All Posts

Most lenders underestimate small business (SMB) credit. They treat it like commercial lending—too complex to scale—or like consumer lending—too simple to price correctly. It’s neither. SMB underwriting sits in the middle, and that’s exactly what makes it powerful—and difficult. “Small businesses are just big humans” In consumer lending, the playbook is straightforward: pull a bureau, run a model, make a decision. The data is standardized, the variables are well understood, an

It’s Friday afternoon before a holiday weekend when you get an alert from one of your call centers. They’ve found an “odd” account application. You review the entry. It passed KYC. It passed your fraud risk checks. Everything seems fine. But by Monday morning you have an assortment of similar accounts that are attempting odd purchases, or exhibiting high risk behavior. Yet the only thing they have in common is they applied in all caps. At least that is all you can see from yo

There's no shortage of enthusiasm right now about applying large language models and AI agents to credit underwriting. We want to be clear about where we stand on that — and then make a more interesting point about where these tools genuinely do add value. What LLMs Are Not For We've written elsewhere about why supervised machine learning algorithms — gradient-boosted models like XGBoost in particular — remain the right foundation for credit underwriting. The reasons are well

Most lenders spend the majority of their time optimizing underwriting, pricing, and collections. Far fewer spend time optimizing credit line. That’s a mistake. At Ensemblex, we consistently see that credit line is one of the most powerful—and most underutilized—levers in a lending business. When managed well, it can drive meaningful improvements in conversion, utilization, and profitability. When managed poorly, it quietly erodes returns through missed revenue or excess risk.

It's a question that sounds simple but turns out to be surprisingly hard to answer, especially in the early months after a model goes live. Part of what makes it hard is timing. In a credit business, the outcomes you care most about take time to materialize. A model trained to predict severe delinquency at 12 months on book can't tell you much about whether it's working after two weeks in production. You have to build toward that answer gradually, using a sequence of leading

A founder's framework for turning target market, risk, and rewards strategy into a credit card product that actually works You've decided to build a credit card. Maybe you're deep in conversations with a sponsor bank, or weighing a BaaS partner against a full build. Either way, somewhere between the pitch deck and the term sheet, a harder question shows up: what should this card actually look like? Practically speaking: What are the terms and features? APR. Credit limits. Fee

Most early-stage lenders don’t have a pricing problem. They have a learning problem. Recently, I worked with a fintech lender that had built a surprisingly sophisticated pricing system: ~600+ pricing cells (term × loan amount × bureau score buckets x customer channel) ~20 distinct price points in one point increments Strong portfolio performance and risk adjusted returns On the surface, things “worked.” But underneath, there was a fundamental issue: They had never run a singl

A model that ranks applicants by risk is only half the job. At some point, you have to draw a line — and how you draw it matters as much as the model itself. Score cutoffs come up constantly in conversations with clients and prospects, and the questions around them are often underappreciated. Where should the line be? What's driving that decision? And once you've set it, when do you look at it again? This post walks through how we think about cutoffs: what they're actually do

When lenders want to expand access to credit without blowing up risk, secured products often look like the obvious answer. Add a deposit. Reduce losses. Open the funnel. Problem solved. In practice, secured and partially secured credit products solve one problem while quietly creating several others. The question isn’t whether they reduce risk — they do. The question is whether they actually create a durable credit business. That answer depends far more on adoption, usage, an

The honest answer — and the four factors that determine whether it takes 60 days or 8 months (or never!). The Question Every Lending Fintech Founder Eventually Asks You've validated the concept. The product roadmap is drafted. The pitch deck is sharp. And someone on your team (usually someone who's done this before) says: "We need to start talking to sponsor banks." The follow-up question comes immediately: "How long is that going to take?" The honest answer is: it depends. B