SBA loans for SaaS and software acquisitions

1. Cash flow carries the debt, not ARR. No lender in the SBA or FLEX market sizes a loan off an ARR multiple. The note is sized against adjusted cash flow after a market rate salary for you, and coverage has to clear 1.25x on a fully amortizing basis, with most credit committees wanting a visible cushion above that. On software targets the add-back conversation is unusually contested: capitalized development costs, founder compensation taken as an S corp distribution rather than payroll, one-time platform migrations, and hosting credits that expire after close. We normalize all of it before the file goes out, because an add-back removed in underwriting removes coverage with it.

2. Revenue durability, measured two ways. Gross revenue retention and net revenue retention answer different questions and only one is a credit question. NRR flatters a deal by letting expansion inside surviving accounts paper over accounts that left. GRR shows what the base does with no help. Private SMB SaaS clusters around 90% to 95% GRR, with better files above 95%, and lenders get uncomfortable below the high eighties because the model then needs new sales to hold the line, and new sales under a first-time owner is exactly the assumption a credit committee will not underwrite. We show logo churn and dollar churn separately for the same reason: losing 15% of accounts that represent 4% of revenue is a very different business from the reverse.

3. Contract term mix and the committed base. A month-to-month book and an annual prepay book with identical ARR are not the same collateralized cash flow. We build a renewal calendar showing what share of revenue is under annual or multi-year contract, when each cohort comes up, whether agreements carry auto-renewal and price escalators, and whether they contain change-of-control clauses a stock or asset purchase would trip. Committed revenue extending past the first twelve months post-close is the most persuasive exhibit in a SaaS credit memo.

4. Concentration and the tail. Any account above 10% of revenue draws a question, above 20% draws a structural response: usually a larger seller note on standby, an earnout tied to retention of the named accounts, or a holdback, rather than a decline. We would rather propose the mitigant in the memo than have a lender invent one after term sheet.

5. Transferability of the product itself. If the seller is the lead developer, the lender is lending against one person's continued goodwill. The mitigants are concrete: a documented codebase, a second engineer who has shipped to production, a transition and consulting agreement with real hours, and a non-compete. We price that risk into the structure at the outset instead of letting it surface three weeks before funding.

The ARR bridge, reconciled to cash: We rebuild ARR from the billing system rather than accepting the CIM figure, and reconcile it to bank deposits and the tax return. The test is whether the reported number is contracted ARR (signed agreements at their contractual rate) or annualized MRR (one month times twelve, which imports every one-time charge, seasonal spike, and unconverted trial). We separate subscription revenue from implementation fees, professional services, usage overages, and reseller pass-through, because a lender discounts each differently.

Cohort retention pulled from the source system: Retention gets built from raw subscription events by signup cohort, not from a summary slide. We want gross revenue retention and logo retention side by side over at least twenty four months, downgrade and expansion isolated from each other, and involuntary churn (failed cards) shown separately from cancellations, since the first is a dunning problem a buyer can fix and the second is a product problem that may not be fixable.

Concentration, contracts, and assignment: We map revenue by customer for the top twenty accounts, then read the actual agreements. Assignment and change-of-control clauses matter more here than in almost any other category: a target where the top ten customers can each terminate on notice at close is a very different risk than one locked to annual terms, and consents need lining up before funding rather than after.

Deferred revenue and the working capital peg: Annual prepay is a cash flow advantage and a diligence trap. Cash collected before the service is delivered is a liability the buyer inherits and must perform against with no offsetting cash if the seller has already spent it. We quantify the deferred revenue balance at close, negotiate a working capital peg that delivers that liability funded, and make sure the lender sees the peg rather than discovering a hole in month two.

IP assignment chain, open source, and third-party dependency: Every contributor to the codebase needs a written assignment, including offshore contractors and the friend who wrote the original prototype. We run an open source license review for copyleft obligations that could compromise a proprietary product, confirm trademark and domain ownership sit inside the entity being purchased rather than the seller personally, and inventory every third-party API, model provider, and platform the product depends on, including the commercial terms and whether they survive a change of control.

Key person coverage and the operating handoff: We document who holds deployment access, infrastructure credentials, and the customer relationships, then build the transition package around the gaps: a written runbook, a defined consulting period, and where the founder is the sole developer, an explicit engineering coverage plan the credit committee can read.