Floor Plan Financing Software: How Do You Catch a Sold Out of Trust Unit Before It Becomes a Write Off?

Why floor plan lending breaks the systems you already own

It is a Tuesday morning and the field audit report lands as a PDF. It says 289 units counted at a dealer group in Ohio. Your servicing system says 312 units outstanding. The nightly dealer management system feed says 296. Three numbers, three systems, and the analyst who has to reconcile them starts by exporting all three into Excel. By Thursday she has an answer: eleven units sit at a satellite lot nobody told you about, eight were sold and the payoff is in transit, and four have been registered to retail buyers for three weeks with no payoff at all. Those last four are sold out of trust. That is a loss you already took, you just found out on Thursday.

Floor plan is not one loan. It is a few hundred loans standing in a trench coat. Every vehicle, boat, coach, tractor or skid steer on that lot is a separate advance with its own invoice amount, advance date, accrual, curtailment schedule, title location and payoff. The credit line is only an aggregate for reporting. Your general ledger sees one exposure per dealer. Your actual risk lives at the VIN or serial number, and any system that cannot hold a unit as a first class object with its own clock is going to hand the join back to a human.

Solifi, NETSOL Ascent and Sopra Banking Software Cassiopae are real asset finance platforms and they do contract accounting properly. Where they thin out is the floorplan module, which encodes a generic curtailment ladder and a generic audit cycle and then expects your programs to bend into it. Your programs do not bend. A captive program with 90 days of free flooring, a used unit schedule that starts curtailing at day 120, and one dealer whose schedule was tightened at renewal after a covenant trip are three rule sets that must coexist in the same portfolio on the same morning. What you actually run is base configuration plus a spreadsheet of exceptions maintained by one person in wholesale operations who is the only one who understands it.

Problem 1: curtailment rules are a negotiation, not a setting

Ask five wholesale lenders what a curtailment schedule looks like and you get five different shapes. Interest free periods subsidised by the manufacturer, then a step to a spread over prime. A first principal reduction at day 180 of 10 percent of original advance for new units and day 120 for used. Aged unit penalties. Model year rollover triggers that pull the whole schedule forward when next year's units land. Different treatment for demonstrators, for units transferred between rooftops in the same group, and for a program you inherited when you bought a competitor's book.

The part packaged systems get wrong is not the arithmetic, it is time. When a manufacturer changes a program in March, what did a unit advanced in January owe on 1 February? If your system recalculates history against the current rule, your billing statements and your audit trail disagree, and that argument with a dealer is one you lose. A custom build models the schedule as a versioned, effective dated rule object bound to the program, the dealer agreement and the unit class, so the system can answer what was owed on any past date without a reconstruction exercise. That single design decision removes most of the exceptions spreadsheet.

Problem 2: the audit is your only real control and it arrives as a PDF

Everything else in wholesale finance is derived. The audit is the only moment somebody physically confirms the collateral exists. And in most lending shops that moment ends as a scanned spreadsheet from one of several audit firms, each with its own column layout, emailed to a shared inbox.

The exception taxonomy is where the money is: unit not found, unit sold and payoff pending, unit at a satellite location, unit out on demo, unit at a body shop, unit wholesaled to another dealer, unit is a duplicate VIN from a re-flooring. Each of those has a different cure clock and a different escalation. Handled in email, cures get missed and a sold out of trust unit hides behind an unresolved not found for six weeks.

A custom build gives auditors a mobile capture app with barcode and VIN scan that works offline, because dealer back lots have no signal and your auditor is not going to stand in the rain waiting for a spinner. Counts post as an audit event, unmatched units drop into an exception queue with a cure deadline and an owner, and the queue is the thing your risk committee reads. Document extraction handles the audit firms who will keep sending PDFs regardless of what you build, turning their layouts into structured unit lists so nobody retypes 300 VINs. That is the honest AI use case here. It is not a model predicting fraud, it is reading documents nobody has time to read.

Problem 3: titles and payoffs are a race you get scored on

Title custody is a state by state mess. Some states run electronic lien and title, some still move paper into a vault, and your operation is probably doing both with a filing cabinet as the source of truth for one half. Meanwhile the payoff arrives as a single ACH or wire covering nine units, sometimes with a remittance advice, sometimes with a subject line that says payoff. Somebody allocates it by hand, and if they allocate to the wrong units the curtailment clocks on the remaining ones are now wrong too.

What a build must do: cash application with unit level allocation rules and a review queue for anything ambiguous, title release triggered by cleared funds rather than received funds, and a hold policy that your credit team controls per dealer rather than per transaction. The satisfying part is that once payoff, title and audit sit in one ledger you can answer the question that used to take two days. Which units on this dealer are unpaid, unaudited, and have a title we no longer hold.

Problem 4: the early warning signals exist, they are just in four systems

Nobody discovers sold out of trust from a dashboard. They discover it from an audit, weeks late. Yet the signal usually existed earlier: a state registration showing retail delivery, a dealer management system sales record, a curtailment payment that slipped from day two to day nine to day fifteen, a sudden spike in line utilisation at month end, an audit exception rate that doubled over two cycles.

None of that requires machine learning. It requires a daily join across sources that currently do not talk, plus an exception engine that raises a case with an owner rather than emailing a report. Where modelling does earn its place is audit frequency: once you have two or three years of exception history, ranking dealers by predicted exception risk lets you audit the risky ones monthly and the boring ones quarterly at the same total audit spend. That is a real return, and it is only available to a lender whose exception history is structured data rather than a folder of PDFs.

What a floor plan build costs and how long it takes

A focused first release, meaning the unit level ledger, effective dated curtailment engine, mobile audit capture and exception workflow, runs $90,000 to $200,000 and ships in 14 to 20 weeks. That is a system your wholesale ops team runs the portfolio on, not a pilot. A full platform adding a dealer portal for payoffs and title requests, electronic lien and title integrations, dealer management system feeds, cash application, early warning scoring and general ledger posting runs $250,000 to $700,000 phased across 9 to 18 months.

What drives the number up in this category specifically: the count of distinct manufacturer and captive programs, because each is a rule set plus its own reporting; electronic lien and title, because each state provider is a separate integration and each has its own certification; dealer management system feeds from CDK, Reynolds and Reynolds or Dealertrack, which are individually negotiated and individually slow; non automotive collateral such as marine, recreational vehicles, powersports or agricultural equipment, where there is no clean sales feed and the serial number conventions are inconsistent; and securitisation, because investor level reporting on a wholesale facility is a project in its own right.

What holds the number down: starting with your largest program and your top 30 dealers by exposure, and leaving the dealer portal for phase two. Dealers will adopt a portal only after the internal system is trustworthy anyway.

Build versus buy, and when Solifi or NETSOL is the right answer

Buy if you fund a few hundred units on one manufacturer program with a standard schedule, your audits are already outsourced with structured output, and your title work is entirely electronic. Solifi and NETSOL Ascent will serve you, configuration will take a quarter, and a custom build would be an expensive way to get the same ledger.

Build when two or more of these are true. You run three or more programs with individually negotiated variations at the dealer level. Your exceptions live in a spreadsheet that one person maintains. You have taken a sold out of trust loss in the last 24 months and the post mortem said the data existed but nobody joined it. You fund collateral classes with no reliable sales feed. Or you securitise, and your investor reporting is assembled by hand each month.

The tipping point is not portfolio size on its own. It is when the exception handling logic, which is where the credit losses actually happen, has become the part of the business that no vendor models and no one person can safely own.

How to choose a developer for wholesale finance software

Ask them to whiteboard the unit lifecycle before you sign anything: advance, accrual, curtailment schedule version, audit event, exception, cure, payoff allocation, title release, and the dealer agreement that governs all of it. A team that has done asset finance draws effective dated rules without being prompted. A team that draws loans and payments has built a consumer lending app and is about to learn wholesale on your budget.

Ask specifically how they will answer what a unit owed on a date in the past after a program change. If the answer involves recalculating from current rules, stop there.

Ask what they have integrated. Electronic lien and title providers, dealer management systems and core banking or general ledger systems are three separate competencies and each takes real weeks. Ask for the named provider and the named document, not a claim about integrations in general.

Ask who owns the code, the repository and the cloud accounts, and get it in the contract before kickoff. At Digital Heroes the client owns everything from the first commit, and you should walk from anyone who treats the repo as their asset. In lending, where a regulator or an investor may one day ask how a number was produced, that ownership is not a preference, it is your ability to answer.