By Sachin Rajat Sharma, CPO, nexus | 24 August, 2020
Are financial institutions ready for hyper personalisation?
Sachin R Sharma with Vipul Gupta ,23 August 2020
You were talking about a holiday and lo and behold!, an advertisement for ‘weekend getaways’ appears on your social media wall. ‘This is spooky, are they listening to me?’ . Well as it turns out a sophisticated algorithm is curating advertisements customised to your very unique profile, timed to your recent activities and web searches.
Another day you wonder – ‘Are they really listening to me? As you struggle to explain your credit card transaction issue to your bank’s customer service agent who seems to be more interested in selling you a personal loan that you do not need.
Despite financial transaction data widely being accepted as the most valuable data set in customer profiling, traditional financial institutions struggle to make effective use of it.
Digital natives are used to relevant, specific and timely marketing interventions. Your social media wall is tailor made for ‘you’ no second user experiences the same feed, in effect you are a hyper customised ’segment of one’
For your bank credit card proposition on the other hand you must fit into more or less one of these categories
● Cashback crazy
● Miles Junkie
● Rewards seeker
● Foodie
● Shopaholic
Too bad if you cut across these categories
The reality is that most of us are a little bit of all, and our preferences constantly change from time to time. As such we will end up having multiple cards with constantly shifting loyalties
This may have been good 10 years back with little digital penetration and most of us dependent on the ‘physical’ media for making our choices. But not anymore, big tech has completely changed the game and banks need to keep up.
So what will it take to bridge the gap, and can it ever be bridged ?
This article lays out a 5 point plan to get there
I. Get the data in one place
The bank started operations in 1961, a time of paper ledgers, in the 80s they introduced the first computer systems and by early 90s they had a state of the art core banking system to manage deposits and payments
A few years later credit cards were introduced followed by an overhaul of lending systems in the late 1990s. The databases of all these systems were maintained independently and nothing has changed in the last 20 years. The only place you can see a ‘consolidated’ view of the customers holding is in the servicing system that has painfully sourced data from a large number of federated databases. No access to a consolidated data set and limited opportunity to drive customer engagement
The problem statement ‘how do you consolidate federated data into an all encompassing, accessible data store or data lake’
Financial Institutions face a three-pronged challenge in leveraging the real power behind the data that they hold across myriad systems – most of them legacy and sometimes running into the hundreds. The problem I see most often is that most of the institutions are solving for only a part of the problem!
Physical availability of data (Breadth and Frequency)
Hurdle: Traditionally the data availability has been primarily dictated by the institution’s ‘Reporting’ needs. This limits the number of data points readily available for consumption and also the frequency of refresh is also in most cases monthly. This fundamentally affects the capabilities of any analytics function because they then have to resort to ad hoc requests to secure required data and spend ~80-90% of the time preparing the data.
Solution: Technologies like Hadoop / HaaS etc provide almost unlimited storage at a fraction of the cost earlier. Financial institutions can leverage these solutions to create a Data lake where ALL the data from ALL the systems can be dumped at a frequency of choice. This would ensure ready and quick access to any data element for the user
Understanding of data:
Hurdle: But will physical availability alone solve the problem? While it might seem so, often due to multiple legacy systems, products and its features are booked in complex fashions across systems. There is limited knowledge of how to filter for the right data and also how to join data elements across systems. So you might have 100k data elements in your data lake, yet not know how to use most of it!
Solution:
– Value your Subject Matter Experts (SMEs)! They are the ones who can partner with business and unlock the true value of the hidden gems in data. Banks need to start treating their System and Data SMEs with respect. In today’s data-driven world they are worth their weight in gold. Banks on account of myriad cost pressures over the last decade (Post-Lehmann) have failed to realise the importance of this aspect and substantial such SMEs have been lost to ensuing expense line management actions. Hope they do not repeat the same mistake post-Covid.
– Data dictionaries – While banks have started creating individual data element dictionaries with some basic descriptions, these are often not enough for actual usage. Banks need to create ‘Data Block’ dictionaries which store logical joins and filter conditions to arrive at a usable data attribute. For e.g. I might need to use 5 different tables in my cards system to arrive at ALL valid transactions by a customer and I also need to know the conditions for identifying only ‘valid’ transactions. This will ensure widespread and consistent usage of data across the organization
The Organisation’s Data Culture
Hurdle: Banks have various units working in silos and often monopolistic ownership of data. This is a strong hurdle in building a data-driven decision-making culture. Any new initiative which cuts across functions seems a mammoth task to navigate in the highly matrixed organizational hierarchy, killing many data dreams even before they are born. Also with a traditional mind-set the focus for the failure of any new initiative is too often focused on blaming individual team members.
Solution: Traditional organisations need to be more open towards ‘test and learn’ or ‘fail fast’ experiments to break new ground. The culture needs to move away from punishing every failure which will be a key enabler for a cutting-edge analytics culture. Data Silos need to be broken and its usage should be democratised with the right controls. Many banks are starting to adopt Agile ways of project execution with teams organised as Tribes and Squads to overcome some of these legacy shortcomings.
II. Make your data work harder
While banking transactions data is undoubtedly valuable, it is not enough for a ‘contextual engagement’
For that you need, search, location and off-us transaction data as well. Going back to the facebook example, not just every like and reaction on facebook is tracked but through a product called ‘Pixel’ facebook also gets access to your activity on partner websites.
In brief you need to make your own data work much harder and have a solid foothold in the data partnership ecosystem
Better usage of internal data
In many cases the bank’s own internal data hasn’t been utilised yet to its full potential. Banks have a wealth of information about the customer revealing Financial, Transactional, Behavioural and Demographical dimensions across time. With physical availability of data, knowledge of how to use the data and powerful processing tools that are now easily available, banks first need to unlock this potential. For e.g. 1) there is a plethora of customer loyalty gaining potential in acting on real time data of ‘failed touch points / transactions’ for a customer and helping the customer out of it at that very moment. 2) Spending pattern changes for a customer over the years might indicate its time to upgrade the customer to affluent banking although he might not have placed the money with us
Data as a currency
Consumers increasingly are willing to provide access to their personal data in exchange for rewards. We walk into a mall and they ask for your details in exchange for free Wi-Fi, we agree to go ahead without giving it a second thought, sounds very familiar isn’t it? Within the ambit of data privacy, banks can increasingly integrate a rewards / convenience approach into their marketing strategies to gain better and informed contextual access to relevant information. For e.g. turn on location services on the mobile app in exchange for location specific discount offers?
Data partnerships
Banks can augment their understanding of the customer through data partnerships and external data (Bureau, Social Media listening, Product & Brand Sentiment Analysis, Alternate data – Utilities etc) of course by doing so responsibly and within the relevant data privacy laws and spirit.
Ability to remarket and build a dynamic 360 view of the customer can only be achieved with data partnerships There is limited understanding of these data partnership eco-systems within the traditional banks and the marketing teams are mostly dependent on agencies to ‘buy’ the right media for campaigns. This gap needs to be actively bridged. Working with google and facebook is not really a question anymore. How to make it effective is the question to ask.
Some readers may argue that banking regulation makes it significantly more difficult to strike data partnerships. This is a fair point, however at the same time with more regulations such as GDPR coming into effect, the playing field between bank and non- bank entities is evening out. It is time to actively engage regulators in a data privacy discussions, where consumers data can be shared safely and transparently for better experiences
III. Engage digitally and track performance
For ‘segment of one’ to work you need your customers to be engaged digitally. While most banks are investing heavily into their digital banking apps, getting all customers to use them have been an uphill task. Read my article Preaching to the converted
The incumbents who have a substantial user base using the mobile apps (at least 50% of active account logging in regularly) need to move from ‘view and transact’ to ‘active engagement’
Also after significant effort the best incumbent banks get 50~60% of their user base on the mobile banking app, while nearly 100% will be using google and facebook services. In many instances these may be the best channels to reach not just new but your existing portfolio customers as well.
The metric for digital engagement measurement also need to evolve – ‘monthly logins’ need to evolve to measuring monthly ‘#digital transaction’, ‘new product applications’ and ‘participation in portfolio campaign’
The metrics for digital engagement should cover the following high level categories at the least:
Customer engagement indicators
‘No of successful logins’, ‘No of transactions initiated’, ‘Trend of transactions – Stable / Increasing / Decreasing by customer’ etc. ‘Count of types of transactions by customer’ et al.
Customer experience
‘No of failed logins’, ‘ No of failed transactions’, ‘No of app crashes’, ‘Time taken for specific transactions’, ‘Time taken to complete applications’, ‘App up-time %age’ etc.
New Business potential
‘New product categories viewed’ and ‘%age conversion’. These can be triggers for the bank’s digital remarketing efforts.
IV. You have the data, run campaigns
Most incumbent financial institutions have strong analytics teams. Large group of data analyst work on business insights for management decision making and find potential opportunities for marketing interventions
X no. clients had >$Y spend in the previous year now that has dropped to $Z – opportunity to ‘win back’
X no. of deposit clients have a high propensity of taking up a mortgage loan – opportunity to x-sell etc
Typically these campaigns cycle would be 2-3 months with high degree of ‘human interventions’
Automating a typical campaign cycle is a major challenge. The good news is there are a number of third party solutions now available for campaign automation (to some extent)
E.g. Adobe Campaign management
Automation is not a luxury anymore, especially if the bank has a mass market client portfolio. The challenge is to automate all components of the campaign life cycle. Not just identifying opportunities but also fulfilling them with the least amount of human intervention .
V. Crank up the marketing
Having the data, partnerships, campaign automation and engagement platforms in place now it’s time to ‘crank up the marketing’
This is perhaps the most challenging piece of the puzzle.
Consider this, Facebook has ten of thousands of advertisers generating content which when delivered to you in the right context feels completely customised to your needs.
How does a bank compete with that?
There are two problems to solve
- How does the bank prioritize between the multiple campaigns that the client may be eligible for at any point in time?
- How does the bank ‘customise’ the content to hyper personalise the message for that particular user?
The former is usually settled in order of ‘business value’.A prioritization ‘committee’ will probably choose an insurance campaign with a higher revenue opportunity over a low margin time deposit campaign.
In the short term target oriented scenario, this may be a reality. However to effect a long term change campaigns should be ‘algorithmically prioritised on ‘Customer lifetime value’.
For example –
The insurance campaign is prioritised, however a high value deposit into the customers account triggers an ‘event’ that brings the Term Deposit campaign upfront. The campaign message shows up in the bank mobile application with an easy process to set up the time deposit.
Recommendation engines – Platforms like Amazon, Netflix etc are heavy users of Recommendation engine algorithms to personalise customer offerings. These are based on customers past searches, purchase behaviour, what content they watch etc. It’s like creating an individualised marketing offer, pitch or nudge for the next action that the customer is most likely to take. Banks can similarly marry the plethora of information at their disposal and curate personalised offers on a ‘Offers for you’ wall for the customer. For e.g. 1) A customer starts buying from MotherCare or Kiddy Palace in Singapore (Indicating that there is a high probability of an addition to the customer’s family). There can be a personalised offer to the customer for discounts / offers for baby and post-natal merchants. 2) Customer starts paying university fees through his account; the bank correlates his income and fees and accordingly offers an educational loan.
Some resource on Recommendation engines:
Recommendation Engines | Recommendation System in Banks
A simple way to explain the Recommendation Engine in AI
This is easy to illustrate, difficult to execute. A lot of elements such as data engineering, AI programming and most importantly and mindset change towards long term engagement needs to come through to achieve this
The second part of the problem – ‘customised content’ is even more challenging. Any customer offer needs to have a business case behind it, $100 cashback for a spending $2000 over the next two months. 10% rebate from the first year premium if you invest more than $1000 per month. For asset products there can be further customisation such as risked based pricing. For every offer marketing content needs to get generated across multiple channels, in app, sms, email, digital banners etc. Not an easy task with limited marketing and business bandwidth
Again innovative third party solutions need to be considered. Content generators from the gig economy can be another source. In brief the marketing content bottleneck needs to be resolved for true hyper customisation
In conclusion, amongst the incumbent banks that we have in view, there are several who have taken great leaps towards achieving better use of data for customer engagement. However the foot can not be taken off the pedal just yet, there is a long way to go before retail banking customer can truly say ‘my bank is listening to me’
…
Disclaimer: The views expressed here are our own and not representative of our organisations in any way. Commercial or brands mentioned in the article are only for creative reference and should not be considered as the authors recommendations.
