Maximize Value Across the Entire Customer Lifecycle
Life in 3D: Using Alternative Data to Power Credit Risk Decisioning
Life in 3D: Using Alternative Data to Power Credit Risk Decisioning
As humans we see the world in 3D. It helps us judge everything we do, from reading people’s reactions so we can respond accordingly to knowing how high we should step to make sure we don’t fall. Yet as a lender you’re often forced to determine credit risk in one dimension, using a traditional credit score. And if there is no credit score, you are inclined to walk away. So, as a lender, how do you take your 1D view of a prospective customer, either a business or an individual, and see in 3D to assess credit risk?
You explore alternative data to fill in the blanks.
Today, alternative data enables you to create a more agile approach to your credit risk decisioning process. It unlocks a new world of information — location, social media presence, text message history, job histories, education, and Internet searches, all of which can play a significant role in driving an alternative data approach to credit risk decisioning. An approach that is not reliant on a traditional credit report. Think of your potential customers as a 3D image waiting to be seen. How will you see these faces and ultimately determine if you should lend to them?
Give Me Some Credit: Growing Your Business Using Alternative Data
Alternative data has the power to assist in achieving growth and expanding your business. How? It empowers you to gain a deeper understanding of potential and existing customers. Using the right types of alternative data allows for a more comprehensive and accurate credit risk picture. Think of this as building out your three- dimensional view of a person…you may currently rely on traditional methods to create a 1D view of someone, and while this can be a solid indicator of credit risk there is often much more to see, especially if no traditional credit score is available. So, how can you use alternative data to drive business growth? 1. Say yes more One area ripe to benefit from the use of alternative data is decisioning for credit invisible, thin-filed, or underbanked consumers. Often a decisioning challenge for lenders, this group includes minorities, immigrants and younger consumers seeking an opportunity to obtain credit but lacking traditional credit scores or a credit history. Alternative data becomes an entry into financial services and brings them into the lending ecosystem. Small businesses are also another customer to consider. SMEs can find accessing credit difficult – paperwork requirements and background checks can be tedious and time-consuming, meaning they don’t get access to funds
quickly, which many small businesses desperately require. Some SMEs even choose to forego the traditional credit process altogether. But alternative data can help overcome these barriers and help entrepreneurs gain access to credit by creating a more accurate, complete profile with a faster turnaround.
In a TransUnion study 1 of lenders and credit providers using alternative data, it was found that:
By using alternative data, these lenders are creating a more robust picture of creditworthiness for a broader group of consumers. In fact, 71% of respondents agreed that alternative data provides them a more complete view of consumer credit risk. 2. No second guessing Improve your accuracy and increase your confidence in your credit pricing. Alternative data can remove doubts, which previously led to disqualifying customers or potentially overpricing an offer to hedge for unknown risks. Alternative data creates new opportunities for lenders by expanding your customer base, but it’s also a chance to improve your understanding of current ones or no-file consumers 87% 64% improve risk assessment among unbanked consumers by using alternative data 67% evaluate non-prime consumers through alternative data use alternative data to evaluate thin-file
by supplementing your current view. Ingesting additional data into your model improves the detail of your 3D profile: rotate, zoom in, zoom out and view from a different perspective. Alternative data opens the door for new possibilities with personalized pricing and product offers, plus the ability to upsell when the time is right. Imagine renovating a house and relying solely on 1D drawings and blueprints – it gives you a good enough picture, but it doesn’t really allow you to imagine the finished product easily (or make changes) the way a 3D rendering can. Relying on 1D to determine a customer’s credit risk is similar – a traditional, static credit score just can’t offer the same agility that a more holistic view provides. When you make a customer feel like you’ve created products and services directly for their needs, you’ve established loyalty, something which can be hard to come by in today’s instant-everything society. The result is new lending opportunities, improved customer experiences and lower, more competitive rates.
Lenders using alternative data experienced a:
increase in offer acceptance 1 48% 38% improvement in pricing 1
But don’t stop there. The value of alternative data is limitless. Discover new market opportunities, incur lower transaction costs, and identify and prevent fraud. Worried about the ROI of implementing an alternative data strategy? Don't be: 64% of lenders experienced 1 tangible benefits within a year.
Exploring the Topography: Using Alternative Data Across the Credit Lifecycle Let’s look at how alternative data can bring value across the customer credit lifecycle: 1. Onboarding/identity verification Alternative data can be used in the onboarding process to simplify the customer experience and create a more suitable offer that is not solely dependent on a credit report. On top of that, alternative data plays a key role in fraud mitigation. The UK alone saw a 66% rise in fraud 2 during 2020. Alternative data sources, such as an email address, phone ownership, photos, and more can be used to verify identity, and in turn, prevent fraud. It is reported that online lenders can improve their fraud detection rates by up to 92% 3 at submission, allowing lenders to cut fraud losses. 2. Underwriting/decisioning Incorporating alternative data points into a credit risk model can help bolster the underwriting process and enable stronger decisioning where you aren’t limited in how you determine risk. Accuracy improves, and alternative data supports more precise pricing strategies. You’ll be able to drive portfolio improvement from two angles – reduced defaults and more competitive products. In fact, alternative data delivers superior modeling results. Credit risk models more accurately score 90% of applicants who would have returned a no hit or thin-file response using only traditional data. 4
While the approval or pricing improvement may look small in isolation, when the results are seen across the entire portfolio, the performance improvements translate into significant financial gains. 3. Upsell/cross-sell Alternative data can identify life events that trigger new financial needs, such as buying a home, becoming a parent, attending college and more. The data can also show a customer’s loan payments made to other lenders, a change in spending behavior, rising income or the timely payment of utility bills, identifying opportunities for new products and services. This information can be used to better understand customers for well-suited upsell and cross-sell offers, empowering you to offer the right products at exactly the right time, driving a personalized and valuable customer experience.
Seeing in 3D: Overcoming the Common Alternative Data Challenges
expertise 1 46%
of lenders lack internal alternative data
Almost half of lenders cite in-house knowledge as a key challenge to implementing an alternative data strategy. A solution that fast tracks your delivery is to engage with an experienced partner. This partner should offer a variety of alternative data capabilities to fill this internal knowledge gap. However, choose your partner wisely; be sure to find one with a proven track record of using alternative data in credit risk decisioning models. Working with a partner who complements your team’s strengths can significantly propel your results by filling the gaps in your 3D image. 64% of lenders not using alternative data cited developing or testing new models as a challenge 1 Effective deployment of risk models isn’t confined to alternative data. Many lenders struggle to develop, test, and deploy credit risk models. Some of the key challenges prohibiting easy model deployment include model recoding delays caused by technology that only supports certain languages, and slow data integration. A Rexer Data Science Survey found that only 13% of companies 5 “almost always” successfully deploy analytics models.
Many companies have turned to SaaS decisioning solutions to overcome these issues. For example, the right solution will support any modeling language, enabling your team to quickly deploy advanced predictive, machine- learning models. The solution should also make integrating alternative data quick and easy, so that you can test new data types in real-time. 47% of lenders find access to external data sources challenging 1 As lenders have historically relied on traditional data types, it’s not surprising that many lenders are daunted by the prospect of integrating new data sources, especially as the type of data required can vary by situation. For example, for a consumer without a credit file, using rent payments can be predictive. And for another consumer, looking at mobile wallet information can be helpful for those customers who don’t typically use a card. Studies have defined 4 the following characteristics of good alternative data sources: 1. Coverage – Consistent and broad coverage in a concentrated market enables easily achievable data collection. For example, what percentage of adults use a cell phone in a region? 2. Specificity – Detailed data for a borrower can create a comprehensive picture, such as on-time and late payments over a significant time period. 3. Accuracy + timeliness – Accurate and frequently updated data is critical. Keep it refreshed.
4. Predictive – The data should contain information that is relevant to the behavior you’re trying to predict, such as checking account data being predictive of repayment behavior. 5. Compliant – The data must comply with existing consumer credit regulations, allowing for regional differences like the Fair Credit Reporting Act, Equal Credit Opportunity Act, or the Gramm-Leach-Bliley Act How do you find and use alternative data sources? Option 1: Vendor by Vendor Work directly with a data vendor. When seeking one, it’s important to find a company that meets the criteria outlined in the five points above. A few questions can help vet potential partners: • Can you integrate the data sources within your existing system? • What is the cost of the data? • Will the alternative data set provide a good ROI? • What’s the data integration time into your system? • Which regions do you cover? • Does the data meet regulatory and compliance requirements? • How many types of data do you offer? Once you find a partner, you’ll need to complete a contract and understand the integration process. Additional questions may include what sort of technology and process changes are needed, what are the internal staffing needs and what’s the learning curve for implementation and analysis?
Option 2: The Marketplace Approach
Use a data marketplace. Through a data marketplace like Provenir’s, lenders can accelerate and improve the accuracy of risk decisions by selecting numerous data sources in the lending ecosystem to meet individual needs. When choosing a data marketplace make sure to select a data vendor that can support your alternative data strategy both now and in the future. Some questions to ask: • Does the data meet regulatory and compliance requirements? • Do you have partners across all regions? • What types of alternative data are available? • How easy is it to integrate data into my decisioning processes? • Will I need vendor support to add data sources to my data feed? • Will I need to maintain integrations? • Do you add new vendors to the marketplace? Once you choose your data marketplace solution, you’ll select your data sources and connect them into your decisioning solution. With Provenir, think of this as your customizable real-time data feed. You can pick and choose data to build out your 3D view of customers all through a single marketplace and API.
Uncovering the Bigger Picture: Exploring Potential Alternative Data Types Whether it’s age, experience, or social media usage, people come in all shapes and sizes. It’s their diversity that makes them strong candidates to be your customers. Combining different data points will vary for all customers and having a dynamic credit risk process is key. There’s plenty of alternative data to choose from to create a full 3D view of individuals and their surroundings. Multiple data types can make the risk decisioning process stronger, but it will take a different combination for different uses. Finding the right alternative data is key. By combining it and creating a credit risk model that maximizes ROI, it can target prospects, expand credit to new borrowers, create new opportunities for current customers and provide an enhanced customer experience through speed to market and less paperwork. And along the way, you’ll benefit from reduced costs, greater competition and lower fraud and portfolio risks. Here are some alternative data types to consider for your credit risk decisioning process. Think of each data type as a building block and imagine the various types of 3D images you can create as you stack those blocks – keeping in mind that data types will vary across regions and as customer circumstances (or your products) change.
Alternative Data Types:
This list of alternative data sources is not exhaustive. This
This list of alternative data sources is not exhaustive. Regardless of which data sources most appropriately align with your business model, the objective is to ingest the right alternative data to build the right credit profile by individual or business and lend based on an adjusted credit risk score without sacrificing speed or financial risk. Here are a few companies finding success with some of the alternative data mentioned above: • Australian-based fintech Jacaranda Finance focuses on short- term lending and uses bank statement data for its alternative scoring system. By working with a financial services insights’ solutions company, Credfin, Jacaranda analyzes and categorizes comprehensive bank statement data to clean, extract and group data for a loan underwriter’s review. Jacaranda found using this alternative data type enabled them to identify customers likely to repay their loans and accurately predict those likely to default. 6 • U.S. fintech Petal is a credit card company offering products for consumers new to credit and typically overlooked by the big banks. Petal analyzes a consumer’s banking history and then creates an alternative credit score called Cash Score, which measures creditworthiness based on one’s income, savings and spending history. 7 • Singapore fintech Credolab works with lenders globally and has found that when working with its clients in Africa, the best source for alternative scoring comes from mobile wallet data as very few consumers utilize either debit or credit cards. This alternative data also serves as a substitute for income, helping to compute predictive value for credit risk decisions. 8 • China’s online-only bank WeBank has a successful unsecured personal loan product. Through a relationship with the internet company, Tencent, WeBank uses their social media data, including WeChat messaging app data, within its strategy and to determine fraud risks. Data analyses are also created to assign in-house social scores to help make loan decisions. 9
Building the Bigger Picture: Put Alternative Data to Work through Actionable Steps Moving from a 1D or even 2D data view to a full three- dimensional vision of your customers can seem daunting, but it doesn’t have to be. Use the actionable steps below to get started: 1. Proof-of-Concept
Define the customers and credit decisions that you want to predict. Are you focused on unbanked customers, small businesses looking for short-term loans or young consumers with minimal credit history? Think about your goals – are you trying to predict which customers will pay their loans on time or are you trying to price loans to be more competitive? Use this information to develop a proof-of-concept project. 2. Develop Your Risk Models + Identify Data Partners Once you have your project, you’ll need to pin down your credit risk strategy and identify data partners. Depending on your business approach, you may either source the model or the data first! Whether you’re building your model internally or using a third-party, you’ll need to source the data to support it by either contracting through multiple data partners or through a data marketplace.
3. Integrate + Automate Create an automated decisioning solution to manage end-to-end straight-through processing for applications. Data from your chosen vendors or marketplace will be integrated into these workflows. With the right technology this should be a quick and easy process. You’ll deploy your risk models into these processes to power world-class user experiences and store the data you need to fully track KPIs across the decisioning flow. 4. Analyze To track the performance of your alternative data risk models you’ll need to analyze the data based on the predetermined KPIs of your POC project. These may include approval rates, average pricing, decisioning speed, credit risk accuracy, default rates and more. With business insights, machine learning models, and visualization tools directly integrated with your decisioning solution, you can track these metrics in real-time. 5. Determine Return on Investment As with any data source, there is an associated cost. To determine the success of your proof-of-concept project you should evaluate whether the increased cost of using more data results in improved portfolio performance. With data showing that most lenders see a return on the cost of alternative data within the first year, you should be confident that investing in an alternative data strategy will be well worth it!
The View from the Cloud
Discover how our Global Data Marketplace can help you find and integrate new data partners and sources. With Provenir’s Cloud Suite, we offer you cloud-native data, decisioning, analytics and insights, all in one platform. Rapidly create sophisticated decisioning workflows, integrate any data source and get real- time business insights. Supporting a wide range of AI and machine learning capabilities across the customer lifecycle, our no-code user interface allows you to make changes quickly – making it easy to launch, learn and innovate, driving the flexibility you need to power business growth.
Supercharge your data strategy
Endnotes 1 https://www.transunion.com/resources/transunion/doc/insights/research-reports/research-report-state- of-alternative-data.pdf 2 https://home.barclays/news/press-releases/2020/08/scammers-take-advantage-of-covid-19--cashing- in-on-nations--unce/ 3 https://www.prweb.com/releases/how_fintechs_can_use_alternative_data_for_improved_predictive_ modeling/prweb18046239.htm 4 https://www.oliverwyman.com/content/dam/oliver-wyman/v2/publications/2017/may/Oliver_Wyman_ Alternative_Data.pdf 5 https://www.rexeranalytics.com/data-science-survey 6 https://www.canberratimes.com.au/story/7316103/is-using-alternative-data-the-future-of- lending/?cs=14256 7 https://news.crunchbase.com/news/beyond-payday-loans-more-startups-and-vcs-bank-on-subprime- lending-alternatives/ 8 https://www.theafricareport.com/107432/will-ai-risk-analysis-really-expand-access-to-credit-in- africa/ 9 https://www.iif.com/portals/0/Files/private/finewdata_cfi.pdf
Page 1 Page 2 Page 3 Page 4 Page 5 Page 6 Page 7 Page 8 Page 9 Page 10 Page 11 Page 12 Page 13 Page 14 Page 15 Page 16 Page 17 Page 18Made with FlippingBook Digital Publishing Software