DEVELOPING POWERFUL SEGMENTATION IN FINANCIAL SERVICES

How to practically identify opportunities that will drive growth. 

Since its popularisation in the 1960s, segmentation has been an ever present tool in the marketer’s tool-bag but despite continual development and refinement, it remains a polarising concept. In a world of increasingly valuable customer databases, direct-to-consumer and online marketing, the ability to quickly and accurately locate a customer in a segment is paramount, but too many methods rely on a complicated ‘black-box’ software solution, and back-end algorithm to ‘type’ customers, that most practitioners cannot explain. This sophisticated software solution can close down the internal dialogue - encouraging us to abdicate responsibility for creating the segmentation itself. The approach we discuss here is intended to put a ‘fierce strategic dialogue’ back into the segmentation development process to deliver actionable segmentation that works.

We have outlined a project here that includes in its core design principles

1. The ability to take action;
2. Statistical processes that encourage & support ‘fierce dialogue’ (rather than replacing it);
3.  Co-creation with the client stakeholder team. 

Designed to address common segmentation pitfalls, this approach delivered a powerful, action orientated, framework that could be immediately embedded within the business. This is an evolution of work previously done to develop a segmentation to support management of a global brand – using a Latent Class clustering approach (Fonseca, Riley & Smith, 2017)[1]. The evolved segmentation utilises a version of the psychological model we used to underpin that framework, but applies it in support of tactical market by market activation.

The process starts by defining what the segmentation needs to identify or explain … through an extensive review of literature, existing research data, and qualitative interviews. A good starting point is our Perspective, Intent and Command (PIC) model. We have developed this model over a number of years – based on our own thinking and other published research in the financial services sector and beyond. This model is the basis for a dialogue with our client working team as to what behaviours, needs or attitudes we need to understand.

PIC focuses on fundamental human drives that underpin and motivate people’s everyday attitudes and behavior as they relate to their ‘financial life’. There are numerous taxonomies and frameworks for identifying core human drives, but we arrived at these 8 Cs as the basis for the framework

• Challenges and goals: The need for people to have challenges and goals that are coherent with how they see themselves and their future lives.
• Congruence: The need to have a positive self-image and be aligned with one’s values.
• Care and support: The need to care for and support others, to help family and or the community.
• Control and competence: The need for control and competence in one’s life.
• Connectiveness: The need to be connected to and be part of a wider group and/or community.
• Contribution: The need to feel valued and make a contribution to society.
• Creative expression: The need to express oneself and be creative.
• Change: The need to constantly change, adapt and respond to differing circumstances.

Different individuals will attach different weights to varying core human emotional drives, but these drives began to guide our approach to detecting deep insights about how people think about their financial freedom and literacy. Work over a number of years has allowed us to refine the model and collapse these fundamental human drives into our three PIC constructs that drive financial decision-making, attitudes and behavior.

• Perspective: how people view and connect with the world – whether they have a more optimistic/worry-free, as opposed to a more pessimistic/more anxious, view of the world, including their view on responding to constant change.
• Intent: where an individual is in terms of the challenges and goals they set. It embraces the congruence they are trying to achieve and the creative expression they want in their lives, including the need to support, care for and contribute to others.
• Command: confidence, control and competence, whether people see themselves as being driven by events or able to manage their lives and contribute to others.

We used data from a subsequent quantitative study to identify variables that (singularly or in combination) show some relationship or correlation with what we want the segmentation to predict i.e. the elements of PIC. To create compelling and distinct segments we need to identify the variables which create the greatest differences in the PIC dimensions. The analytical process covers 4 core tasks:

1) Build a data matrix - to test potential market frame variables using CHAID[2]; identifying characteristics that are predictive of PIC AND can be used as a basis for marketing efforts designed to reach each segment.

Exhibit 1.

CHAID is sometimes used as a segmentation method in its own right, however in this context it is used to detect interaction between variables, identifying independent (predictor) variables given a target dependent variable (Exhibit 1.). In this sense it is being used to inform the dialogue, identifying those variables that explain differences in the elements we wish to understand i.e. PIC. We are then using those variables that correlate with or explain PIC attitudes and enable marketing action (or proxies for strong variables that are less actionable), create a market frame – i.e. a hierarchical interaction of those variables that best predict different combinations of PIC.

2) Create possible market frames - Our understanding of core human drives underpinning and motivating people’s everyday attitudes and behaviours - informed and supported the development of the PIC model, which in turn formed a framework to guide our thinking and learning as the segmentation frame evolved and nascent segments began to emerge. PIC is predicated on our collective human needs: to have challenges and goals that are coherent with how we see ourselves and our future lives; to care for and support others – to help our family and/or community; to be part of a wider group and/or community (the we vs. me ‘connectiveness’ dimension); for control and competence; to be creative, express oneself and be able to make a contribution to society; and to constantly change, adapt and respond to differing circumstances (reflecting people’s outlook on the world) – and the frame must capture and reflect differences on these key dimensions i.e. the interaction of the variables that we choose to use on the frame must pull customers apart on these PIC dimensions.

One would typically create multiple frames to test which best meet the project objectives – in this instance the PIC criteria. These frames were shared and reviewed with the client team. These co-creation sessions rely not only on the quantitative data but also, crucially, the business context and resident market knowledge of the participants. Creating the frames is an iterative process of test and review. We review how each axis, and both in combination, drive differences on the PIC dimensions, looking to find the combinations which are most differentiating and provide most interesting variance (Exhibit 2).

At the heart of traditional financial services marketing is the notion of which stage people are at in their life. Our analysis (not surprisingly) identified age, marital status and presence of children within our short-list of variables. This allowed providers to offer products to individuals at key life-stage moments, for example when getting married, having their first child, sending their children to university, about to retire, and so on. Clearly, we needed to have life stage as a fundamental lens underpinning a customer focused approach but also looked at the more core emotional drives that motivate attitude and behavior – but again using CHAID to understand what ‘observable’ and more actionable variables can be used as proxies for different attitudes etc. which we could also use to complete our ‘framing’ of the market.

3) Identify segments with the framework - how do cells in the frame coalesce around the PIC criteria – to create segments with distinctive behaviour, and segments that can be easily identified. The creation of the frame and collapsing the constituent cells to create segments is an iterative process driven by data – looking for similar PIC profiles and collapsing these cells into nascent segments (Exhibit 2).

These hypotheses formed a living document that was constantly revised, fine-tuned or improved as it was tested, and new information and/or insights gleaned from an extensive program of research became available. This perspective on a framework allowed us to focus the subsequent research and analysis – critical in a study with this breadth of scope – to ensure we were not searching for an answer in a vacuum or attempting to analytically ‘boil the ocean’.

Exhibit 2 – Simplified Illustration

4) Create initial segment profiles - utilising all available data to develop high impact materials that address the business challenge. Like other segmentation methodologies – this project delivered the expected outputs including: segment profiles (who, what, when, where, why & how); data dashboards – key metrics for each segment; segment summaries - descriptive information for each segment at a glance; day-in-the-life stories – easy to digest pen portraits and segment info-graphics sheets. However we were also able to develop effective activity models – identifying the economic opportunity within each segment tied to the firm’s actual revenues in each segment, which is much easier to achieve if membership of a segment is readily identifiable. We also identified a series of KPIs to monitor by segment, to assess the impact of marketing interventions.

The outcome of this work was a dynamic segmentation that was actionable in terms of the marketing levers that the financial services company would have had at its disposal, and which delivered a measureable positive impact on the future of the business … specifically 

• Clear, holistic, view of the customer landscape as well as deep knowledge of the customers’ needs, wants, beliefs and associations, as well as current behavior
• A clear view of the customer opportunities available and their attractiveness (quantified value of the potential) - an understanding of what it will take to capture these opportunities – barriers to overcome and drivers to leverage
• The ability to tag their own customer database to allow more focused and targeted communication of tailored product offers
• A view on the ‘sensors in the ground’ that this company would need in order to track and manage performance in priority segments

Today, customers need to enjoy life now whilst also taking control of their financial future. They want transparency and openness in looking at solutions that will give them financial freedom, control of their lives, and help them to achieve their life goals. Our client needed to better understand this within a framework that would allow them to put their finger on the individual – send them specific messages and offer them tailored solutions that enable them to achieve their specific ambitions. The solution is a balance of robust analytics and pragmatism – recognizing how the business does its marketing and the marketing levers available to it. 

[1] Fonseca, Riley, Smith – “Are You Insured, Scarlett? ‘I Can’t Think About That Right Now … I’ll Think About That Tomorrow’. How MetLife imagined a new future for the insurance industry … and is delivering it today”. ESOMAR Congress Paper 2017

[2] Chi-square Automatic Interaction Detector (CHAID), developed by Gordon V. Kass in 1980, is a tool used to discover the relationship between variables. CHAID analysis builds a predictive model, or tree, to help determine how variables best merge to explain the outcome in the given dependent variable.