We specialise in customer analytics, information, and solutions for the customer lifecycle management. In terms of the customer lifecycle, we start with a Pre-qualification standpoint where we discriminate customers in terms of riskiness and the likeliness of them to take an offer if given an offer. This essentially is a function of data and Analytics, says Subhrangshu Chattopadhyay, National Sales Head, CRIF India.
Implementation of this in a solution which is configurable and flexible. The next thing is customer acquisition or underwriting, we look at two things: Intent and Capacity. These represent itself in a number of analytic customer solutions in terms of risk, fraud, app scores, profitability, income as well as personality assessment.
Fundamentally if you have data, you can build analytics and should be able to deploy it in such a manner that you can respond to the market. This is where IT solution comes in, where you can configure every analytical figure you have built, put the policy on top of it, cut-off stimulations and can configure underwriting strategy.
Post-Acquisition what are the key problems you should solve
- Look at behavior to predict loyalty
- The response of the customer when you make an offer
- If out of 100 respondents, you accept 60, what are the portfolio losses you going to take
Use cases are the same, in lending, it is for portfolio losses and for insurance it is the claim ratio. Use the data, do Analytics, deliver it in a solution where you can put Analytics plus policy, go-to-market strategy making sure that solution is configurable and is vendor-independent.
In terms of collection, what do you solve for? You solve for channels, cash-accounting, allocation strategy and feed on the street. Once you solve all of this you get the same thing used data, built analytics put it in a solution that is configurable and help customers become independent of CRIF change request.
Talking about the acquisition, he said; you qualify the right customer, make the right offer at the right time and the channel through which the offer approval is likely to be highest. By doing this you increase efficiency and standardise policy. You do all this through policy rule optimisation because analytics scorecard are information assets that are non-changeable for the medium term. Because you built it in certain data sets to look at a certain period in the future and unless the model has enough bad you don’t deliver competitive advantage.
How you deliver a competitive advantage? It’s the flexibility through optimised policies because enough bad scorecards takes time. Enough bad scorecard which are void of any external intervention. Like when you have demonetisation and Goods and Services Tax (GST) in the 12-month period it means all your scorecards are delayed by 24-months. Because that bad is not necessarily an externally independent bad. It’s an event bad. So, you generate advantage through policy rules and then series of scorecards whichever makes sense for your organisation in the stage of evolution. It could be riskiness at product level which is called bureau look-alike scorecard, an app-score if you have enough vintage, fraud-prevention score-card, a customer profitability score-card or income segmentation scorecard which can deliver a competitive advantage in business loans.
In terms of customer management, the first thing you want to predict is customer lifetime value. Does your organisation look at the customer as a single transaction or series of transactions through which you will finance the lifetime needs?
There are Non-banking Financial Companies (NBFCs) which has changed their vision statements from the former to the latter. You increase customer loyalty, by doing so you also increase profitability and another aspect that is Early Warning Signals (EWS) so you avoid additional exposure to those people who are likely to show stress. With this, you have a possibility of balanced transfer, customer retention, lifetime probability you look into either Indian Accounting Standards (IND AS) or IFRS 9 framework because lifetime computations have to be linked and the provisioning you have to make today for the loss of tomorrow. So, 100 percent of goods in the model doesn’t deliver any competitive advantage. The competitive advantage of the model comes when you have bad.
In terms of collections, what decision engine should do is actually segment, separate bad from the good. For every customer, it should actually predict in early collection phase whether tomorrow you have to make provision for it. And what at the contract or customer level is going to be loss given default and exposure at default. That is capital set aside. You need to also look at channel allocation, basic segmentation of which customer is more likely to repay if you put them through the dialler vs which customer is more likely to repay if you put them through agency A or is your agency A is performing better in delinquency bucket so that agency B could be shifted to agency A as well.
You impute the probability of recovery which is nothing but looking at roll back rates at your portfolio and benchmarking that with whoever is your relevant competitions. Now, you have these decisions, what workflow you can implement based on these decisions. User or agency management, rule-based allocation, integration with dialler, determining whether you collect through cash or through noncash. If you collect the cash letting the cashier in the branch know how much cash has been picked and how much to expect.
How does Analytics help?
Analytics by itself is an optional module that delivers an advantage. Information Technology or software is an optional module where if you have an interface which allows you to be flexible delivers a competitive advantage. It is not necessary you do it in one go. Whatever is the choice it is unto the evolution stage in which your organisation is.
How do we work with our customers?
We first identify the challenge. Challenge can be in terms of effectiveness, efficiency, or cultural challenge. Once that is identified, then we look into solution recommendations which are not always the big bang approach. It is to say what are the components that are optimal for you at this stage and what you should measure over the short-term and medium-term to look at your next phase of the rollout. Do you really need a scorecard to do better analytics then the answer is no? You may have a series of policy rules which if you optimise gives you better analytics at which time it is perspective and not predictive. And based on that you can move to predictive. Predictive is scorecards.
Implementation is configurable, integration is channel-agnostic and the key reason why people work with us is that we bring bureau data, and Analytics on top of the data. We also bring software solutions which you can host Analytics and policy. We do this in multi-layer engagement model which measures performance, so it’s not just built and transfer.