This interview is with Travis Vayssie, CFO, Swipe Solutions.
What’s the story of how you moved from traditional finance into building a tech-enabled matching business, and what pivotal decision most changed your trajectory?
The catalyst was watching my parents get crushed by a predatory loan. They needed money fast, didn’t know their options, and ended up with terms that took years to escape. That experience stuck with me.
I was working in lead generation and saw how the lending industry operated behind the scenes—how some lenders competed for borrowers while others preyed on them. The information asymmetry was massive. Borrowers had no idea which lenders might approve them or what rates were reasonable for their situation.
The pivotal decision was choosing to build for the borrower instead of just selling leads to the highest bidder. In 2018, I launched Swipe Solutions with a simple premise: show people their actual options before they apply, protect their credit scores from unnecessary inquiries, and filter out the predatory options entirely.
That meant walking away from easier money. Selling leads without qualification pays fine. Building technology that genuinely helps borrowers find fair terms takes longer and costs more upfront. But seven years and 50,000 borrowers later, it was clearly the right call—both ethically and as a business.
Building on that, describe a moment when a financial objective (margin, cash, or growth) collided with engineering reality—how did you resolve it, and what changed in your roadmap as a result?
Early on, we were sending every lead directly to our lending partners without pre-qualification. Volume looked great on paper, but our funding rate was terrible—under 1%. Partners were getting frustrated with lead quality, and we were burning cash on traffic that converted to nothing.
The growth objective stated “more leads, faster.” The engineering reality indicated “we need to build a qualification layer first, and that’s a three-month project minimum.”
We faced a hard choice: keep pushing volume to hit short-term revenue targets, or essentially pause growth while we rebuilt the funnel. We chose to build.
That meant implementing a pre-qualification system that scored leads before they ever reached a lender. Credit score ranges, income verification, and employment status were all captured upfront. Unqualified users were redirected to credit-building resources instead of wasting everyone’s time.
In the short term, revenue dipped. We were rejecting leads we previously would have monetized, even if poorly. However, within six months, our funding rate jumped significantly, and partner relationships strengthened because we were sending them borrowers who could actually get approved.
The lesson reshaped our entire roadmap: engineer for quality first, then scale. We’ve never deviated from that sequence since.
On your matching engine, how did you design it to balance approval rates, CAC, and unit economics, and what single change produced the biggest lift in margin or conversion?
The matching engine runs on a simple principle: don’t pay to acquire leads you can’t monetize. That sounds obvious, but most lead generation operations optimize for volume first and figure out quality later. We flipped it.
The system scores every user before we spend a dollar on them. Credit tier, income range, employment type, loan purpose, even geographic factors—each variable feeds into a qualification score. Users below the threshold receive educational content and credit-building tools. Users above the threshold see lender matches ranked by predicted approval likelihood for their specific profile.
This change eliminated our Customer Acquisition Cost (CAC) problem. We stopped competing for bottom-of-funnel traffic that looked cheap but never converted. Higher quality traffic costs more per click, but the unit economics work because funding rates justify the spend.
The single biggest lift came from adding loan purpose segmentation. Our data showed emergency loan seekers faced 93% rejection rates, while debt consolidation applicants saw only 22% rejection—with identical credit profiles. The same borrower, completely different outcomes based on why they needed money.
Once we started routing users differently based on stated purpose and adjusting lender matching accordingly, conversion jumped dramatically. We published that research publicly because the finding was too important to keep to ourselves.
On the risk and regulatory side, share a time you tightened credit or product eligibility—what signals triggered the move, and what was the measurable impact on growth and losses?
We’re a matching platform, not a lender, so our “credit tightening” happens at the qualification layer—deciding which borrowers we’ll send to partners versus redirect elsewhere.
The trigger was chargebacks and partner complaints. We noticed a pattern: borrowers who listed income below $25,000 annually and selected “emergency/unexpected expense” as the loan purpose had abysmal funding rates—low single digits. Worse, some who did get funded defaulted quickly, which came back on our partner relationships.
The signal was clear. That segment wasn’t being served well by our lender network, and forcing the match helped nobody.
We raised the floor. Users below that income threshold with emergency loan purposes now get routed to credit union resources, payday alternative loan information, and credit-building tools instead of our standard matching flow. We’re transparent about why—these options genuinely serve them better.
Growth impact: we lost roughly 15% of our lead volume overnight. Painful.
Quality impact: partner satisfaction improved measurably, funding rates on remaining leads increased, and we stopped contributing to a cycle where desperate borrowers took loans they couldn’t sustain. The math worked out within two quarters, and we sleep better.
Shifting to data, what are the 5–7 metrics you review weekly to manage margin durability, and how did you instrument the stack (tools, pipelines, dashboards) to trust those numbers?
Seven metrics hit my desk every Monday:
- Funding rate—leads that convert to actual funded loans. This is the number that pays the bills. Below 2%, we’re losing money on traffic.
- Qualification rate—percentage of visitors who pass our pre-qualification screen. Too high means we’re too loose. Too low means we’re filtering out viable borrowers.
- Revenue per lead—not per funded loan, but per lead. This metric forces us to account for the 98% that don’t fund.
- CAC by channel—what we paid to acquire each lead, broken down by traffic source. Some channels may look cheap until you see their funding rates.
- Partner acceptance rate—are our lending partners taking what we send? A declining acceptance signals quality drift before it impacts revenue.
- Time to funding—how fast approved borrowers actually receive money. This metric matters for borrower satisfaction and repeat business.
- Credit tier distribution—the mix of excellent, good, fair, and poor credit coming through. Shifts in this distribution predict margin changes weeks in advance.
The stack is deliberately simple. PostgreSQL holds everything. Python scripts pull daily snapshots and push them to Slack channels—no fancy BI tool that becomes its own project. We tried Databox, removed it, and went back to automated Slack reports.
I trust the numbers because they come straight from the database, not transformed through three tools first.
When deciding build vs. buy for core finance/fintech components, tell us about one vendor you replaced with an in-house system (or vice versa)—why did you choose that path, and what would you do differently now?
We ripped out Databox and replaced it with homegrown Slack reporting. This was a classic case of a purchased solution becoming more overhead than value.
The appeal was obvious—pretty dashboards, integrations with everything, and clients love seeing real-time metrics. However, the reality was different. We spent hours configuring data connections that broke whenever our schema changed. The metrics we actually needed required custom calculations that conflicted with their aggregation defaults. Additionally, nobody checked the dashboards daily; the information was too removed from where work happened.
So, we built simple Python scripts that query PostgreSQL directly, calculate the metrics that matter, and push formatted reports to Slack channels every morning. Total development time: maybe two days. Infrastructure cost: essentially zero—runs on cron jobs we already had.
The switch eliminated a $100/month subscription, but more importantly, it eliminated the cognitive overhead. Metrics show up where we already work—no logging into another tool, no remembering passwords, and no fighting with visualization settings.
What I’d do differently: trust that instinct earlier. We kept Databox for months, hoping we’d “figure out” how to make it valuable. This was a classic example of sunk cost thinking. The moment a tool requires effort to use, it’s already failing. Simple and present beats sophisticated and ignored every time.
Build when the problem is core to your business. Buy when it’s a commodity. Dashboards felt core but weren’t.
Looking ahead to 2026, if you had to place one bet that CFOs should make to improve matching accuracy and monetization in a tougher credit environment, what would you implement in the next 90 days and why?
Segment by loan purpose before you segment by credit score.
Everyone obsesses over credit tiers, but our data shows that loan purpose predicts approval likelihood more dramatically than most CFOs realize. Emergency borrowers face 93% rejection rates. Debt consolidation applicants with identical credit profiles see a 22% rejection rate. Same FICO, completely different outcomes.
In a tightening credit environment, lenders get pickier about why someone needs money, not just whether they can repay. Urgency signals risk. Debt consolidation signals someone actively managing their finances. Home improvement signals asset investment. These distinctions matter more when lenders have fewer dollars to deploy.
The 90-day implementation: instrument your funnel to capture loan purpose at the first touch, before you spend acquisition dollars. Route each purpose to lenders who actually approve that use case. Stop sending emergency loan seekers to lenders who will reject 93% of them—you’re burning CAC and frustrating borrowers simultaneously.
Then, build purpose-specific content and landing pages. Someone consolidating debt needs different information than someone covering an emergency. Match the experience to the intent.
This isn’t sophisticated data science. It’s basic segmentation that most platforms ignore because credit score feels more “financial.” But in 2026, the operators who match borrower intent to lender appetite will capture margin while everyone else fights over declining approval rates.
The data is already in your funnel. Use it.
Thanks for sharing your knowledge and expertise. Is there anything else you'd like to add?
One thing I’d emphasize: the lending industry has an information problem disguised as a credit problem.
Millions of Americans get rejected not because they’re unqualified, but because they apply to the wrong lenders, for the wrong products, at the wrong time. Our research shows that 42% of rejected borrowers give up after one denial—never discovering that different lenders evaluate applications completely differently. Persistence through multiple applications increases approval odds by 340%.
That’s not a credit score issue. That’s a knowledge gap. And it’s why we publish our research publicly, even data that doesn’t directly benefit our business. The 93% emergency loan rejection rate, the state-by-state disparities, and the 580 score threshold where approval rates triple—this information should be accessible to anyone trying to navigate the system.
If you’re a borrower reading this: check your rates with soft-pull tools before applying anywhere. One rejection doesn’t mean all rejections. And if someone guarantees approval regardless of credit, walk away—that’s not a lender; that’s a trap.
If you’re building in fintech: the unsexy problems matter most. Matching algorithms, qualification logic, and borrower education—this infrastructure determines whether someone escapes a financial emergency or spirals into predatory debt. Build accordingly.