The rise of the self-adaptive behaviour models is changing the game when it comes to customer profiles and the understanding of their individual preferences.
For a long time, the market agreed to accept pre-defined clusters of profiles, called “personas”, to determine someone’s behaviour. People were viewed on a spectrum of conservative to aggressive according to their investment options or risk tolerance. Businesses then created a strategy for each persona, or clusters of customers. The problem with this is that each client is shoehorned into a persona, rather than seen as someone with personal needs. In other words, strategies are shaped around an imaginary and static persona and the inconsistencies that make each client unique are ignored.
Most financial institutions and vendors claim that using personas enables a business to personalise offers for clients and scale-up in a cost-controlled way. For this reason, many companies invest heavily into this strategy as a data-driven marketing effort. Even with the evolution of artificial intelligence and machine learning, the most advanced technique in banks I have seen recently, is based on personas, but the number of demographic identifiers has increased. For instance, instead of conservative, today you can be called a “father of a family” or a “female executive”. More effort is spent capturing new personas than understand how people in existing personas react to client offers. And by effort, I mean millions of dollars.
In reality, preferences are different due to individuality. It means that a persona can only be perfectly accurate if there is a one for each individual client, which obviously eliminates the scalability of a persona-based strategy. It’s always better to think about each client individually, with their own lifestyle, behaviour, and reaction to stimuli. It’s also important to recognise that people change over time.
Let’s take two examples: think about a young boy in his 20s with a low-income job and no savings. Now imagine a man in his 70s, also with a low income, and equally no savings. Both would be classed conservative regarding their financial state. The common conservative persona would put both into the same profile. But it’s totally different when we step away from the static point of view and consider what they want to do about the future of their personal finances. In other words, if your decision system doesn’t know why they are equal but will respond to different nudges, it could make your marketing offers irrelevant. So, how to do you create more efficient marketing campaigns?
The answer lays in self-adaptive behavioural models. This novel approach, based on machine learning, does not pre-define an expected average behaviour of a user (persona). Instead, it follows the individual. It is fed not only by past data, but also current transactional data and becomes a live indicator, linking the past with present and anticipating the future behaviour and needs of a consumer. Ranking a customer’s risk tolerance can then show not only how they will use their money, but why. The models adapt automatically, not only with data generated daily by the transactions of an individual, but also with the imputes created by the customer accepting or not your actions. Individual propensity is the future of the profiling mapping.
Another reason to adopt these models is that you, as a bank, can save significant budget by not having to track and maintain the perfect personas, which require constant updates in a constantly changing world. Needless to say, these models will be relevant in different regions and countries as opposed to the personas-approach. Indeed, it’s a huge effort required from teams of data scientists. Self-adaptive behavioural profiling models do it automatically, self-adapt, learn with its own mistakes and all in real-time in a much more efficient way.
Your customers will thank you for presenting offers that match their needs, and you will benefit from an even greater low-cost scaling capacity
Pablo Morales | CEO of London Analytics.