Six months ago I wrote here that interest-only periods make Year 1 coverage look good on almost every deal, and that Year 3 is the real stress test. Since October 1, the SBA goes further than that. SOPredactedtook effect that day. Most of what I have read about it is a list of changes. I wanted to know what it does to a real model, so I ran one. Take a $2.4M business. With fees and working capital, total project cost is $2.55M. The stack is 10% cash, a 10% seller note at 8% with 24 months interest-only, and $2.04M of SBA at 10%. Adjusted earnings after a market GM salary: $410K two years ago, $430K last year, $540K trailing twelve. The business is growing. This is a deal you want to like. Here is how most of us have been modeling it. Year 1 debt service is $343.9K, because the note is interest-only. Divide $540K by that and you get 1.57x. Comfortable. Now do what the SOP says. Coverage has to be met on the last fiscal year or the average of the last two. Trailing twelve does not count, and the lender cannot rely on projections. So $430K, not $540K. $430K over $343.9K is 1.25x. It still looks fine. This is the trap. Debt that is not on full standby and is interest-only must be underwritten at an amortization of no more than ten years. So the note costs $37.1K in the test, not $20.4K. Debt service is $360.6K. Coverage is 1.19x. That clears the old 1.15x minimum. It fails the new 1.25x. Same business, same price, same lender. (In the spring, I wrote that the SBA minimum was 1.25x. It was 1.15x then. On an acquisition, it is 1.25x now.) Three ways out: Cut the price by about $118K, to roughly $2.28M. Bring about $105K more equity, which takes the SBA loan down. Or move the note to full standby. Debt service falls to $323.5K and coverage is 1.33x. The seller collects nothing until the SBA loan is retired, though interest can accrue. That is a real trade, and it only works with a seller who wants the headline number more than the cash. What I changed: I stopped opening models on trailing numbers. I run the last fiscal year first, then the two-year average, with every seller note fully amortized. If it does not clear there, the trailing number is an argument to make to a lender, not a result. The seller note is where this hides, because an interest-only period looks like cushion. When a broker quotes you coverage, ask what earnings it uses and what it assumes about the note. Disclosure: I built Deal14, the underwriting tool I use for my own search, and it runs this exact test. The full example with every number is on its blog: redacted Curious how the lenders here are handling this in practice. If trailing twelve runs well above the last fiscal year, how much of that gap are you willing to carry as an adjustment?