Getting a loan approved is a big deal in itself. One is already fuelled with the aspirations due to which the loan is required, and the positivity is so high that nothing can kill the joy.
But then, comes the entire loan approval process, when and where the spirit gets significantly hit at the very least. The funniest of the parts being that one has to prove that they do not really need a loan to get a loan, hence, all the credit scores, debts, liabilities, assets, income statements, etc. need to be shown to make oneself loan worthy. Not just this, the proof of employment, residency, a suitable guarantor, the regularity of the first two and some other things need to be arranged for as well.
Though, all these are the things most people going for a loan would have in place, they often create a significantly bad customer experience. One might feel that since the banks keep calling us all the time asking for our loan requirements, it would be a cake walk to be through the process, but the data speaks otherwise.
We know this would sound unbelievable but about 35% of the total loan applications get rejected annually. What make this number even more disastrous is the fact that about 60% of these 35% applications get rejected due to unavailability of documents/guarantor or a certain credit score which is not as per expectations.
The result- the banks lose. To further define the magnitude of this problem, the Indian banks alone stand at losing about INR 15,000 crore owing to the aforementioned.
A very critical thing to note here is that most of the processes involved in sanctioning a loan are heavily manned; it would be a redundancy to say that they are manned by humans and with customer data being stored across different platforms, it becomes really difficult to consolidate them all at one place at a faster and more customer convenient speed.
However, technology has always been a saviour when it comes to handling data. This time is no exception, and with the abundance of data that we have now across channels, Artificial Intelligence (AI) is emerging out as the saviour.
Let us have a look at how AI can help in loan processing in specific.
Or, as we call them- filling up the loan form and gathering the documents in order to prove one’s capability of being able to pay the loan back. With so many data that needs to be filled, a dedicated employee is usually required to assist the customer with the filling process. One of the major Indian banks even introduced the system whereby an assistant, available on call could come over to the customer’s place, have the information filled in a tab and get the formalities done. This particular bank in point saw an increase in its loan application to the tune of 33% post this feature was inaugurated. The same was then followed by self help kiosks- a hardware based virtual assistant which did not really interact but provided one with enough options to select what they wanted. Once the basic things were done, the person was then redirected to a counter where the remaining process could be worked upon.
This example can be cited as the very first advent of AI in the business. However, that was back in 2013 and Artificial Intelligence has seen significant advancements since them. With Virtual Assistants getting humanized by the day, basic assistance is the commonest of the solutions at offer.
The differential however now is the extent of help. Virtual Assistants have been developed which can not just help one with the basic form filling but also help find where to access what documents and in multiple cases download and aggregate them for the person themselves.
Cross verification of these documents with the records is another domain in which advancements are being made. Based on NLP and DL algorithms, some of these AI based systems can also raise red flags for the customer themselves, telling them what needs to be done as a measure to mitigate the concern and have a hiccup free loan processing.
Considering that the speed of such automated systems is usually way faster than that of a human, it results in significant reduction in time and effort. Thus, while the customer gets a better experience, the firm saves on the money and improves efficiency.
Loan Interview and Due Diligence
There was an episode in one of the famous Netflix series- Black Mirror which talked about how a chip inserted during the insurance claim interview could work through your brain, capturing all the memories that you had in store. Now, that is still at a sci-fi stage but Sentimental and Trait Analysis has been made a possibility. We had mentioned in on of our earlier blogs (give link to the machines comprehend human emotions blog here) that how machines can be made adept at understanding the human emotions and then be made to respond accordingly.
Similarly, it goes with trait analysis where machines pick up cues about how a person is behaving or even other visual cues to understand the basic nature of the person concerned. Both these analysis (sentimental and trait) when combined with a carefully thought psychographic analytics test can make it very difficult (if not impossible) for a person to forge their way through the loan interview and commit a fraud later.
Thus, the probable defaulter gets red flagged in the beginning itself and the due diligence gets further options to be corroborated.
Once the loan has been processed, the next challenge comes in the form of loan servicing. If we talk about the Indian context in specific, it is still not uncommon that the bank agents come to collect the cash or cheque. In most loan payback cases, especially the ones where an auto-detection has not been programmed, manual intervention becomes a necessity and more often than not, due to unavailability of the payer, the visits have to be rescheduled- thus wasting time, money and resource.
Moreover, considering the individuality of the data at hand, the entire process becomes heavily human dependent considering that the relevant information needs to be sourced from separate channels. A date warehouse can be of great assistance in this regard. A CDP more so because it will not just contain information about what the consumer is doing right now but will also keep a record of any financial or otherwise transaction, he/she got into ever. Thus, the entire process which involved dedicated employees aggregating the data into a CRM so that the communication loop could be started, would then become automated with the employees’ bandwidth being utilized elsewhere for more productive pursuits.
For a financially less educated audience, such systems can be responsible for generating necessary awareness as well, as well as the past transaction history can help the banks determine the amount that can be loaned.
We cannot deny the fact that human intervention would still be required but the extent would be reduced significantly.
The last in this blog, but evidently the first domain that gets touched by any technological advancement – “automation and consolidation of the documentation involved.”
An AI based system can assist the clerical department to cut costs by developing bills, reviewing the documents sent against the required, creating tickets, charge slips, maintaining and monitoring the list of investments, automating credit score requests and storing the responses (as well as updating the records once the same is received), verifying stock transactions, determining credit worthiness, etc.
The list simply goes on and on; the idea however is, if you can name one documentation process, there is an utmost certainty that the same can be automated.
(AI has data at its foundation and with the data increasing over time, it is obvious that AI would be seen in bigger roles than the ones it currently has assumed.
At Racetrack, our endeavor is to keep pace with these advancements and hence the ones mentioned above have already been incorporated in our solutions. What lies next, only time will tell.)