Skalar underwrites startup marketing spend for a 10-cent spread on every dollar it advances

A 1.1 times collection is not an interest rate until you know how long the money is out. In the example Skalar gave Crunchbase News, a company spends $10 to acquire a customer who then pays $1 a month for 30 months. Skalar takes the first $11 before the company keeps the rest ^[7]. Those eleven dollars arrive one at a time, so the average dollar advanced is back in about six months. Ten cents earned over six months on a dollar works out near 20 percent a year ^[1].

The gain and the loss run on different scales. Skalar's maximum return on a dollar is 10 cents; its maximum loss is the dollar. A book earning 1.1 times on its good cohorts breaks even once 9.1 percent of principal goes to zero ^[2]. Severity is rarely total, though. In the churn case Cardenas described, a customer who cancels at month eight has paid $8 against the $10 advanced, a loss of 20 cents on the dollar ^[8]. At that severity about a third of the book can churn early before the spread is gone ^[3].

So the diligence looks like a credit shop's. According to co-founder and COO Daniel Castrillon, Skalar reads detailed transaction data to establish what a company spends to acquire a customer, how long those customers stay and how much they pay. It updates its assessments as new data arrives ^[9]. "We have become experts in understanding these types of risks and when they are sufficiently predictable and sufficiently profitable to be underwritable," he said ^[10].

The flexible timeline has conditions attached. Repayment is tied to the revenue rather than a maturity date ^[11], and co-founder and CEO Sebastian Cardenas told Crunchbase News, "We only get repaid as they get repaid" ^[12]. But Skalar sets minimum revenue targets, can require faster repayment when a company falls short of them, and can stop advancing further capital ^[13]. A borrower building a hiring plan on the next tranche is exposed to that second clause.

More than $125 million across seven companies over 12 months averages $17.9 million each ^[4]. For most companies at that stage, that is a sales and marketing line that would otherwise come out of an equity round or the operating account ^[2].

If the seven names behave like the example, Skalar is selling roughly 20 percent money against cohort data and the equity it does not take carries the pitch. If acquisition costs rise across the book at once, the misses arrive together and nine cents of buffer does not absorb much. Rising acquisition cost is one of the estimates Cardenas said the terms rest on ^[14]. The third version is duller, and I think more likely. Companies with 30-month retention and clean attribution are also the companies best able to fund marketing from cash, so the selectivity that makes the underwriting work shrinks the market it works in.

Realized loss severity by cohort would settle which version this is, and a book started in January does not have much of it yet ^[2].