When COGS snaps to a round number, the margin gain isn't real
Gross margin was 68% for five straight quarters. Then it was 74%. Revenue grew the way it always had, headcount in the cost-of-revenue line barely moved, and yet the company suddenly kept six more cents of every dollar. The founder's narrative is operating leverage. The number tells a different story, and the tell is in the cost of revenue itself: it landed on $1.20M, almost exactly, the quarter the margin moved.
Real costs are messy. Hosting bills, support salaries, payment processing, and onboarding labor do not coordinate to produce a clean total. When COGS snaps to a round number, something upstream of the business has changed. Usually it's the definition of COGS.
What an investor checks first
The instinct is to celebrate a margin gain. The discipline is to ask whether the gain came from the company doing the same work for less, or from the company calling the same cost something else.
There are only a few ways gross margin jumps six points in a quarter without a pricing change or a genuine infrastructure shift. A cost that used to sit above the gross margin line moved below it. A cost that was expensed got capitalized. An allocation rule changed so that part of support, or hosting, or implementation now counts as operating expense rather than cost of revenue. Each of these moves margin without touching a single customer or contract.
The clean number is the fingerprint. When a real cost base produces $1.20M, it's a coincidence. When a reclassification produces it, someone picked the line and the rest followed.
The four moves that manufacture margin
Reclassifying support out of COGS. Customer support is the most contested line in SaaS accounting. Move it to G&A and gross margin rises while the company spends exactly what it spent before. Defensible in some models, but it is a presentation choice, not an efficiency.
Capitalizing what used to be expensed. Implementation labor, and increasingly internal software work, can be capitalized and amortized over years instead of hitting the current period. The cash went out the same week. The income statement just doesn't show it in the same place anymore.
Shifting hosting into a different bucket. A committed-use cloud discount or a reserved-capacity prepayment can be spread or deferred in ways that flatter the quarter it lands in. The trendline smooths, but the smoothing is an accrual decision.
Pulling professional services out of the denominator. Bundle low-margin services into a "subscription" number and the blended margin looks like pure software. Strip them out of COGS entirely and it looks even better. Neither move makes the services more profitable.
How to tell leverage from relabeling
The question to put to any margin improvement is simple: did the dollars change, or did the line change?
Operating leverage shows up as a cost base that grows slower than revenue while every category stays in the same place. You can trace it. Revenue up 30%, hosting up 18%, support up 12%, and the mix holds. The improvement is distributed across the cost structure, and it persists, because the underlying economics actually improved.
A reclassification shows up as a single line that drops while another rises by a similar amount in the same period. Gross margin improves; operating margin barely moves, because the cost didn't leave the company, it just left the numerator. The cleanest check is to look one line down. If the savings are real, the bottom of the income statement reflects them. If the savings are cosmetic, the savings disappear by the time you reach operating income.
The timing is the other giveaway. Genuine efficiency arrives gradually as the business scales. An accounting change arrives all at once, on a clean number, in the quarter someone decided to make it.
Why the round number matters more than the margin
None of these moves is necessarily wrong. Capitalizing development can be correct under the accounting standard. Support genuinely belongs in different lines for different business models. The problem is not the reclassification. The problem is presenting it as performance.
When an underwriter spots a six-point margin gain that traces entirely to a moved line, the read is rarely "they cheated." It's "they don't know which of their numbers are operational and which are accounting." That doubt then spreads to every other metric in the deck, because if the headline margin is a definition choice, what else is.
A founder who flags the change first keeps control of the story. "We moved support to G&A this quarter to match how we manage the team; gross margin on a like-for-like basis is up two points, not six." That sentence costs nothing and converts a red flag into a sign of rigor.
Levian computes your gross margin and operating margin from the P&L and trends both over time. When a margin gain at the gross line doesn't survive to operating income, the two charts make it visible: the improvement that looked like leverage on one line and disappeared by the next. You see the same gap an underwriter would find, and you see it first.