Sixty Percent More Dinners

The auditor's email arrived in February. Travel and entertainment had grown sixty percent year-over-year. Headcount had grown twenty. Could the company explain the discrepancy? It could not — not because the answer was embarrassing, but because nobody in the building knew the discrepancy existed. The category had been growing all year, distributed across hundreds of small expenses, none of which triggered review on its own. The aggregate became visible only when someone put two annual totals side by side.

The investigation turned up nothing scandalous, which is what makes the case instructive. Sales travel had grown because the company added enterprise customers in geographies that required in-person support. Customer dinners had migrated from neighborhood restaurants to high-end ones, venue by venue, without anyone approving the shift as a policy. Two offsites were booked at venues costing twice the prior year's. Recruiting travel grew because the candidates got more senior. Every decision was defensible. (This is the annoying part: expense drift produced by bad actors is easy to fix. Expense drift produced by reasonable people making reasonable local decisions is structural.)

At this company, the total came to roughly $180,000 of spend nobody had planned or attributed. Against the revenue base, manageable. The real cost was the timing: the company learned about the drift at audit, the one moment when the numbers are compared across years and nothing can be adjusted, instead of nine months earlier, when the operational decisions producing it were still live.

The mechanism deserves to be stated precisely, because it governs more than T&E. Each expense is approved by a manager operating in their own decision context. Say a sales manager approves a $400 customer dinner, weighing it against the deal it might advance — a sensible trade. Now say the same manager approves thirty similar dinners over a year: $12,000 of category growth that no one is tracking, because no single approval crossed any threshold. Multiply across managers and functions, and the company acquires a spending trajectory that nobody chose. The failure isn't any approval. It's that approvals were the only control.

The counter-mechanism is aggregate review, and it is small enough to fit on one page. Finance produces a monthly report: each major expense category, current month, trailing three months, year-over-year growth, variance against expected. Any variance above a threshold (ten percent above expected is a common calibration) requires the category owner to explain it at the leadership review. Not defend it in the adversarial sense. Explain it. Half of the explanations will be "the business grew, this grew with it," and those take thirty seconds.

The named owner is the load-bearing piece, the same way an unowned collections process quietly stretches receivables. Marketing owns marketing spend. Engineering owns tooling. Sales owns travel and entertainment. People owns recruiting. An owner who knows the variance question is coming approves differently: not more stingily, more deliberately. The visibility changes behavior before any enforcement does. Companies that run spend without attribution discover the same thing in reverse: what nobody reads, nobody manages.

Why is this discipline rare? Because its cost is visible (a report, a thirty-minute standing review, a few awkward explanations) while its benefit is invisible: the drift that never happened, which appears on no P&L line. Leadership teams systematically buy the visible saving and pay the invisible cost. The audit flag is the invoice arriving.

The one-page report, ready to copy:

• Rows: every expense category above 2% of monthly opex.
• Columns: current month · trailing 3-month average · same month last year · YoY % · expected growth % (headcount or revenue growth, whichever the category should track) · variance vs expected.
• Owner column: one name per row. A function, not a person, is not an owner.
• Trigger: variance > 10 points above expected → owner brings a three-sentence explanation to the monthly review: what grew, why, keep-or-adjust.
• Standing slot: thirty minutes, same week the books close. If it isn't on the calendar, it isn't a control.

Run it for two quarters and the February audit email becomes a formality: the variance is real, and you approved it on purpose.