[Product Roadmap] With easy-to-use models, how neobank Kaleidofin aims to address the needs of over 600M unbanked Indians
During their stint at Dvara, Suchitra Mukherjee and Puneet Gupta realised that none of the existing financial service products addressed the need for micro-savings or micro-insurance.
A large number of products and processes had inflexible financial solutions that were designed for middle to upper-middle-class households with relatively stable incomes or sufficient buffers to absorb cash flow mismatches. They were rather unsuitable for customers who have uncertain and fluctuating cash flow.
Thus, the duo started Kaleidofin in 2017. It is a neobank that offers simple, well designed financial solutions to address the needs of over 600 million underbanked and unbanked Indians engaged in the informal economy. Its primary focus has always been on increasing access and usage of formal financial services for informal sector customers.
With significant socioeconomic changes like JanDhan accounts, Aadhar, and increased access to mobile phones, low-income households (LIH) were better able to access finance.
“We felt, while the issue of accessibility may reduce, there was a need to design digital financial services for this segment. However, we were aware that financial literacy was limited in this sector, and we could not have a direct ‘product’ approach. So, we adopted a financial goal approach, where we started conversing with customers about their goals such as education of a child, building a house etc.,” Puneet tells YourStory.
He says Kaleidofin aimed to make the process of selling financial products such that every customer interaction helps them in meeting their goals such as sending their children to school, making improvements to their homes, or investing in their enterprises.
Beginning with the goals
“Towards this end, we have three key offerings – Kaleidofin Goals, KiScore, and KaleidoPay. Kaleidofin Goals is a savings- and investment-led solution that helps individuals achieve their goals through their own savings, supported by risk protection through thoughtfully linked insurance policies; KiScore and KaleidoPay are technology products that ease the process of credit issuance and payment collection for financial firms such MFIs and NBFCs,” says Puneet.
The team introduced Kaleidofin Goals, their first flagship product, in 2017. They realised that while most households have a bank account and typically have many goals such as saving for a child’s education, house improvement etc., they resort to informal financial products such as saving in gold or unregistered chit funds. This is largely due to a lack of access to formal sources of finance or a lack of understanding of products such as pensions, mutual funds, etc.
“This cognitive pressure of planning is on our backend algorithms and machine learning tools. Based on primary and secondary data collection, we create individual customer personas to determine the right financial solution,” says Puneet.
For distribution, Kaleidofin has a B2B2C model with partnership networks consisting of MFIs, SFBs, CBs, NGOs, and other FIs.
“We partnered with network partners such as MFIs, co-operative banks, and corporates to provide a last-mile touch point to customers and reduce any tech friction. In May 2021, we had 110,000 customers and had tie-ups with 40 entities,” says Puneet.
“The trust network helps us leverage the existing trust and behaviour, whereas their agents help us iron out any technology barrier we may face with the end customer. As per our understanding of the market, there is no other firm with a similar offering for this sector,” he adds.
While the core principles of the products remain the same – designing digital financial services for the micro saving and micro-insurance sector – they have added new features based on user feedback and feedback from sales and field agents.
"We improved upon the existing AI mechanism to check for blurriness, format, etc. We today have less than one percent rejection rate,” says Puneet.
The scoring model
In December 2019, Kaleidofin introduced a credit scoring model called KiScoreTM, a machine learning AI-based algorithm that helps underwrite customers for loans. The model is designed to gauge the total ability of households and enterprises to borrow and helps in understanding risks associated with lending.
The model also helps assess risk-based pricing using variables such as demographics, financial transactions, assets, and other behavioural information to be able to create a strong lending algorithm. KiScoreTM has been used to score over 750,000+ customers and underwrite over Rs 3,000 crore.
KiScoreTM is offered as a B2B product to other NBFCs. Today, over six NBFCs are using the solutions to underwrite customers. The first set of loans were given out in February 2020, and after initial testing, the model started to scale post the lockdown.
“Immediately after the lockdowns in April last year, a lot of informal sector customers survived by meeting their household expenses by using up their working capital. Several such customers needed working capital urgently for opening up. KiScoreTM was able to help customers raise more flexible loans to help kickstart their enterprises again,” explains Puneet.
The team unlocked a key growth hack in recognising that there are no solutions to measure risk in a data-driven, scalable fashion for the targeted customer segment as often there is no FICO or credit score equivalent.
Kaleidofin has invested deeply in data ingestion, collection, and predictive models that are customised to how the customer segment behaves – allowing them to have deeper insights and enabling customers to better access credit.
Enabling digital payments
The third product, KaleidoPay, enables digital payments such as EMIs and insurance premiums for Grameen banks, cooperative banks, and other MFIs. It is a unique offering that allows customers who are unable to use UPI-based payments to pay digitally. Launched in June 2020 in a limited mobility environment during the pandemic, it allowed customers to pay their existing loans to various lenders digitally.
This way, Kaleidofin seeks to make the payments environment more inclusive for customers who don’t have access to tech-based payments.
“While financial products geared towards long-term saving, the importance of credit cannot be understated or ignored. With limited availability of credit products that are flexible and are suitable for LIH, customers borrow from informal means such as chit funds or moneylenders,” says Puneet.
He adds the team has designed systems in a manner that they can be horizontally scaled with ease to involve millions of users. Kaleidofin is also investing in backend technologies that allow us to decouple microservices for independent deployment cycles and efficiently process parallel requests asynchronously.
Building customer-centric systems
“Our primary focus has been to build customer-centric systems and processes. Our backend investments are geared towards that. For instance, we noticed that customers from rural/semi-urban sectors respond better and faster to voice-led systems. My team is working on product design to use voice and IVR to lead a large part of instructions on the app,” says Puneet.
Puneet adds they have invested deeply in frontend experiences to aid agents and customers located in areas with low internet connectivity and using lower-end smartphones. The focus is on ease of adoption as the consumer’s focus should only be on achieving their financial goals.
“We ensure investments in strong AI/computer vision techniques even on the frontend apps so that users are provided real-time feedback when they upload the wrong documents and do not face rejections down the road. We have invested in simplified user experiences, allowing users the choice of using WhatsApp vs app-based flows vs voice for various flows,” explains Puneet.
The team realised that this customer segment was also more likely to sign up if they don’t have to take time off work and travel to distant locations to handle their financial needs.
The team used these insights to focus on reducing turnaround times at every step:
1. Kaleidofin invested in a strong Touch-to-Tech model. Initial face-to-face interaction through partner agents allowed us to scale up fast without investing in deep agent networks while also providing a win-win for partners. The team also made deep investments in AI/ML/Computer Vision-based tech to ensure that KYC and payment documents collected were accurate and field agents did not need to commute to have the documents approved from the local branch.
2. Leveraging rich data sets of customer history on loans and questionnaires customised for different geographies/demographics across India allowed us to make personalised product recommendations and credit underwriting recommendations – providing customers access to loans they would otherwise find hard to access and ideal savings solutions for them
3. They invested in consistent ongoing customer service to build brand recall and predicted usage patterns to build multi-channel customer connects.
“Building our backend to decouple microservices for different and overlapping use cases has allowed us to reuse these components across different product lines, thereby best using our engineering team as well as efficiently process parallel requests asynchronously. We have invested in highly scalable queueing systems that allow us to build retry flows within our own services and to build a resilient backend that interacts with numerous third-party systems to support myriad flows such as KYC, payments, mutual funds, loans, insurance, credit scoring, etc,” explains Puneet.
The team actively uses AI, OCR, and computer vision techniques to help identify anomalies in various onboarding use cases such as KYC and payment setup. The team has also made significant strides towards robotic process automation.
“We are continuously investing further in deep-tech that allows us to automate all processes fully and achieve complete straight-through-processing. Developers use Python, Keras, and TensorFlow for the development and testing of models, especially deep learning,” says Puneet.
The team aims to continue to innovate on how to service this customer at their doorstep – physically as well as using other channels of customer engagement. They are looking at AI-based technologies that can help with customer engagement – improved communication using NLP techniques (voice, IVR, video, chatbots) and near real-time issue resolution.
“We also continue to evolve our high-tech high-touch model by building assistance networks from within our own customer base – leading to expanded reach, better service and organic referral-based growth. We are currently in the consolidation phase as far as Kaleidofin Goal, KaleidoPay and KiScore are concerned,” adds Puneet.
The startup is also focusing on achieving these in vernacular languages.