Anonymised, illustrative composite. A perfectly healthy seasonal business tripped its own debt-service covenant because the covenant was measured on the wrong window, not because anything had actually gone wrong.
At a glance
A buyer acquired a Nova Scotia marina and boat-storage operator — a genuinely seasonal business, busiest May through September and largely dormant December through February — financed with a CSBFP-backed term loan carrying scheduled annual debt service of $164,000. Trailing-twelve-month seller's discretionary earnings at closing stood at $205,000, a 1.25× coverage ratio. See the debt service coverage ratio glossary entry for how this ratio is normally read.
The lender's covenant package, drafted from a template built for a non-seasonal business, tested debt-service coverage against “the most recently completed month, annualised” — taking one month's cash flow and multiplying by twelve, rather than looking at trailing twelve actual months. In February, the marina's dead month, discretionary earnings ran roughly $6,000 for the month; annualised, that implied $72,000 for the year against $164,000 of scheduled debt service — a 0.44× ratio and a technical breach, on a business whose actual trailing-twelve-month coverage had not moved from the 1.25× it closed at.
The lender's own credit policy, informed by the kind of coverage benchmarks deavo's disclaimed illustrative bands describe — a DSCR floor of roughly 1.25× measured on seller's discretionary earnings for a business of this size, with the underwriting basis typically shifting from SDE to EBITDA around the $1 million revenue mark — had set the covenant at 1.25×, tested monthly. See deavo's financing page for the illustrative bands themselves, reproduced here with the same disclaimer deavo attaches to them: not a benchmark, illustrative only.
$205,000 ÷ $164,000 = 1.25× on a trailing-twelve-month basis — passing, by design. $72,000 ÷ $164,000 = 0.44× on the single annualised month — failing, by an accident of measurement window rather than by anything the business had done.
Nothing in tax or corporate law governs how a credit agreement defines its own covenant-testing period; that is purely a matter of contract, and it is where this file actually turned. A covenant that annualises a single month treats every month as if the business generated that month's cash flow all year — a reasonable proxy for a business with flat monthly revenue, and a false signal for one that is not. The fix was never the business's performance; it was the measurement window the credit agreement had chosen.
Before the first testing date, the buyer's advisor took the trailing-twelve-month numbers to the lender and asked for the covenant to be re-based to a trailing-twelve-month test, consistent with how the loan had actually been underwritten. The lender agreed, amending the credit agreement before any test date had passed and before any default notice had gone out. The facility's pricing and amortization were untouched — only the measurement window changed.
For a related closing-stage financing condition on a different file, see financing conditional on a client contract renewal, and for the covenant concept itself, the financial covenant glossary entry.
Left uncorrected, the February test would have generated a technical default notice on a facility that was never actually in financial distress — triggering, at minimum, a formal waiver or forbearance negotiation, legal fees on both sides, and a black mark in the loan file that could complicate any future request to increase or refinance the facility. All of that avoidable cost would have traced back to one undefined phrase in the covenant, not to one dollar of underperformance in the business.
The credit agreement's covenant clause never used the words “trailing twelve months” anywhere — only “the most recently completed month, annualised.” For a non-seasonal business the two produce nearly identical results and the gap is invisible until the first seasonal business runs the same template through the same clause. Reading the covenant's own measurement-period definition against the target's monthly revenue pattern, before signing, would have caught it before the first trough month arrived.
A 30-minute call is enough to tell you whether AI pays for itself here.
