The dashboard said 37x. We told the owner to trust her books instead.
The goal
More online orders and subscriptions for a neighborhood flower brand on a modest budget.
What we found
Early reports showed returns as high as 37x on one campaign. Then we found a purchase tag firing twice: Google was reporting about double the revenue the store actually saw. A “checkout” event was also set as a primary optimization goal, so the campaigns were chasing the wrong thing. Later, a Display leak spent budget on 10,332 impressions that produced no orders.
What we changed
Fixed the goals, closed the Display leak (10,332 wasted impressions in July, 228 in August), moved budget toward the search terms and landing pages that actually converted, and started reporting the store’s numbers next to the platform’s.
What happened
When a report line showed more revenue than the owner saw in her own books, we wrote back: trust your books, not our report line. Her own numbers tell the story we’re comfortable with: subscriptions went from 3 in February 2025 to 10 in February 2026. In her words, “Our subscriptions have increased compared to last year.”
What happened next
August 2026 was a better month for the business, and paid ads weren’t the reason; organic search was. We said so, and moved the plan accordingly.