This website requires JavaScript.

50% of loan rejection happens on AI: CASHe CEO

Our default rate is 1% every loan cycle, says Ketan Patel

Nov 6, 2019 by Ruiyao Luo
50% of loan rejection happens on AI: CASHe CEO

Ketan Patel has built four companies from scratch while he was working with Kotak Mahindra Bank.

I have been a startup specialist within Kotak. For me, this is an extension of what I've been doing in Kotak, says Ketan Patel, the CEO of CASHe.

In an exclusive interview with The Passage, Ketan Patel spoke about India’s digital lending space, millennials spending pattern and his aspirations for the company.


The Passage: How did you manage to grow so fast?

Ketan Patel: We are looking at the segment that nobody wants to give loans to. The segment itself is still young. We are very small compared to what the opportunity in the segment is.

We have decided we will keep the user interface so simple, and decision-making so fast, that people who take loans from us will refer their friends and people who they know to us. We want to keep it as simple and as quick as possible. So, that helped. The user interface has been seamless and the entire journey for the customer is on the app. And in the best case, we end up giving a loan in 45 minutes.

We have a very strong referral programme where we encourage our existing customers to refer their friends to us. So at this point, 30% of our new customers come from reference.

The Passage: What made you get into lending space?

Ketan Patel: We realised the lending companies were looking at bureau scores to lend. We felt bureau scores were not the right indicators of deciding whom to lend, whom not to lend, because bureau score is a diagnosis. It treats you on your past. And past may not be the right way to judge a person. A person may have committed mistakes in the past, but what we look at is prognosis. We treat the person as he is today and looking at him today if he will be able to pay the money back in the future.

The Passage: How do you assess the credit worthiness of your prospective customers?

Ketan Patel: We have two algorithms. One is Social Loan Quotient, which looks at the mobile data and social behavior of the person to assign a score. The other is a goodness measure where the financial parameters of the customer are considered against the requested loan size, his salary, the tenure of the loan etc. And based on the score he gets in goodness measure and SLQ, we give loans.

The Passage: Do you look at CIBIL scores? Where do you get your data from?

Ketan Patel: We don't look at the CIBIL score.

Phone gives you a lot of data. The kinds of app the person is using, is this his primary phone, whether he's using the phone for dialing, does he receive calls, the number of SMS, his lat long data. Assuming somebody's saying, he's working in one part of Bombay and staying in some other part of Bombay, but his lat long shows he has not been to either of the two places in the last one month, means this is a guy who's trying to defraud you.

The Passage: Do you read the content on the phone?

Ketan Patel: No, we don't. It is just a data point. We don't read the SMS, we don't want to read it. It doesn't make any sense for us. But we would look at the number of SMSes. The count is more important, because the count will show whether this is a primary mobile for you or not or whether you're wiping off your mobile.

The Passage: Do you vet the user’s social media accounts?

Ketan Patel: We look if the user is on Facebook, LinkedIn etc. We don't want to know his friends, whether he has friends or not. But we look at how long has he been there. Because, if somebody creates a profile a day before to apply for a loan, that’s a red flag for us.

The Passage: What kind of loan categories do you have?

Ketan Patel: We have a 61 day product, a 90 day product, a 180 day product, a 270 day product and a one year product. For one year loan, the customer gets maximum four times her salary, in a 270 day loan she gets three times of her salary, in a 180 day loan she gets two times his salary.

We give loans between 5,000 - 2 lakhs rupees. The average ticket size is 40,000 rupees.

At this point in time, our default rate is 1% every loan cycle. All our products are equated monthly installments.

The Passage: Walk us through SLQ and the role of AI in it?

Ketan Patel: AI is the gatekeeper. 50% of the rejection happens on AI. The entire SLQ has been created internally by our data scientists.

We wanted an alternate way to judge a person. We felt that this is the best tool, a social loan quotient or the social behavior factor is the best tool to judge a person.

What we do is once we get the data, we put you in one of the communities. At this point in time, there are 90 communities. What we mean by communities, is people with similar characteristics.

And we see how the community performs. If the community has performed well, you will get a better score. If the community has not performed well, you will not get a good score. If you don't get a good score, we will not be able to give you a loan.

We end up approving around 8% of the people who download the app.

The Passage: How big is your data and engineering team?

Ketan Patel: So my tech team is 44 people and my data science team has six members.

The Passage: What’s new on the tech front besides improving on the SLQ system?

Ketan Patel: We are trying to do something on blockchain tech. We have a launched a system called Buddy Transfer, where a person can transfer money to his friend using tokens, and the token can be cashed in. The system is on trial.

The Passage: What is your target demographic?

Ketan Patel: Salaried millennials who make around 200,000 rupees salary per annum, and are between age group of 24 to 30.

The Passage: If you are using SLQ to assess the credit score, why is a fixed salary a prerequisite for loans?

Ketan Patel: For SLQ or for that matter any algorithm to work well, you need data. Salaried was an easier segment for us because fixed income is coming on a monthly basis. So, we started with that. But soon, we are going to start with self employed and self employed professionals as well.

The Passage: How’s CASHe profiting from your AI expertise?

Ketan Patel: We are learning, because at the end of the day, my NPA has become one third of what it was one year ago. And the test of the AI is how well can it predict behavior. As long as the NPAs keep going down, we know it is working.

The Passage: With more players entering the lending space, especially from China, do you think the government will bring more regulations?

Ketan Patel: I think there will be regulations regarding data usage and regulations regarding interest rates soon.

I don't think people will be allowed to charge more than the credit card overdraft charges, which is around 3.25% per month, which roughly translates to 40% per annum.

The Passage: Are you looking to raise funds?

Ketan Patel: We are in the midst of a fund raise. We are looking to raise USD 20 million. We broke even in November last year. In the first quarter, we ended up with a profit of nine crore rupees. I think we will end the year with a 40 crore rupees profit.

The Passage: How do you plan to use the funding?

Ketan Patel: We will use the capital to get into the self employed and self employed professional space - only to lend though. We don't have any marketing expense. So whatever money we raise will go to lending.

The Passage: How do you stand out from the competition?

Ketan Patel: Firstly, we are not a bureau based lender, we don't look at scores, and we have our own algorithm to allocate loans.

Secondly, the process is so seamless that the app is very easy to operate. So we've ensured that we don't make it difficult for people to drop out.

Thirdly, from a lending point of view, we are the only profitable company.

The Passage: How do you reach out to the customers?

Ketan Patel: Completely digital. We do have channel partners, we do a lot of stuff on LinkedIn social media.

50% of the customers we acquire are organic and 50% comes through channel partners. So all loan aggregators are my partners, and they refer customers to us.

Ruiyao Luo

Ruiyao Luo is a Beijing-based tech reporter. She focuses on emerging startups and tracks the trends in the startup industry in India and China. She can be reached at

Follow Ruiyao Luo