Spending when the shelf runs out, not when the month begins

Industrial consumables are not bought when an advertisement is seen. They are bought when the stockroom runs low, which for this brand's customers happened on a rhythm of five to nine weeks depending on the plant. Demand therefore arrived in waves that repeated per account and, aggregated across thousands of accounts, produced a visible weekly and monthly shape.

The advertising ignored all of it. A flat daily budget spread the same money across every day of the month, which meant paying full price for clicks on days when nobody in the customer base needed anything, then capping out on the days when they all did. The account looked stable and was quietly buying its worst traffic at the same rate as its best.

We treated the reorder interval as the planning unit instead of the calendar month.

• Consumption curve reconstructed from three years of order dates, per SKU and per order size, to find where the waves actually fall.
• Budgets shaped to the curve — daily caps raised sharply inside reorder windows and pulled back between them, rather than averaged flat.
• Scheduled budget rules by day of week, since procurement concentrates Monday to Thursday — search advertising has no hour-of-day bid control, so the pacing is done with the budget lever that does exist.
• Separate campaigns for first-time and returning demand, so the expensive discovery traffic is not paced against the cheap replenishment traffic.

ACoS fell from 39% to 26% while revenue grew 67%, because the same budget now lands on the days when a purchase is actually imminent.

Wasted spend on zero-conversion days dropped by roughly $22k a month, and repeat-order share rose twenty-seven points — the pacing keeps the brand visible precisely when an existing customer is deciding whether to reorder or to try whatever is cheapest that week. Nothing about the listings, the products or the prices changed.$