Making 'emotive banking' a practical reality
We all know that the world has changed as a result of coronavirus, and the needs and expectations of customers have changed too. In this 'new strange' reality, banks are going to have to step up their game in delivering customer experiences that demonstrate that they know their customers as people and not just account-numbers. As we saw following the global financial crisis, failing to focus on customers (rather than just financial or operational challenges), can leave the door open for more strategically-minded competitors to grab market share.
Whilst many of you have spent the last few months virtualising your businesses: for example, switching customers to digital and voice banking - a few of you have also seen this as an opportunity to take an accelerated transformation program still further - developing deeper insights into what customers really care about now, and using those insights to deliver a more personalised, relevant experience (others have recognised that they need to recalibrate their understanding of customers to reflect a post-crisis mindset).
This is where 'Predictive Behavioural Analytics' plays a role - it explains why apparently similar customers behave differently, what they are likely to do next, and what you can do to enhance the customer experience in response. By putting customer emotions at the heart of your understanding of them, you lay the foundation for 'emotive banking' - a more empathic, human banking experience.
About a month ago, Maveric Systems and I presented a webinar on 'Maximising Customer Value in Banking Using Predictive Behavioural Analytics' (you can access the recording here: https://lnkd.in/gcCZa97). This followed an earlier video with me and IDC's Gina Smith and Michael Araneta on the same subject (you can watch the recording of that here: https://youtu.be/zO_cltr42iI). In the webinar, Maveric and I discussed what predictive behavioural analytics is, why it is important and how a bank can start its journey.
The first thing to address is what we mean by 'behaviour' - for many data scientists, this term is used to describe a collection of 'observations' (a data set). By analysing the data, you can develop models that predict an outcome based on observations - an 'effect'. However, for predictive behavioural analytics, you should also be interested in the motivation ('cause') behind the effect - what prompted the behaviour that you observed. This is a fundamental shift in focus for most analytics teams. Also, because you are focused on motivation, the outcome is a de-facto prediction - when customers are motivated by 'x', they are more likely to do 'y'.
People are motivated by a range of factors; what they need (functional requirements), what they want (behavioural/emotions requirements), the influence of 3rd-parties (e.g. recommendations / star-ratings), their internal biases (both cognitive and unconscious), what they are trying to do (the context) and their experiences. It is the blend of all of these factors that results in a decision that itself results in an action. Whilst this can appear a daunting array of things for you to consider, it is actually relatively straightforward to implement (albeit not necessarily trivial).
When you have a more comprehensive model of how customers think and behave, you can take more focused, proactive and personalised actions that powerfully resonate with your customers, resulting in increased revenues, reduced costs, greater loyalty and all of the other business benefits of increased customer satisfaction. For example, knowing that a valuable customer might be feeling dissatisfied and taken for granted, may stimulate you to make a proactive approach to that customer before they start thinking about defecting to a competitor.
What it will take
Data Sources - both structured and unstructured data, covering both operational and experiential facets of customer data.
Data Pipeline - handling the ingestion, management and storage of an analytical data set
NLP / NLU - A method of extracting meaning (especially emotive) from verbatim narratives: it is this capability that identified what customers care about and why
Behavioural Modelling - using the 6 dimensions I noted earlier to score the relative importance of each factor & develop predictive models
Event Stream Modelling - identifying when a 'moment that matters' has occurred to and which customers
Interface to Operational Systems - so that the results of modelling can be applied to operational decision-making
My advice would be to start off with a pilot / proof of value that focusses on a specific facet of your business and from which improvements at a strategic level can be made. This should normally take no longer that 9 weeks to deploy. It will also allow you to identify other future application areas.
Emotions are fleeting and personal, how can an accurate model be built at the customer level? My advice is to focus on the event that causes the emotion, not on the customer, then consider each significant event in the context of prior events for each customer - you are looking for subtle adjustments along a continuous timeline, not a seismic shift.
What are the challenges of handling valid data sets? Predominantly that there is often not an effective Customer Data Platform (CDP) that integrates the necessary data - it often resides in multiple siloes, or has significant challenges with accuracy and / or completeness.
How can I take the micro-level insights of behaviour and apply them strategically? As you analyse what is motivating the behaviours you observe, common 'causes' become apparent - by asking customers the right question (and it really is just one question) customers will tell you what is on their mind and why. If enough customers give the same response, you have something that needs a systemic response. Nonetheless, it can still influence operational questions at the individual customer level.
How can you know what an individual customer is feeling without asking them directly? There are aspects of human behaviour that are predictable - behavioural economics has shown us that there are over 150 ways our 'rational' thinking lets us down. That said, no model is perfect, nor does it have to be - it just needs to be better than what the customer expects or the competition can deliver.
- Customer emotions are detectable from verbatim narratives and we know that they influence the decisions customers make
- Emotions are part of decision-making framework that can be modelled to predict outcomes
- By predicting what a customer will do and knowing why, you can take a proactive action to optimise an outcome
- Implementation is straightforward, although not necessarily trivial
- You can start small - at the strategic level - and move into an automated operational model when ready
Don't forget, you can watch a video of the full webinar here: https://lnkd.in/gcCZa97
Peter is an award winning expert in using a combination of data and behavioural sciences to lead transformation in the field of Experience Management (XM); encompassing Customer Experience (CX), Employee Experience EX) and Partner Experience (PX) .
Over the last 3 years,...