Predictive Analytical models can be built to use Transaction behaviour of customers. Products suggestion have to be relevant when we predict based on data available.
Real time location usage
Customer real time location tracking coupled with predictive model will help to provide relevant suggestions. Prediction can be done based on transaction behaviour. But to get this correct we need to ensure that its relevant for the customer. Relevancy can
be done using customer location tracking. This can be done using mobile banking application installation. While the app is installed we need to integrate the mobile banking application with Google maps access. This is currently used by most of the apps as
part of the installation (Apps requires the customers to agree the usage of Google maps, phone books, face book page, twitter etc., and this is legally allowed when done with consent).
Example of this combination
1) When a customer does a transaction in baby store for the 1st time it’s a prediction that new member of the family has arrived (or) about to arrive. This prediction can go wrong when we are relying only on transaction data. Customer could be buying a gift
for his (or) her friends kid. In order to make this prediction validate we need to use the real time location data to verify. Using the mobile banking application access to Google maps we can validate whether the customer was present in any hospital recently.
Positive results of both the outcome would help to guess product very precisely. This can be further validated using the customer previous transactions which could have been frequent pharmacy transactions (or) hospital within last 1 year.
2) Repeated visit to a new location can provide insights to working on different location. This can give opportunities in suggesting products like travel insurance, add on card to family etc.,
3) This location usage can be used even to identify Fraudulent transactions. When a customer not present in a location from which a POS transaction happens can be a fraud trigger.
Social Media behaviour
When new relationships are created we should try to get customer Face book / Twitter like social media id’s as part of account opening. This should be used to send custom built messages to make the customer follow the bank face book page. Social media behaviour
will be very much helpful to validate the predictive analytical outcome.
Example for this combination
When a customer frequently visits a super market like Wal-Mart for his / her regular purchase we need to check their face book behaviour in terms of likes (or) comments for Wal-Mart (visit to Wal-Mart can be identified using transaction behaviour & Face
book behaviour can be checked when we make the customer to follow the bank). This combination would provide insight to the bank to predict a reward program specific to Wal-Mart. Transaction pattern would let us know Wal-Mart is used for purchase but Social
media behaviour would lets us know whether customer will stay with Wal-Mart before we propose the product. If customer comments a negative feedback to Wal-Mart will help to decide that prediction can go wrong.
Face book location ping can be used to identify customers visiting locations accurately. These will give new opportunities when a customer uses other bank card in those locations which can’t be identified using transaction based data. Same way customer likes
can be used to offer more precise products.
Social media can even show some important events in customer life like job promotion (can be verified with increase in salary if credit happens in the same bank), child birth (with transactions in hospitals/ baby store etc.,), Child leaving for college (transaction
in university buying application form) coupled with face book congratulations / following of new college.
Mobile Banking App with location based services
Mobile Banking application can use the location and provide nearby ATM’s / Branches / Restaurant’s that will give reward points for card linked to account. Predictive model combined with behaviour analysis and location should be used precisely for suggestions.
Australian Bank Westpac is providing location based alerts based on mobile banking like giving the partner bank ATM’s when a customer travels to United states (to save from withdrawal charges), Customer at Airport asking him to inform the bank country of visit
to avoid oversead card transaction rejections etc.,
Bank are looking at App stores as next step to cover the customer in 360 degree using mobile Banking. All the above have to be combined in predictive analytics to suggest right product at right time. This model needs to capture every transaction and every
interaction through various channels. The outcome needs to be constantly updated based on daily behaviour.
In addition to this Predictive & behaviour analytics can be used for the following:
- Segmentation – various phase of relationship, demographics, behaviour, attitude, Peer group with similar combinations
- Personalised communication & product offering which will reduce the marketing cost
- Identification of new ATM’s (or) kiosk (or) advertisement hoarding based on transaction concentration
- Portfolio management strategy based on behaviour, demographics and attitude for wealth management products.
- Customer survival time
- Customer profitability like online channel usage only, phone banking channel, mobile banking channel (or) branch channels.