Do we need Human employees in a Bank/FI for customer acquisition?

There is a lot of buzz around Artificial Intelligence, Machine learning, Big Data, Algorithm these days in mainstream media, thanks to popular tech companies like Google, Facebook etc. Speculations are doing the rounds on which professions would cease to exist in future due to this technology shift. The discussion ranges from impact on Education (teacher-less school) to Journalism (robot-generated reports) to English literature (robot generates poem!). How about the banking/lending industry and its associated workforce? Would we get into a situation in future where we just need a basic minimum workforce to operate a bank/financial institution instead of the current large manpower? Or are we already into it?

Most of the financial institution’s (FI) biggest challenge any day is to have better profitability and an efficient system. The major negative contributor against both of them is its large workforce and the inherent inefficiencies that grow over the years. The operating expenditure of salary cost starts eating into the profitability. The work queue reduces the efficiency. We try to solve these problems by adding more people into it or by doing a new role definition for existing people or trying to tweak around the process to have lower manpower in doing the repetitive jobs. Is this going to continue or is there a new way of doing things getting built around the corner?

If we look at the modus operandi of an FI in an over simplistic way, the basic purpose of an FI is to 1) Generate funds at X% rate from market or bank or some source, 2) Lend the money to customers at Y% rate and 3) Collect back the money from the customers. If we focus only on #2, the proposition is — can we reduce human intervention as much as possible or are we at optimum already?

1) Marketing — getting hold of a probable customer: In mature markets, we are already seeing efforts in moving from traditional marketing to digital marketing. We are in a situation where the first interaction of a customer with an FI is happening over digital mode, be it in Social media or at Loan aggregator pages. It’s not through the earlier traditional models. If an FI focuses only on urban customers, in all probability they have already gone down with the traditional Marketing budget and man power.

2) Sales — making the deal and collecting data about the customer: The primary job of a Sales rep is to convince a customer with the rates and offering and making him agree to a particular scheme from the FI.

Sales Executive used to fill up long hard-copy forms for collecting the relevant details of the customer. These used to be eye-balled by Operations officers at back-office who would generate a digital footprint of the data in the Core system via data entry. This model is changing. Sales team is generating the data himself via Mobile based CAS solutions. In many occasions, even the customer himself is finding the FI, sharing his information himself.

The Loan aggregator sites are playing the intermediary role in showing the offering from various FI online. Customer is generating his own data footprint in the system there. In some scenarios, the customer may still call up a Sales guy who can come to the home / workplace of the customer and fill up all the relevant details on his behalf in the Sales rep’s Mobile solution. In either way, the data entry profession (Operations) at back-office is fast coming at risk.

3) Credit Analyst — finding credit worthiness of the customer: Whenever a new customer comes for a loan to an FI, the standard operating procedure has been to search for him/her in the FI Internal database for possible frauds / delinquency earlier and also to search in external credit bureaus. Based on the search result, the Credit Analyst would take a call whether the customer is worth a loan or not.

These search triggers to Credit Bureaus are already automated as most of them provide API/Web services to call their services real time. The results are available in XML (machine readable format). If we want to automate the decisioning also, its just a matter of putting the CA brain into a rule engine.This may sound simplistic. But possibly this is the biggest block so far in going for a human-interaction-less banking.

The decision quality would depend on two factors — availability of data and a robust rule definition. This is set to improve in coming days with more availability of data. Many 3rd party sites provide a huge data source about potential customers already and based on customer consent, FI core system can fetch results about a particular customer from there. The rule engine may need to learn itself while it gets trained with more and more data every day.

Instead of depending on a static algorithm/pattern, machine learning approach can be used to learn and improve on the data everyday for a better search and also predicting whether this customer would default in future or not. India still lags on this front due to unavailability of good data. There are instances of similar operation in mature markets already.

Parting thoughts:

Are we already in a position where we can go for a human-interaction-less FI? The algorithms are getting built and computer scientists are raring to go to solve this. But the implementation would depend on the data availability mostly. Will it happen in near future? This has already started happening in some markets. Traditional Financial Institutions may not like to jump into this mode immediately, but alternate banking models are already taking foray into this models. It’s possibly a matter of time when you would see the customer acquisition process in banks/Financial institutions being run completely by Algorithms and a few Data Scientists!