Patricio Pazmiño,
Director of Analytics and AI,
Kin Analytics

The typical underwriting process is far from perfect. It is characterized by manual work and subjectivity, and the process is very prone to human error. To improve this, many lenders have adopted credit scoring models, which is a great start. Nevertheless, there is no reason to stop there as many other steps in the underwriting process, both pre- and post-scoring, can be enhanced and streamlined
with the help of artificial intelligence and data.

This article will cover a “common scenario” experienced in each step of the underwriting process and what each step could look like with technology solutions.

CREDIT APPLICATION SUBMISSION:
• Common Scenario: Credit applications are usually submitted in the form of hard copy documentation, such as PDFs, scans or even handwritten forms. Analysts must manually identify and extract relevant information and enter it into the loan origination system (LOS). This tedious process is prone to errors and demands too much time.

• Ideal Scenario: Leveraging AI, such as computer vision (i.e. optical character recognition), this process can be fully automated. Key fields from hard-copy documents can be identified with precision and automatically uploaded into the LOS. That means new records can be created in minutes, not hours, with barely any human intervention.

KNOW YOUR BUSINESS AND KNOW YOUR CUSTOMER
• Common Scenario: Typically, a sales rep, credit analyst or submission analyst conducts a manual background check on a company applying for credit. They go through public sources like the Secretary of State’s website or perform Office of Foreign Assets Control (OFAC) checks, manually verifying the company’s legitimacy. Given the time-consuming nature of this process, this step can be skipped altogether, leaving the door wide open for fraud.

• Ideal Scenario: Lenders could tap into the many know your business and fraud-check products available in the market or build their own connections to relevant sources. Automatically pulling relevant data (through APIs) from places like the Secretary of State or OFAC lists ensures every new application is validated. If something doesn’t look right, the system can immediately alert the analyst, preventing bad applications from moving forward.

CREDIT SCORING
• Common Scenario: Credit bureau reports are pulled (sometimes manually), and credit analysts spend hours scanning pages of data to identify potential risks. The probability is high that credit analysts may overlook important information or interpret information differently from other analysts, leading to inconsistencies and subjective decisions. Two applications with a similar level of risk might get different outcomes depending on who is doing the analysis.

• Ideal Scenario: Predictive analytics streamlines the process. Credit scoring models, both traditional or machine learning (ML)-based, can objectively and consistently assess every incoming application based on data-driven and objective factors. This cuts down on subjectivity, reduces the chance of human error and saves time. Analysts can focus on more complex cases in which models alone aren’t enough. Also, these models provide valuable risk metrics — such as probability of default and expected losses — which enables other applications, like risk-based pricing.

DEAL STRUCTURING
• Common Scenario: Once an application is approved, the deal must be structured, including setting the price. Most lenders rely on basic pricing matrices that don’t always reflect the customer’s risk level. They might base pricing on factors like loan term or down payment, but rarely on the actual financial structure or risk involved. This leads to suboptimal pricing decisions, which can hurt a company’s competitive advantage.

• Ideal Scenario: With data analytics, lenders can implement risk-based pricing models. These models use the outputs of the credit scoring model and the lender’s cost structure to recommend optimal pricing. This not only ensures profitability but also helps the lender remain competitive by offering fair prices based on actual risk analysis.

COMMUNICATION WITH CUSTOMERS
• Common Scenario: After a deal is approved, analysts manually send out approval or rejection letters. In many cases, it takes days (at best, several hours) for the customer to hear back. In today’s highly competitive market, those delays could cost lenders the deal.

• Ideal Scenario: If all the steps above have been automated, decision-making times can drop from days to minutes or even seconds, if auto-decisioning is used. With faster response times, lenders can stay competitive while improving customer experience. Even for the more complicated deals that still require human intervention, the gains in efficiency, accuracy and decision-making quality will be undeniable.

CHALLENGES IN ACHIEVING THE IDEAL SCENARIO
Getting to the “ideal scenario” we have described above isn’t easy. It requires time, resources and a deep commitment to transforming your company from the ground up. But, here’s the thing: no matter where you currently stand in terms of technology maturity, there will always be actions (even small ones) you can take to start moving towards this transformation. For example:

• Treat Data Like Gold. Invest the right resources in improving your data collection and storage capabilities. Every customer interaction, document or external source contains valuable data points. Even if they don’t seem important now, they could be critical when you’re ready to implement predictive analytics. Make an effort to store every piece of information you can.

• Move to the Cloud. Cloud services are at the core of technological advancement. The sooner you move to cloud-based solutions, the better prepared you’ll be for adopting data-driven applications. It’s not just about storing data; it’s about being nimble and scalable for whatever comes next.

• Bring in Expertise. Whether you decide to develop capabilities in house, bring in external help or a mix of both, having expert guidance is key in making sure you take the right steps.

• Listen to Your Internal Customers. Your employees are on the front lines, dealing with the daily challenges created by your current systems and processes. Get obsessed with gathering their feedback. They can point out immediate pain points and help you focus your tech efforts where they’re needed most.

• Stay Engaged with the Market. Talk to other lenders, see what they’re doing and learn from their experiences. Our industry is collaborative — take advantage of that. You can save time and avoid pitfalls by leaning on the knowledge of others.

In today’s competitive equipment leasing industry, sticking with outdated underwriting processes just won’t cut it. AI and advanced analytics offer powerful tools to streamline workflows, reduce errors and make smarter, faster decisions. While the journey to the “ideal scenario” won’t happen overnight, even small steps can lead to significant improvements. The future of underwriting is already here, and those who embrace it will gain a critical edge in efficiency, accuracy and customer satisfaction. •

Patricio Pazmiño is director of analytics and AI at Kin Analytics.