Data-Driven Lending: How Lumos is Changing the SBA Game

For years, SBA lending has been built on a mix of experience, intuition, and past performance. Credit officers make calls based on their historical insights, lending teams lean on market familiarity, and sometimes, decisions come down to “how things have always been done.”

But in 2025, that’s not enough.

With rising competition, increased regulatory changes, and an evolving risk landscape, lenders need more than gut feelings. They need data.

That’s exactly what we explored in our Wake Up and Smell the Data webinar with Brett Caines, CEO of Lumos and former CFO of Live Oak Bank. Brett is leading the charge in bringing data-driven decision-making into SBA lending, and the insights he shared were nothing short of eye-opening.

The SBA Lending Landscape is Shifting—Are You Keeping Up?

Let’s talk about the elephant in the room: SBA default rates are climbing.

According to Lumos data, default rates have been steadily increasing since 2023. Some lenders were prepared for it, but others? Not so much.

So what’s driving this? A few key factors:

📌 Rising Interest Rates: More than 80% of SBA loans are variable rate. With prime increasing 525 basis points in 18 months, debt service coverage ratios have been squeezed.

📌 Post-Pandemic Market Correction: Stimulus efforts in 2020 and 2021 kept many struggling businesses afloat. Now, as liquidity dries up, some of those businesses are defaulting in what Brett calls a “catch-up effect.”

📌 Non-Bank Lenders Entering the Space: With more non-bank lenders getting licensed, their impact on SBA default rates is becoming clearer—and in many cases, concerning.

📌 Old-School Decision-Making: Some lenders are still operating on outdated rules, avoiding entire industries because of one bad experience years ago. But the data shows that some of these industries have bounced back. And, the opposite is true. Some of those great industries or franchises from years ago may now be substantially weaker.

The takeaway? The market is evolving, and the lenders who can adapt, leverage data, and make informed decisions will be the ones who thrive.

The Problem with “That’s How We’ve Always Done It”

One of the most interesting discussions we had was about lenders who avoid entire industries based on past experiences.

For example, some institutions refuse to finance certain types of small businesses because they had a bad batch of loans seven years ago. But when Lumos steps in and presents fresh data, they often realize the industry has stabilized—or even become an opportunity.

This highlights a fundamental issue in SBA lending: many credit decisions are made based on old information.

Imagine trying to navigate using a five-year-old GPS. You’d miss every road closure, every new development, and all the shortcuts. That’s essentially what lenders are doing when they don’t integrate up-to-date data into their decision-making.

The Rise of Non-Bank Lenders—And Their Default Rates

Another major talking point? The role of non-bank lenders in the SBA space.

Non-bank lenders have been growing fast. The SBA has expanded licenses for them, but the data shows a concerning trend:

🚨 Default rates among non-bank lenders are significantly higher than banks and credit unions.

While non-bank lenders bring flexibility and speed to the market, the numbers suggest that they may also be taking on riskier loan profiles than traditional banks. As more non-bank lenders enter the SBA space, this is a trend to watch closely.

The Future of SBA Lending: Predictive Analytics & Automated Decisioning

One of the biggest questions we tackled in the webinar was:

"How do we move toward faster, more efficient decision-making without sacrificing risk management?"

Brett shared insights into predictive analytics and automated underwriting as key tools for the future.

✅ Predictive Default Modeling – Lumos can assess risk based on billions of data points, rather than just relying on a lender’s past experience with a few hundred loans.

✅ Automated Decisioning for Small Loans – While no lender has fully implemented a “score-and-go” model for SBA loans up to $500K, Lumos is seeing institutions start to use risk thresholds to streamline approvals or route applications appropriately - conventional, SBA, or refer to another lender (e.g. CDFI).

✅ Portfolio Risk Monitoring – Instead of waiting for defaults to happen, lenders can now proactively spot risk and adjust their strategies in real-time.

What Lenders Need to Do Next

The takeaway from this discussion was clear: data-driven decision-making isn’t just the future of SBA lending—it’s the present.

Lenders who want to stay ahead should be asking:

➡️ Are we using real-time data to make lending decisions?
➡️ Are we leveraging predictive analytics to anticipate risk?
➡️ Are we adjusting our credit box based on fresh market insights?

If your institution isn’t already integrating data analytics, automated decisioning, and real-time risk monitoring into your SBA strategy, it’s time to start.

Because in a market that’s evolving this fast, sticking to the old ways of doing things isn’t just outdated—it’s risky.

🔹 Want to learn more about how data can transform your lending strategy? Check out LumosData.com.