How to Spot High-Risk Borrowers Before They Default

Online lenders often experience high default rates due to several factors. These include a high number of applications, limited personal interaction with borrowers, and increased risk of fraud.

How can this situation be improved, and the quality of the loan portfolio enhanced? Accurate identification of high-risk customers in digital lending is the first step toward success.

This article will explore the types of applicants who fall into this category and key strategies for identifying high-risk borrowers.

‍

“Red flags” for lenders - who is considered a high-risk borrower?

Identifying potential defaulters using traditional credit scoring can be challenging. 

This method relies solely on borrowers' credit history and ratings, making it limited and providing only a one-sided assessment of applicants.

Overcoming these digital lending challenges is possible by utilizing alternative data in credit scoring. 

‍

Five clear signs of high-risk borrowers

#1. Lack of financial history

A person’s financial history is a traditional source of data for credit scoring. This includes bank accounts, credit accounts, transaction history, overdue payments, and more.

Why might the absence of such data be suspicious?

According to recent data from the World Bank, 76% of the working-age population have accounts in banks or alternative financial institutions.

Another source, Our World in Data, claims that in some countries, this figure reaches 99% or more.

A complete lack of financial history may signal a high default risk to lenders.

#2. Previous instances of fraud

Some applicants may already have been flagged for attempted fraud. Naturally, granting loans to such individuals is strictly off-limits for lenders.

The most common fraudulent schemes include the use of synthetic identities and identity theft.

The first involves combining real and fabricated data to create a fictitious person for committing illegal acts on their behalf.

According to TransUnion, in 2024, banks incurred losses amounting to $3.2 billion due to synthetic identity fraud. The number of accounts opened under fictitious identities also reached a record high.

Identity theft, along with its subsequent use for illegal purposes, is no less widespread.

The findings of a Regula Forensics study revealed that more than half of all businesses affected by identity fraud incur losses exceeding $300,000.

Specifically in the banking sector, the situation is no better – 49% of respondents report losses of over $300,000.

#3. Suspicious transactional data and behavioral patterns

Credit scoring with alternative data, and particularly digital footprint analysis for lenders, provides information about users' online behavior and the transactions they have made.

Such data can be useful as it helps form a psychological profile of the applicant and reveals their consumer habits.

Let’s look at a few examples that may cause a lender to doubt the reliability of an applicant:

• Unusual transactions. These could include expensive purchases on online marketplaces, large money transfers, opening numerous accounts, and more.

• Changes in income and expenses. Sudden deposits from unusual sources or large sums being cashed out can also be a red flag.

• Actions indicating a tendency toward default. For example, buying things at night could indicate impulsiveness, while playing in online casinos may suggest a tendency to take risks.

#4. Living in high-risk regions

Some regions are considered high-risk, leading to extra checks for applicants from those areas.This is explained by different economic indicators in various regions, even within the same country.For example, the credit default rate varies significantly across different states in the U.S. While in Iowa, it barely reaches 13%, in Mississippi, it exceeds 39%.

Economic stability in the region, as well as the crime rate and unemployment level, also play a role.

#5.  Inconsistencies in the provided data

Credit scoring models with alternative data are capable of detecting discrepancies between the information provided in the credit application and reality.

What might this involve?

‍Application data. This includes full name, email address, phone number, information about employment, income level, education, etc. This data should match the details provided by the applicant.

‍Borrower's photos. Innovative scoring systems analyze users' avatars on social media, comparing them with each other and with the provided selfie. If the photos feature different people, it will negatively affect the digital credit score.

Applicant's location. The actual geolocation of a client can be determined in several ways, such as by the IP address of their device. If the application provides different information, it is grounds for further verification due to high credit risk.

‍

Negative consequences of lending to high-risk borrowers

Ineffective credit risk assessment and granting a loan to a borrower with a high-risk level can lead to the following consequences for the lender:

‍1. Increased costs. A high default rate negatively impacts the profit margin of the lending institution.Another important business metric impacted is the acquisition cost (AC). In other words, the lender wastes money on attracting clients and processing applications from potential borrowers who turn out to be fraudsters or defaulters.

‍2. Damage to company reputation. Lending to fraudsters or unreliable clients can harm the lender’s reputation, which may reduce the number of trustworthy borrowers.

‍3. Legal issues. Credit organizations are required to comply with AML/CFT regulations. If it is proven that these requirements are not met, the company may face fines or even the revocation of its license.

‍4. Loss of time and resources. If the lender spends too much time and resources on handling fraudulent applications and dealing with defaulters, it will lead to operational inefficiency.

‍

Best ways to identify high-risk customers

To timely identify high-risk borrowers, online lenders are actively adopting an innovative approach to credit risk management. 

Here are the main strategies you can implement.

Using alternative data

Numerous studies have proven the effectiveness of scoring models based on a combination of traditional and alternative data.

This is illustrated by the RiskSeal’s chart below:

Alternative data includes users’ activity on social media and other websites, purchases on e-commerce platforms, information about paid subscriptions, utility and rent payments, and more.

For example, RiskSeal provides its clients with the following data set:

• Email and phone lookup data
• Social media accounts
• Online platform registrations, including local ones
• Location insights
• Online behavior metrics
• Digital credit score in real-time
• Photos and user names comparison

Before enriching your scoring model with alternative data, make sure the provided data set suits your needs. To do this, some providers offer the opportunity to test their products.

E.g., RiskSeal offers its clients a free proof of concept (PoC), during which you can assess the effectiveness of our services.

Want to choose a reliable alternative data provider that suits your needs? Check out the list of top industry representatives: Top 15 Alternative Data Providers That Serve the Credit Industry. 

Digital footprint analysis

This method involves studying the online presence traces of applicants. These include activity on various web resources, including social media.

The scoring system RiskSeal specializes in digital footprint analysis. We return hundreds of data points about potential borrowers to our clients by examining the information they leave behind while using the internet.

Behavioral analysis

In the context of credit scoring, this refers to studying the applicant's consumer and financial habits. Anomalies found in these habits can indicate the likelihood of default or fraud.

AI tools for risk management take into account the following indicators:

• Borrower's geolocation
• Device used to log into the account
• Time the loan application was submitted
• Consumer's financial habits
• Internet connection characteristics

Any inconsistencies detected can serve as a “red flag” for the lender.

Identity verification

AI-based identity verification solutions deliver excellent results.

For example, the RiskSeal ensures a highly accurate verification of an applicant's identity. This is achieved through the advanced technologies it uses:

• Face recognition – comparing photos of the applicant posted across different online accounts and, if required by the loan terms, with the applicant's selfie.
• Name match – verifying the consistency of names across all profiles owned by the applicant.
• Location match – determining the applicant's actual location and comparing the found information with what was provided in the application.

If the lender detects any discrepancies, the applicant will be deemed high-risk.

Digital Credit Scoring

‍Digital credit scoring is an excellent alternative to traditional methods of assessing an applicant's creditworthiness. 

When creating such a credit score, data from digital footprint analysis is taken into account. As a result, the obtained scores are more accurate and up-to-date.

RiskSeal’s clients receive ready digital credit scores for applicants. This simplifies the process of evaluating their reliability.

Geolocation and device monitoring

This method allows for the detection of signs of potential default or fraud:

• The borrower’s residence in regions pre-identified as high-risk.
• Application submission from devices previously used for criminal purposes.

Ongoing monitoring

This method relies on continuous monitoring of borrower actions after their application to the lending institution. To assess existing risks, the following helps:

• Tracking suspicious transactions.
• Automatic alerts for abnormal consumer behavior.

‍

Case study by RiskSeal

RiskSeal has extensive experience in identifying high-risk customers in digital lending.

We will share our insights through a case study of a large lending organization in Mexico (name withheld due to an NDA).

Client challenge

The client's issue was the difficulty in identifying borrowers who were unable to repay their loans, which resulted in a high default rate.

Solution

We analyzed email addresses, phone numbers, and geolocation data of the applicants, providing the lender with hundreds of data points on potential borrowers.

After enriching their credit scoring models with our alternative data, they reviewed 750,000 loan applications. 

Key results

Here are the results the company achieved:

• 17% reduction in default rates
• Approval rates increased by 2x
• 26% reduction in KYC spending

At this stage, productive collaboration with this client continues.

‍

Key takeaways

To summarize, here are the key points to focus on when discussing identifying high-risk borrowers:

1. There are several indicators that directly or indirectly point to a high-risk level. These include the lack of financial information, abnormal consumer behavior, discrepancies in data, and more.

2. Lending to high-risk borrowers can lead to negative consequences for the lender. These go far beyond financial losses.

3. You can implement several strategies for identifying high-risk borrowers. One of the most effective is digital footprint analysis.

RiskSeal helps its clients identify high-risk applicants. We enrich their credit scoring models with alternative data that optimizes the process of assessing the creditworthiness of potential borrowers.