January 2018 will become a milestone in global financial history: PSD2 reforms coming into force will drive innovation within a centuries-old service model. Customers will benefit from easier switching from their primary bank and enjoy new alternative applications as main touch points with banks. At the same time, banks will face the challenge of not losing such touch points and staying competitive. Giving a customer a reason to stay with his or her current bank through providing customised communications could become a key target. As such, banks should consider a strategic move away from just being a reactive utilitarian tool to being a proactive personal financial assistant with the use of AI to maintain long term relationships with its customers.
Beyond PSD2 context
As of today, banks are at risk of becoming commoditised: similar services are provided at nearly the same prices with almost the same level of quality at any bank. The Open Banking environment is expected to boost competition among banks and fintechs with effective customer engagement being at a forefront. If a bank finds a way to attract a customer and make him or her stay in the long run, it will gain from the PSD2 challenge and increase profits. To stand out, a bank needs to become an inevitable part of a customer’s financial life.
How could a bank build reliable long-term relationships with customers? First and foremost, a bank has unique knowledge about a customer’s transactional behaviour – the data that could help understand a customer’s needs, interests etc. The appropriate usage of internal transactional data with AI and ML tools could help a bank detect and predict many things about a client: from simple ideas like breakfast eating preferences to more informative observations like travel habits or a new car purchase.
Just having the knowledge however, is not enough. It is important to use this knowledge and bring value to a customer. No doubt banks know how to offer their products contextually, a customer needs a new banking product, a card or a credit for example, infrequently – perhaps once or twice per year. However, a person makes transactions on a daily basis. With this in mind, a bank could use external merchants’ data to introduce appealing recommendations or tips that would help a customer to make wise financial decisions in everyday life.
Recommend as opposed to advertise
Comparing to direct in-app, e-mail or SMS advertisements, personalised tips do not have a promotional nature – they deal with a customer’s financial and lifestyle context aiming at increasing a customer’s engagement with the bank and cross-selling of banking products.
For example, a bank may help a customer to reduce daily spending. It may give contextual tips with valuable information about those merchants which a customer uses on a regular basis, letting him or her to stay with the same services but paying less:
“You often travel by Virgin trains on weekends where Off-Peak tickets are valid. Combining such tickets with standard fares when needed could save you about £68 for the last 3 months”
Tip scenarios vary depending on a customer’s situation and needs, and may also deal with budget monitoring, analysing and optimisation; prediction of future payments and things to-do; highlighting relevant services and many more.
Hidden opportunities within the data
To grow a customer’s Lifetime Value (LTV), in terms of the long term relationship with a bank and usage of more bank products and services, a bank needs to come up with a relevant tip for each customer pretty often, at least once a week. And of course tips for one customer at different points of time should be different. How could it this be achieved? A manual approach will take too long and be hard to manage at scale. Thus, automation will be a crucial solution to creating an appropriate and appealing content mix of recommendations for specific target audiences (tip receivers).
Huge amounts of useful information for customers is publicly available at merchants’ websites, like prices, discounts, special offers etc. It is possible to crawl through web-sites with the usage of Natural Language Processing methods, in particular Named Entity Recognition, searching for information potentially relevant to a customer. This information could be transformed into personalised communications, or tips, to be organised and sent automatically.
All tips generated will be automatically associated with real life situations for the customer. Nevertheless, from time to time a customer could receive a non-appealing tip. When such cases happen, the tip-generating system will take a customer’s feedback into consideration (swiping notifications in the app, for example) and relearn targeting models to avoid distribution of similar tips in future, thus maintaining customer satisfaction.
As a result, retail banking customers can receive highly personalized tips based on the mix of internal banking data and publicly available data about merchants from 10,000+ external sources. The tip-based approach allows for the creation of a service that understands a customer's financial situation better than anyone else and to help customers predictively and proactively solve everyday tasks relating to their finances. This boosts Net Promoter Score (NPS) and engagement with the client, enhances trust and increases the sales of core banking products by increasing the number of useful touch points with a client.