decision, and should depend on potential benefits and costs. Some benefits are primarily commercial. Others primarily help you build a warm relationship with your customers. The most promising projects usually score high on both.
One key determinant of costs is your company's internal data infrastructure. You rarely need to invest heavily in expensive new software to start building the right digital products and algorithms. But before we can get started, we need to know what data your company is capturing, in which frequency, format, and so on.
Mapping your company's data infrastructure helps you identify the opportunity areas with the lowest costs - often those where data is readily available. For opportunity areas where new data need to be collected, it will help you determine the associated costs.
Second, for each high-priority opportunity area, make sure you can confidently assign one or more KPIs. These KPIs must be measurable at scale in real consumer behavior. Your data scientist need access to large-scale data to build the most effective machine learning models.
We're often asked how to create KPIs to measure customer satisfaction. We rarely recommend surveys -- new behaviors can't be tested by asking people their opinion, and surveys are slow, expensive, and often impossible to scale.
We suggest to create a minimum viable product and study real customers in their real environment. When we have established, through thick data, how and why the product provides value, we then search for behavioral thin data that enable us to create scalable measures of customer satisfaction. These measures fit naturally into your company's data science pipeline.
Third, you need to carefully consider how to pilot a new product. This requires analyzing how valuable different groups of customers are for your company, as well as considering how sensitive they might be to experimentation.
STEP 3: Train your organization, not just your algorithms
With a clear objective and the right data available, it's time to begin implementing analytics in your organization.
In our experience, analytics projects often fail because they, and the data scientists that develop them, become siloed and disintegrated from the rest of the organization. When we talk to executives, they often express a degree of disappointment -- the products and models built by their data scientists often fail to support the organization’s broader strategy. And rarely do they feel that their data scientists are sufficiently embedded in the organization.
This is why starting with foundational insights is key to getting analytics right. Foundational insights allow your teams across the organization to unite around a common goal and a shared understanding of how to create value for your customers. And by cooperating about how to transform insights into useful algorithms, you empower your organization to think of new creative ways to incorporate their knowledge into analytics and data science.
This is how you move beyond the hype. By developing original hypothesis. By capturing the data that enable your products to stand out. And by creating a mature organization that gradually integrates analytics and data science -- one well-defined project at a time.
We recently piloted a project for a large energy company who wanted us to help build a data-driven customer engagement model. We worked jointly with the company's marketing and data science teams, starting with foundational insights and gradually moving to the development of new machine learning models.
While the initial results of our work are promising, the biggest impact has been the steady replacement of siloed thinking with an organization that is curious, creative, and collaborative. Gradually, the data scientists felt comfortable moving out of their comfort zone and start integrating with, and learning from, other teams. And by integrating data science throughout the process, the marketing and strategy teams were learning how to provide the right inputs to the data scientist's models: