What is Alternative Credit Scoring and How Does It Differ From the Traditional

Traditional credit scoring based on credit history and ratings may not always be applicable. 

Imagine the guy in his early thirties, leading a successful creative agency. He wants to grow, so he needs a loan for new talent and tech. However, banks are hesitant because he doesn't have a long credit history or a typical job path.

This shows a problem with traditional credit scoring methods: they miss out on people like this guy. 

But there's a new solution: alternative credit scoring.

We'll explain how these new methods differ from traditional ones and how they can help you identify those valuable customers. 

Let's dive in.

‍

Why financial inclusion needs alternative data beyond the traditional

The Federal Reserve Board reports that 6% of Americans are unbanked, and another 13% are underbanked. Due to outdated credit models, banks will not be able to lend to 19% of the US population.

The issue is that the situation with unbanked people worldwide is much worse than locally in the US. 

‍Global Finance Magazine reports that the top five world's most unbanked countries are Morocco, Vietnam, Egypt, the Philippines, and Mexico:

This suggests that millions of people cannot rely on obtaining credit from a lending organization using traditional credit scoring.

Another standard requirement for borrowers is formal employment. 

However, recent data indicates that there are over 76 million freelancers in the US. Like credit-invisible people, they may not be able to obtain funds from a bank regardless of their creditworthiness.

To serve these demographics, credit institutions require credit risk digital transformation. They need to supplement their credit scoring algorithms with non-traditional data.

‍

Understanding traditional credit scoring models

These scoring models use traditional data to evaluate potential borrowers:

• Payment history. Records of payments related to financial obligations. It includes information on the date, amount, and timeliness of payments.

• Outstanding debt. The amount the borrower owes to one or more creditors.

• Length of credit history. The period during which a person has been using credit. Typically measured from the opening of the first credit account.

• Types of credit in use. Forms of credit the borrower has utilized, including installment loans, credit cards, mortgages, etc.

• New credit. Information including recent credit inquiries or new credit accounts opened by the borrower.

• Amounts owed or credit utilization. The ratio of total debt to total credit limit across all credit accounts.

Based on this data, traditional credit scoring generates a credit score representing the client's creditworthiness. 

For example, FICO scores range from 300 to 850:

The borrower's chances of getting credit and receiving favorable lending terms increase as their credit score improves.

‍

The impact of alternative credit scoring on determining creditworthiness

Traditional scoring models fail to objectively assess borrowers and significantly limit the target audience of lending organizations. 

Another problem is that traditional credit assessment does not consider the constantly changing behavior of users.

People of different generations have entirely different attitudes towards lending. For example, there are more freelancers among Gen Z and Millennials, and they are more technologically savvy.

Their lifestyle radically differs from older generations like Baby Boomers and Gen X, so traditional methods are insufficient to assess their creditworthiness. 

Learn the key findings of the U.S. Gen Z by TransUnion in the diagram below.

By incorporating non-traditional data, lenders can consider borrowers' behavioral patterns, particularly those that vary based on their age group. 

This type of credit assessment enhances the effectiveness of scoring models and aids in making well-informed decisions regarding credit applications.

‍

Alternative data sources to transform credit scoring

To expand the coverage of consumers with their services, modern credit organizations use different credit scoring models, each of which utilizes specific types of alternative data.

Digital footprints

According to Data Reportal analytics, there are more than 5.5 billion internet users in the world. This is almost 68% of the global population. And every year, this number increases by 2.5%. 

Such a global reach is advantageous for credit institutions.

The fact is that every internet user leaves a digital footprint. This is all the information that remains publicly accessible after a person interacts with various online resources.

Lenders use this information to enrich their scoring models. That is, they take it into account when assessing the creditworthiness of potential borrowers.

Here are the key types of alternative data available as a result of digital footprint analysis:

#1. Email data

This is information that can be found by knowing just the applicant's email address. It includes:

• Email inbox activity (deliverability of emails)
• Linked profiles on social networks and other resources
• The age of the email account
• Domain information
• Data breaches or presence in blacklists

#2. Phone number data

Another piece of information that a potential client is required to provide in a credit application is their mobile phone number. By knowing it, the lender can track:

• Suspicious numbers (virtual SIMs, disposable numbers, and burner phones)
• The subscriber's location based on the operator's code
• Whether the number is included in high-risk databases

#3. Location data

Hiding one's actual location is one of the likely signs of fraud. Here are the factors that will indicate this:

• Use of anonymizers (VPNs and similar services)
• Discrepancies between the data provided in the application and the results of digital footprint analysis (mobile operator code, IP geolocation, social media data)
• Inclusion of the IP address in blacklists

#4. Social media data

Social networks contain a lot of information about their users.

The mere absence of registered accounts is suspicious, as statistics show that the average user has about 7 social profiles.

Additionally, other information is valuable:

• Profile type and the content posted
• Career and education data (on professional networks like LinkedIn)
• Geolocation tags to verify the stated location

#5. E-commerce data

According to statistics, 2.71 billion people worldwide actively make online purchases. Data on activity on e-commerce platforms is a valuable source of information for creditors:

• The frequency of purchases, types of goods, and their cost indicate the consumer's material status
• Payment methods, transaction times, and cart abandonment rates can provide additional data on a person's character traits

#6. Paid subscriptions data

Having paid subscriptions is already a positive sign for the lender. However, other facts should also be considered:

• Regularity of payments
• Duration of subscriptions and changes in their status (upgrades, downgrades, cancellations)

ML algorithms

Artificial intelligence (AI), and specifically machine learning (ML), is widely used to enhance modern scoring systems.

These technologies allow for the optimization of risk assessment and forecasting, improve the decision-making process, and effectively prevent fraud.

The RiskSeal scoring system offers several methods of assessing potential borrowers based on ML algorithms:

1. Face recognition. This method allows for comparing publicly available photos of the applicant found online.

For example, if different people appear in the profile pictures on Facebook, Instagram, and LinkedIn, it may be seen as a sign of fraud.

‍2. Name match technique. The names used by the user across different online platforms are also compared. If they don't match, the borrower is assigned a high-risk level.

3. Anomalies detection. ML algorithms are capable of processing vast amounts of data, analyzing it, and identifying patterns.

This allows for constructing a complete picture of the applicant's consumer habits and detecting atypical patterns. Their presence is another sign of potential fraud.

Payment data from third-party systems

This term refers to information about any payments other than those made for loans and credit cards. These can include:

• Utility payments
• Rent payments
• Phone bills
• Internet subscriptions
• Insurance payments
• Other periodic expenses

This information can be provided by the following third-party services:

• Banks
• Payment systems
• Fintech platforms
• Utility companies
• Mobile network operators
• Internet service providers, etc.

Information showing timely and full payments and a high credit rating can confirm their creditworthiness.

Personalized models for specific population groups

There are segments of the population that traditional banks prefer not to lend to, such as freelancers, former defaulters, elderly, or very young people. 

Non-traditional credit scoring considers these factors when building scoring models, resulting in more accurate assessments.

‍

The 10 advantages of alternative credit scoring

Alternative credit scoring enhances the competitiveness of lending organizations by offering a range of benefits.

1. Expansion of customer base. Alternative credit decisioning allows extending credit to individuals without a credit history or with a low credit score.

2. Personalized borrower assessment. Lenders can incorporate various analyzed parameters beyond traditional credit history into their scoring models. For instance, the borrower's occupation, age, marital status, etc.

3. Adaptability. Machine Learning algorithms learn from new data and adapt to it. This ensures that non-traditional credit scoring models remain relevant in changing market conditions and consumer behavior.

4. Broader access to credit. Allows more borrowers, especially those without traditional credit histories or with low scores, to access loans.

5. Customized scoring models. Enables the development of unique scoring models tailored to a specific business’s needs, enhancing decision-making accuracy.

6. Personal approach to each customer. Alternative credit scoring methods provide more detailed information on borrower behavior than traditional credit assessment. This enables offering personalized banking services to customers, such as tailored loan interest rates.

7. Reduced loan origination costs. Automation and efficiency gains help lower costs associated with loan processing, potentially leading to more favorable interest rates for customers.

8. Bias and error reduction. Automated systems minimize the biases and manual errors that can occur in traditional credit scoring processes.

9. Flexibility of software. Digital products for alternative credit scoring support the customization of threshold values and age requirements, allowing compliance with lender conditions. They can also expand the pool of potential borrowers.

10. High level of data confidentiality. The use of non-traditional data implies obtaining consumer consent. Many borrowers are willing to share personal data, indicating a high level of trust in alternative scoring systems.

In addition to the obvious benefits, lending organizations that use alternative credit scoring models may face certain challenges. You can learn more about this from our guide to enhancing scorecards with alternative data.

‍

Comparing traditional and alternative data in credit scoring

You should use both traditional and alternative data to create effective scoring models. These data significantly differ but complement each other to achieve maximum efficiency.

‍

Alternative data performance at RiskSeal. Case study

RiskSeal enriches scoring models with alternative data. We analyze applicants’ digital footprints and provide over 400 digital signals.

Our latest project has shown that the effectiveness of credit scoring is maximized when both alternative and traditional data are used.

About the project. This was a collaboration between RiskSeal and the fintech company AvaFin, specializing in innovative consumer financial solutions. 

The challenge. The Mexican branch of our client identified a significant untapped market. They established that 50% of the local population had limited or no established credit history. This presented a valuable chance to develop innovative assessment methods where traditional data was unavailable, allowing AvaFin to serve this financially underserved segment.

Solution. AvaFin reached out to RiskSeal to apply alternative data sources and improve their scoring models.

Result. RiskSeal provided AvaFin with various data points based on the web presence of borrowers. Among them:

1. Digital footprints from social media and messengers, entertainment, gambling, travel, educational platforms, and custom regional platforms.

2. Email lookup signals – deliverability verification, email age, domain data analysis, data breaches, and disposable domain identification.

3. Phone number lookup signals – disposable numbers, data breaches, digital and social profile registrations, telco details.

4. Signals for identity verification – applicant’s avatars and photos for face recognition, name variants, location data, and email and phone number data links.

At the end of the collaboration, our clients noted an increase in decision-making speed, access to a wide range of data, and excellent fraud detection.

‍

Key takeaways

Alternative credit scoring is a method of assessing potential borrowers using non-traditional data sources.

Here are the key insights we discussed in this article:

• Broad data consideration. Traditional credit scoring considers only the financial information of borrowers, while alternative scoring systems use a broader range of data.

• Expanding credit access. The use of alternative data promotes financial inclusion, as enriched scoring models enable lending to unbanked individuals.

• Digital footprint analysis. Non-traditional data sources include users' digital footprints and payment data other than credit payments.

• Benefits for lenders. Alternative credit scoring has several advantages, including expanding the lender’s customer base, offering a personalized approach to each client, reducing costs, and minimizing errors.

• RiskSeal’s unique approach. The RiskSeal scoring system specializes in analyzing applicants’ digital footprints. We provide our clients with 400+ digital signals, enabling them to effectively identify fraudsters, increase approval rates, and reduce default rates.