FundKite Did $70M With Five Engineers. Its Worst Defaults Come From Debt Settlement Firms
Alex Shvarts runs a $70 million revenue-based finance business with five engineers and writes the underwriting code himself. Banks, he says, cannot underwrite a pizzeria.
On this page
Nine out of ten of FundKite’s defaults are not businesses in trouble. They are businesses that got a phone call.
They’re going in because the debt settlement companies call them and say, basically, stop paying, we’re gonna negotiate for less. That’s the problem. Because the merchants that really have a problem call us — we just lower the payments based on their sales. We work with them.
Alex Shvarts, CEO, FundKite
The thesis. FundKite did about $70 million of revenue in 2023 with 90 people and five engineers, because the founder writes the underwriting code himself. The constraint in small-business finance is not capital — it is reading a pizzeria’s bank statements fast enough to make an offer, and banks cannot do it.
The product is not a loan
FundKite does revenue-based finance: buying future receivables at a discount, then collecting a fixed share of sales as they arrive.
Let’s say we bought a $100,000 worth of receivables for a lump sum payment of $90,000 today. And we’ll collect 10%, or whatever that fixed ratio is, of your receivables. So if you did 5,000 in sales this month, we’ll collect $500. If you did 3,000, we’ll collect 300.
There is no term. Shvarts models eight to sixteen months and says it is very unpredictable. The discount runs from 10% to about 28% depending on risk.
Latka presses on the effective interest rate, and the exchange is the clearest explanation of the category on any of these tapes. Shvarts distinguishes it from factoring, which underwrites a buyer against a purchase order — his example is a manufacturer with a million-dollar Home Depot order who needs $500,000 to produce the goods, where Home Depot writes the cheque to the factor.
With us, we’re buying receivables. We don’t know if they’re gonna have them. What happens if their sales go down? It’s a hurricane, it’s COVID — there is no guarantee that they’re gonna have receivables.
California and New York require an APR disclosure at origination. Shvarts complies and says it is not always accurate, because the term is not fixed: “the better way to look at this is the cost of capital. I’m getting $90,000, I gotta pay back $100,000, my cost of capital is $10,000.”
The funnel, and who gets turned away
Customers are businesses doing $1–5 million a year in gross sales, across restaurants, e-commerce, medical, agriculture, supermarkets and manufacturing. The exclusions are more interesting than the inclusions, and each has a mechanism behind it.
- Trucking and transportation — “the biggest problem they have is accounting for the cost of goods, which in their case might be fuel. They’re booking the jobs at one rate, they’re not getting paid fast enough.”
- Construction, pulled back — delayed payments and stalled projects; when rates went up, commercial developers could not refinance.
- Bail bondsmen — excluded outright.
Losing deals happens to OnDeck, Shopify, PayPal and Square. Shvarts is dismissive of much of the rest: “a lot of them are kind of lead gen — they’re telling you they’re doing it, and then they send it out.”
Reading the bank statements
The underwriting philosophy is the whole business, and Shvarts explains it with a fast-food metaphor.
You take a Burger King in New York City and you take a Burger King in Toledo. It looks like it’s a Burger King — two different operators, different costs, labour, rent, everything. And you’ve got two different people that run businesses and cash flow two different ways. So you really have to go through those bank statements line by line to understand true sales revenue and how they operate.
Credit comes second, and he is emphatic about why.
You can have somebody with 550 credit — they’re just utilising a lot of their credit and their score is down. You can have somebody with a 780 FICO score and they have no credit, they’ve got one credit card. So credit isn’t always a true reliable factor for us. It’s how they got there.
Which is the gap he thinks banks cannot close, even as regulation pushes them back toward small business lending: “banks are not equipped to underwrite a small business. They can’t underwrite a pizzeria. They don’t understand how to look at bank statements like we do.”
The loss numbers
Life-to-date write-offs run about 6.8% of receivables purchased, through the end of 2023. Roughly 12–15% of the portfolio hits some collection issue, and about half of what enters default or collections is recovered.
The mechanics of a default are unglamorous. FundKite has permission to debit the merchant’s account and does so on a reconciled schedule — but the merchant can block it.
We send a notice, this is your reconciliation, we’re gonna debit $92 from your account tomorrow. And then all of a sudden they block it. You reach out, you email, you call. If they go dark, that’s a breach and we pursue legal.
There is a UCC filing on the receivables and arbitration if it comes to that. Shvarts says they have worked out a policy that makes suing cost-effective even on small balances — and that if a business genuinely closes, they do not sue at all: “you just want some verification that they’re closing down, and you write it off. That’s the risk you take.”
How the money is raised
FundKite has capacity to deploy about $30 million a month and has raised almost $200 million in debt and participation. What it has not raised is the thing most fintechs raise.
We don’t have any warehouse facilities or senior lines. We don’t take on capital like that. We have investment partners that participate in these transactions.
At the operating company it issues structured notes — Crunchbase records $16.8 million in October 2023, since increased — with a structure worth noting.
Note return = stated interest + contingent interestA fixed coupon plus upside on how the portfolio performs. FundKite’s own fees are charged regardless.
Cost of capital is 12% or less against 13–15% that Shvarts says is typical in his space, and he does not think there is room to push it lower right now.
The economics, done his way
Latka tries to model the margin as revenue minus defaults. Shvarts reframes it around velocity.
Let’s say your average factor is 1.30. You take $100 million, you’re buying $130 worth of receivables. You’re going to turn that money probably one and a half to two times a year. So with $100 million you’re not gonna buy $130 million — you’re gonna buy 200, maybe $260 million of receivables, because you’re collecting money and putting it back out.
Net interest margin lands around 16–18% in a good year. Revenue for 2023 was about $70 million, before cost of capital — a figure the GetLatka profile records for the year, and one Latka only gets to after his own back-of-envelope lands at a third of it.
Five engineers and a CEO who still writes code
Ninety people, five of them engineers, in a business processing five thousand applications a month. Asked whether that is really automation, Shvarts answers by describing his own week.
I’m like the CEO who’s first to tech. I’ve built our platform, architected everything, and I know it needs to be done quickly. I’m able to bridge all that together very quickly, as opposed to explaining to a developer who’s gonna sit there and write this out for thirty days. We want stuff in days that other people take months.
The automation is the growth plan: going from 5,000 or 7,000 applications a month to 50,000 or 100,000, with the labour and processing cost falling as it scales. The 2024 goal is $300 million deployed — “$100 million would be below my expectations” — and well over 100% growth.
The newest channel is credit card processing, aimed at merchants the big platforms leave behind.
Unless they invite you to take money, you can’t get a Shopify loan if you’re sitting on their platform. And they cap you out. So we’ve had a lot of Shopify merchants turn to us and say, hey, Shopify only gave me this, but I need this. And these are good businesses with really good track records.
On who eventually buys a company like his, he sees exactly two candidates: a bank that needs a working underwriting and funding stack, or a hedge fund with an enormous amount of money and no vehicle to deploy it through.
Asked what he wishes he had known at twenty, the founder who still writes the underwriting engine gives the only possible answer.
Studied code back then. I started to study code way, way older in life. Should’ve done code when I was 20, but we only had, like, Commodores and Atari.
Sources Alex Shvarts’s interview with Nathan Latka, recorded 7 March 2024; revenue row from the GetLatka FundKite profile; the October 2023 note raise as recorded on Crunchbase and discussed on the tape.
