A lending fintech has been growing for two years. It has customers, it has a loan book, and it finally lands the meeting it wanted: an international credit fund wants to get to know it. Before talking rates, the fund asks for the loan-level data. The analyst runs a single test, loan by loan: opening balance, plus disbursements, minus principal payments, has to equal the closing balance.
In hundreds of loans, it doesn't. The conversation doesn't end over the rate or the risk. It ends right there.
No one stole anything. Most likely someone mistyped a number, or a system stopped updating a field. But from the outside, a typo and a serious problem look exactly the same.
Now, the opposite scene. Not long ago in Colombia, a well-known cattle rancher raised money from many investors for a business backed by heads of cattle. The story was good, the name was trusted, and the numbers looked right. No one checked whether the animals existed. They didn't.
An honest company with messy data goes without funding. A well-told story about a nonexistent asset gets it. The market is confusing order with truth.
Everyone has errors
Alejandro Arenas evaluates lending fintechs in Latin America for CIM, an impact credit fund based in San Francisco, and he said it plainly on All You Can Debt: he has never come across a company without errors in its data. Not one. There was a time when half of the deals fell apart because fixing the information with the originator was impossible; today he hopes it's below 30%.
The most common error is the simplest: balances don't reconcile. And his advice to founders is just as simple: before sending your data to a lender, think about what analysis they'll run, and run it yourself first.
It sounds like an operational detail, but it's the core of the business. For a credit fund, data is the only way to see a portfolio's past, and the past is all it has to estimate the future. That's why, Alejandro says, equity is for proving a product works and debt is for scaling it once there's a track record. How much track record depends on the product: a factoring portfolio at 60 or 90 days turns over several times in a year, while a 60-month auto loan, at month twelve, has shown almost nothing. Either way, if the data doesn't reconcile, that track record doesn't exist, even if the loans did.
The perfect spreadsheet
The problem is that the market is using order as if it were proof. A dataset that reconciles brings peace of mind, and one that doesn't raises suspicion. But reconciliation only shows that the numbers are consistent with each other, not that they correspond to anything real. The rancher surely had flawless spreadsheets too.
Alejandro gave a clue about how this works in practice when he talked about the backup servicer, the standby administrator that steps in if the originator fails. In one of his deals, the originator was liquidated, and even though they had the right to replace it, they decided to keep working with its team, because no one knew the portfolio better. The only case in which they would activate the backup without hesitation, he explained, is fraud: when trust in whoever services the loans breaks.
So everything depends on being able to tell an error from a fraud in time. And that isn't solved with a tidier spreadsheet. It's solved by looking at the source.
What the data has to prove
Alejandro closed the conversation with a book, When Genius Failed, about the fund of Nobel laureates that collapsed in the late nineties. His takeaway: no matter how smart and prepared you are, things always happen that no one had in the model. The way to protect yourself isn't to trust the model more, but to trust what feeds it.
At Vaas, we verify the existence, ownership, and cash flows of every asset against the source, every day. That lets an organized fintech prove its order is real, and lets a lender tell a typo apart from cattle that never existed. Balances that reconcile are the first step. Balances that reconcile with reality are what turn data into trust.
The full conversation with Alejandro Arenas, including how one fintech helped importing SMEs shrink their cash cycles from six months to one or two, and why he believes Colombia's usury rate ends up making credit more expensive for the very people it tries to protect, is on the second episode of All You Can Debt.
▶️ YouTube: https://sl1nk.com/8trd9gx
🎙️ Spotify: https://l1nk.dev/cc0dfvi
How Vaas verifies the asset is actually there: https://getvaas.com