Is deciding which customers are qualified for financing without having a human being review the file a trustworthy long-term strategy for bankers to follow?
Software-driven, automated lending decisions might give the Risk and Underwriting teams a feeling of unease. However, if lenders can use auto-decisioning to augment human-based underwriting when delivering lending decisions to small business clients, and make better decisions even faster as a result, is it not worth a look?
As a way to optimize, digitize, and scale a lending operation, financial institutions should go through the process to figure out how auto-decisioning can work for their specific situation. In this article, we’ll explore how bankers can do this without opening themselves up to undue risk, especially at a time of considerable volatility.
What exactly is auto-decisioning?
Auto-decisioning software can make lending decisions on loan applications on its own. Your unique credit policies serve as inputs for the algorithm used by the software to make lending decisions. Auto-decisioning further enhances the loan origination process by tapping into an ecosystem of technology. For example, the ability to integrate with 3rd-party technology and data sources (e.g. pull in real-time information from the credit bureau’s like Dun & Bradstreet, Equifax, and Experian to help provide credit scoring data) and advanced analytics which provides insights into what input conditions might need to be adjusted. You can choose how much automation your bank wants to use for lending decisions.
Auto-decisioning software can provide real-time lending decisions to a borrower for a range of products including term loans, Small Business Administration loans, lines of credit, business credit cards, business checking accounts, trade finance, commercial real estate loans, and more.
How auto-decisioning can impact your lending operations
Auto-decisioning software, and the supporting technology infrastructure, for business loans can augment your existing human staff and take your lending operations to the next level.
Consider the following hypothetical scenario: after reviewing a cost-benefit analysis, a bank decides to use auto-decisioning to automate small underwriting decisions between $5,000 – $50,000. Some of the advantages they can expect would be the following:
- Increased Application Volume: customers can fill out an application, and submit required forms and documentation, using a self-service, digital portal. The software gives the bank the ability to track pending items and communicate directly to applicants from the platform. This can increase the volume of applications the bank can process and the bank can redeploy employee resources to other areas. For example, figuring out the next segment of the lending operation that can benefit from auto-decisioning.
- Increased Loan Closure: The bank can teach the auto-decisioning software to make underwriting decisions based on its exact credit and risk requirements. This can increase loan underwriting speed, loan closure speed, and increase the total number of loan closures which helps the bank scale their lending operation with preferred risk.
- Increased Lending Upsell: Based on the credit risk of the applicant, the bank can automatically serve up related products that fit the bank's risk profile.
- Increased Risk Analysis: Since the system is storing data from a higher number of applicants and making more credit decisions, built-in analytics systems can aggregate this data and provide management with deeper insights. These insights can improve the inputs that the auto-decisioning system is using and provide data to decide which other segments of the lending operation can be automated.
Auto-decisioning can help your entire lending operation scale with preferred risk and strengthen your loan portfolio. The sky's the limit because auto-decisioning software can be used on products such as term loans, Small Business Administration loans, lines of credit, business credit cards, business checking accounts, trade finance, commercial real estate loans, and more.
There are downstream benefits to implementing auto-decisioning:
- Time savings is a key benefit. Bank staff members can be redeployed for higher-level work now that auto-decisioning is supporting lending operations because less manual intervention is required.
- By removing the need for human intervention in some or all of the lending decision process, you can lower underwriting costs, shorten the underwriting process, and lower the risk of human error.
- Auto-decisioning allows the bank to maintain control as it is not an all-or-nothing proposition. Meaning, you can still have manual approval on decisions. You don’t have to go from a fully-manual approval process to a fully automated approval process. You can gradually move towards a fully-automated approach, have a blended approach, or any other variation you can think of. It can flex to the specific needs of your loan processing.
- Integrate an adjusted loan pricing strategy when the macroeconomic environment changes; such as in times of high inflation or rising interest rates. For additional information, please review our article titled Business Loan Pricing Strategies for Banks When Rates are Rising.
To learn how auto-decisioning has streamlined the digital lending process and improved customer satisfaction at several real-world financial institutions, take a look at some of the case studies from Biz2X clients.
A common concern with going from manual lending decisions to automated lending decisions made by a machine is how to maintain control and not expose bankers to undue risk. There are a variety of ways to test auto-decisioning in your environment while maintaining control and mitigating risk, but we offer three approaches as a testing framework.
Strategies to test auto-decisioning capabilities
So far we’ve covered what auto-decisioning is and how it can impact your lending operation. Now, let’s review three high-level tactics to determine which form of auto-decisioning might work best at your financial institution. You can also view these three tactics as an incremental, multi-step approach that allows you to take advantage of the benefits of automation in the lending process over time (i.e., implement part one, optimize, implement part two, optimize, etc), without risking too much and maintaining full control.
Before jumping into a test, your team must decide what to test. There are a few ways to think about this because you likely want to test something that can have a high impact and demonstrate to leadership the power of automated loan decisioning. Here are some questions you can use to help you decide what to test:
- What is the most manually-intensive underwriting process?
- What is the most error-prone underwriting process?
- Which credit and business banking products take the most time to review and approve an application?
- What credit and business banking products do you wish you could sell more of if you only had more resources?
- What credit and business banking products do you want to expand into but currently don’t have the resources for?
All of these are interesting places to start your test because they could yield high-impact results. Once you know what to test, consider using the below three-level testing framework to assist:
Level 1: Decisions generated by machines are used as an input by human underwriting
Level 1 testing allows the bank to turn on auto-decisioning with a segment of its lending products as a starting point. Final credit decisions are still made manually by humans which can lower risk during the initial evaluation of the software. To expand on the example above, use auto-decisioning to automate small underwriting decisions and see how the software performs. This is key because it allows the Risk and Underwriting teams to:
- Get comfortable with the functionality of the software, how to enter and adjust inputs, integrate 3rd-party data, etc.
- Get comfortable with automation in the lending process in a semi-live environment.
- Gather data for internal reports likely needed to help get final approvals from the decision-making committee which is reviewing various other lending solutions. For more information on selecting the right solution, review our article titled Six Things to Do Before Selecting a Business Banking Platform.
- Incorporate experimental techniques based on machine learning that equip your underwriters with insights and recommendations on each credit file.
- Use the results from auto-decisioning to make a loan decision. In other words, step through the motions because going from manual to automated (or semi-automated) will result in new workflows for banking staff and will require training.
Level 1 is essentially laying the groundwork for the next two levels. This level also serves as a proof of concept for internal stakeholders to give the green light to move further into automation. Some institutions may stay at Level 1 for years, ensuring that there is a high degree of comfort in the decisions being generated by the machine before proceeding to Level 2 or Level 3.
Level 2: Decisions made by machines are checked by humans before implementation
As a bank progresses with their use of auto-decisioning technology, the data gathered from Level 1 has informed and prepared Underwriting for Level 2. At Level 2, auto-decisioning is making approval decisions in a live environment with specific guardrails in place, and humans are checking the results after the fact. For example, a term loan has been reviewed and approved by the software, the Underwriting team receives a notification, the Underwriting team then reviews the file to confirm the correct decision was made, and the Underwriting team provides feedback to management or via the software (which improves future decisions by the machine).
At Level 2, the lending operations team is still in the trenches with the software on each approval (albeit, hopefully just confirming the results). This approach allows the bank to de-risk and maintain control while turning on more automation at the same time.
There is no universal rule for levels 1 and 2 to determine how long you should be at each level. The timing of each of the first two levels will come down to your financial institution and how long it takes to gather the necessary data to decide whether to move on or not. Discuss with your team ahead of time how much data they would like to see and what performance history must look like in order to advance to higher levels of automation.
Level 3: Decisions are made by machines and audited by humans to improve results and tweak input conditions
At Level 3, data and metrics are reviewed in aggregate over a period of time to improve results by adjusting inputs into the software. For example, instead of Underwriting reviewing every individual approval made by the software, they are now reviewing hundreds or thousands of decisions made over the last month or quarter. The team will see these results rolled up into a single report and can make holistic decisions about what needs to change from there.
When your business is risk management, a crawl-walk-run approach can be highly effective when considering automated loan technology. It allows you to scale and digitize your lending operations using the latest fintech while maintaining full control of the process. This type of digital transformation can even enable your bank or credit union to open up additional product lines without needing to make a large investment in additional staff.
To Auto-Decision or Not to Auto-Decision? You Decide.
Auto-decisioning technology for business loans and credit products can help bankers make better-informed underwriting decisions by reducing risk and approving more qualified borrowers. The technology also enables financial services companies to scale their lending operation by being able to handle an increase in loan applications, decrease processing time, increase the total number of loan closures, and increase upsell opportunities. Technologies like Biz2X’s advanced credit & risk analytics tools all you to choose how much automation your bank wants to use for lending decisions. keeping bankers firmly in control of the process.
To see Biz2X in action, connect with a funding platform specialist for a customized demo of Biz2X for your bank or financial institution.