The Beanie Babies That Changed Business Lending

Introduction

Instead of following traditional rules and models that made it impossible for banks to provide loans for a huge segment, Kathryn Petralia has decided to do something different. When founding Kabbage in 2010, they were perceived as pioneers who started to use data, automation, and advanced technology to do what banks couldn’t—assess risk in real-time and fund the businesses no one else would. So how come 15 years passed since then, technology has never been more advanced, data has never been more accessible, and yet – small businesses still struggle to get the capital they need from their bank?

“If the people executing the strategy don’t benefit from it, they’re less likely to embrace change. Saying no is safe, but saying yes carries risk—and risk-averse cultures make innovation difficult.”

‍

Interview

Omri: You've had a front-row seat to the evolution of small business lending. What inspired you to start Kabbage, and how did you identify the gaps in the industry?

Kathryn Petralia: It’s an unusual story, especially in fintech. It all began with an API that eBay launched in 2007, which provided third parties with real-time access to seller, transaction, and payment data.
That API sparked an idea for my co-founder, Rob. He wondered: Could this data be used to make lending decisions for eBay sellers? That was the key breakthrough—leveraging technology to automate lending decisions in real time. So it wasn’t the typical story of a struggle to get a small business loan for ourselves.
But beyond the technology, we were serving a segment that traditional lenders didn’t even recognize as businesses. Imagine walking into Bank of America in 2009, sitting across from a loan officer, and saying, I sell Beanie Babies on eBay. They’d probably say, Thanks, but no thanks.
Without realizing it at the time, we built a solution tailored for the smallest businesses, a market that banks largely ignored. And because we automated nearly every step—from application to funding—we were able to serve them cost-effectively, unlike traditional lenders.

Omri: When you first started, how did you approach building a new underwriting model?

Kathryn: Our approach evolved over time. When we launched in 2010, we focused on e-commerce businesses, using real-time data to assess risk. While we didn’t call it cash flow underwriting, that’s exactly what it was.
When we started, our lending model was designed to be as accessible as possible. As long as the seller had consistent eBay revenue for at least a year, they qualified for an initial loan of $500 or more. But as the portfolio matured, around 2013, we saw issues—some of the smaller borrowers weren’t stable. To improve our model, we analyzed the lowest-performing 20% of borrowers—those who struggled the most with repayments. We used that data to adjust our risk criteria, ensuring we could better predict which businesses were likely to succeed. Instead of lending only to the low-risk applicant, we tested a broad range of scenarios and refined our approach over time.

Omri: And if a bank hired you to mentor them today, what lessons could they take from your success to better serve neglected market segments?

Kathryn: I’d say there may be two key lessons. First, understanding the need. Today, the need is bigger than ever. There are 34 million small businesses in the U.S., but only six million have employees. The rest—sole proprietors and independent contractors—are often ignored by traditional lenders. Even among the six million employer businesses, 90% have fewer than 20 employees, but only few financial institutions serve them. The reality is that millions of small businesses still struggle to access capital, just as they did when we started.
Second lesson would be automation. It’s the only way to cost-effectively serve small businesses. Manual applications aren’t scalable—having people walk into branches and fill out paperwork won’t work. Instead, banks need to leverage third-party data to mitigate risk.
This is where fraud becomes a major concern—many applicants alter financial documents, whether it’s bank statements, proformas, or other records, making risk assessment even more difficult. Automation plays a critical role here—by using real-time data verification, banks can detect inconsistencies, flag potential risks, and make faster, more accurate lending decisions.

Omri: So why do banks still hesitate to trust automation?

Kathryn: There’s a saying: The spirit is willing, but the flesh is weak. Many banks want to embrace automation, but internal dynamics make it difficult. At Kabbage, we saw this firsthand working with large banks. When we expanded internationally, the CEO of a major bank personally invited us to help launch a fintech initiative. But when we started working with their local teams, we hit resistance. The small business lending teams in some parts of Europe, for example, had no incentive to change how they operated. That’s the core issue—incentive alignment. If the people executing the strategy don’t benefit from it, they’re less likely to embrace change. In banks, saying no is safe, but saying yes carries risk—and risk-averse cultures make innovation difficult. Without a mindset shift on risk tolerance, traditional institutions will struggle to modernize.
But even when banks want to modernize, the next challenge is getting the models right. You need to ensure you’re not missing opportunities by rejecting the wrong applicants.

Omri: And how did you ensure you weren’t missing opportunities? 

Kathryn: We worked with almost every major data source.We had to trust our data and technology, and balance risk with opportunity. I don’t think we missed a key segment, but there’s a limit—some businesses are just too risky.
That said, FICO was a frustrating necessity. Our investors needed it for historical correlations, especially in securitizations. For example, we found that below 620, default risk was high. But above 660, FICO didn’t matter much.
In many cases, careful personal credit management translated to stronger business finances.

Omri: Did you focus on building ongoing relationships with borrowers, or was it more of a transactional process?

Kathryn: The structure of the product was key—that’s what really built the relationship. From the very beginning, we provided a line of credit rather than a one-time loan. Borrowers could access funds repeatedly, up to their approved limit, and we continuously underwrote them in real time. If we saw a decline in revenue, especially one that didn’t follow a seasonal trend, we could adjust their access accordingly.
This model naturally created long-term relationships. Many customers stayed with us for years—we had borrowers in 2019 who had been with us since 2012. They kept coming back because the product was simple, fast, and predictable.
Even businesses that qualified for traditional bank loans often preferred Kabbage. Banks required annual financial updates, site visits, and lengthy processes, making access to credit uncertain. Other fintech lenders, like merchant cash advance providers, offered one-time loans with prepayment penalties—forcing borrowers to take more than they needed and pay fees on the full amount.
Our line of credit eliminated these pain points. Customers borrowed only what they needed, when they needed it, with no hidden costs. That flexibility was more valuable than any personal relationship with a banker or customer success rep.

‍

“Without realizing it at the time, we built a solution tailored for the smallest businesses, a market that banks largely ignored. And because we automated nearly every step—from application to funding—we were able to serve them cost-effectively, unlike traditional lenders.”

‍

Omri: That’s interesting. I imagine many bankers in our audience might be wondering—if Kabbage and others have mastered automation, what does that mean for the future of lenders and underwriters in banking?

Kathryn: People are still essential, just in different roles. You still need humans to verify models and ensure compliance—even if traditional loan officers aren’t doing manual underwriting anymore, they can transition into customer success roles or other areas where human interaction still adds value.
Even at Kabbage, we used machine learning models almost from the start—what many would call AI today. But we still needed humans to review outputs, especially from unsupervised models, to make sure we complied with fair lending laws and avoided disparate impact.
So, while automation is reshaping the role of lenders, humans remain critical to ensuring the system works ethically and effectively.

Omri: For those thinking about starting a tech company in SMB lending, would you advise them to compete with banks or partner with them?

Kathryn: Neither. We weren’t competing with banks—we were serving a market they weren’t reaching. And when we did partner with banks, it wasn’t always a good fit. The problem came down to incentive alignment.
To work with banks, we changed aspects of our model, but that was a mistake. We should have stayed focused on what we did best—identifying customers, assessing risk, and maintaining relationships. Banks bring brand recognition and capital, but fintechs should retain control over the customer experience.

Omri: Because banks wouldn’t let you keep control of it?

Kathryn: Exactly. They wanted to, and they were bossy about it. It wasn’t just banks—larger partners acted the same way. For example, we partnered with Alibaba, allowing small businesses to use Kabbage at checkout instead of a credit card. We had a similar partnership with Airbnb, but in both cases, we lost control of the customer experience. These companies were bigger brands, and they had strong opinions about how things should work. In the end, those partnerships didn’t perform as well as the programs we ran directly. When you lose control of how customers engage with your product, you risk diminishing the value you bring.

Omri: Any key takeaway for the bankers in the audience?

Kathryn: Small businesses need more than just capital—they need long-term financial support. The sooner you serve them, the stronger the relationship, because early-stage business owners often struggle to access the tools they need to grow.The more services you provide, the more valuable you become to your customers.

This is something banks and fintechs should think about—small businesses don’t just need loans, they need a full suite of financial tools to manage their operations efficiently. The more you integrate into their daily business, the more likely they are to stay with you long-term.

Omri: That’s a great perspective—helping SMBs grow isn’t just about lending, it’s about empowering them with the right financial tools. Thank you, Kathryn.

‍

Kathryn Petralia- Bio

Kathryn Petralia is the Co-Founder of Keep Financial, a fintech pioneering flexible compensation. Prior to Keep, Kathryn Co-Founded Kabbage, an SMB fintech acquired by American Express in 2020. Before Kabbage, Kathryn spent 15 years with fintech and ecommerce startups. In 2018, she was named to Forbes’ World’s Most Powerful Women.

Kathryn serves on the boards of CARE, Tricolor, CoreCard, Cloverly, PadSplit & The Woodruff Arts Center. 

‍