Which channels produced borrowers who funded and came back?

A fix-and-flip investor who likes your process will come back for the next project, and the one after that. A productive borrower relationship can run to a dozen loans, and the acquisition cost was paid once, on the first.

Measuring cost per funded loan therefore overstates the cost of the good channels by roughly the number of times their borrowers return. It is the same arithmetic error that makes subscription businesses look unprofitable if you only count the first month.

Matching on the borrower rather than the loan attaches every subsequent origination to the channel that found them, which usually separates the channels far more sharply than first-loan economics do.

Origination points arrive at funding, interest accrues monthly through the term, and exit or extension fees land at the end. A loan that extends twice earns materially more than one that pays off on schedule.

Exporting each as its own row against the same borrower lets the report rank channels on total yield rather than on origination alone. Channels that produce borrowers who extend are worth more, and nothing else you have will show it.

Nobody pays these rates for the pricing. They pay for certainty and a closing date, because the borrower is standing in front of an auction deadline, a contract that expires on Friday, or a bank that has just pulled out three weeks into diligence. That urgency dictates the behavior: the borrower rings, they want somebody who understands the deal on the first call, and they are comparing lenders on how quickly a term sheet appears in their inbox rather than on a landing page. A form submitted at eleven at night is a lead; a call answered in four rings is a loan.

This matters for measurement because the phone is where the intent lives, and the phone is the weakest point in almost every lender's reporting. The advertising platform records that somebody clicked an advertisement. The origination system records that a loan funded and what it earned. Between those two records sits a conversation nobody wrote down, held by whoever happened to be at the desk, and in a business where most serious inquiries arrive exactly that way, that missing link is not an edge case to be tidied up later. It is the main route to revenue.

A call export supplies the caller's number with the campaign attached, and the funded loan file supplies the same number with the points attached. Joining them on a phone number the network captured, rather than on anything a borrower typed, is why lenders who already run call tracking tend to reconcile a high proportion of their fundings on the first attempt. Those who do not can still start from the platform export and usually add call tracking once the first report shows how much of the book arrived by phone.

Interest that accrues is not interest received. A borrower who stops paying in month four still shows a growing balance in the servicing system while the lender receives nothing at all, and what happens next takes months: forbearance, a modification, a default notice, a foreclosure, occasionally a property taken back and sold. Some of that principal comes home with the accrued interest attached and some of it does not, and the outcome is not known for a year or more after the origination that a marketing report already scored as a success.

Channels differ on this, and they differ in ways nobody would guess without looking at the outcomes. A source producing experienced sponsors on their eleventh project and a source producing first-time renovators acting on the promises of a weekend seminar can originate at similar volumes, at similar points, in the same quarter, and then perform completely differently over the following eighteen months. The origination figure flatters both equally and identically, which is precisely what is wrong with using it to decide where next quarter's budget goes.

Exporting cash actually received, with a revenue-type column mapped as Product / service, puts the two on an honest footing. Points collected at funding, interest received, extension fees paid, and recoveries on a loan that went wrong all become rows against the same borrower, crediting the channel that produced them. The ranking that comes out is a statement about credit quality by source as much as about marketing, and for a lender it is difficult to think of a more useful pair of facts to have in one table.

A lender's constraint is capital, not demand. There are stretches where the facility is drawn, the fund is committed, the warehouse line is at its limit and the credit committee is turning down deals it would have written gladly two months earlier. In those periods, generating more inquiries does not generate more revenue; it generates declined applications, wasted underwriting time and borrowers who will remember being turned away when they come back with a better deal. Marketing spend in that window has a negative return, and every lead metric in use will report it as a strong month.

The inverse case is equally real and equally unmeasured. When capital frees up because a run of loans paid off early or a new facility closed, the lender needs deals immediately, and the channels that can be turned up quickly are worth considerably more than their average cost per funded loan suggests. Responsiveness becomes a property of a channel that matters as much as its price, and no report a lender currently runs measures it, because every report treats the month as the unit and the capital position as somebody else's problem.

Ranking on yield collected per period, with funding dates on every row, lets a lender line the marketing up against the capital cycle rather than against the calendar. The useful question stops being which channel is cheapest on average and becomes which channel produced deployable loans at the specific moments there was money to deploy. That is a different question with a different answer, and it can only be settled from the lender's own funded records, because no advertising platform knows or cares whether the line was drawn that week.