Google Ads judged on dresses sold, not conversions reported

The problem

A wedding dress is a considered, multi-thousand-dollar purchase with a months-long decision cycle. Ad platform conversion columns count form fills, lag by weeks, and cheerfully take credit for people who were always going to walk in. Scale on those numbers and you will scale the wrong thing.

What we did

We built the measurement before we trusted the spend. Everything the ads touch is ours — the website the clicks land on, the booking platform behind it, and the CRM the orders live in — so every ad click ID is stored against the booking it produces and joined to actual dress orders. Then we hold the account to that number instead of the dashboard’s.

Two findings that paid for the engagement

• A 17× overpay hiding in plain sight. A branded ad group was buying clicks on the studio’s own name at $4.57 while the dedicated brand campaign bought the same clicks at $0.27. Fixing the cannibalization recovered about $350 a month that was being spent to compete with ourselves.
• More budget produced fewer bookings. A 19-day stretch at the raised budget delivered roughly 850 paid sessions and zero bookings. We cut the daily budget back and got 5 bookings in 13 days from half the traffic. Spending more was actively buying worse traffic.
• The gap every platform hides: a bride who sees the ad, phones, and walks in leaves no click ID at all. Call tracking exists precisely because the dashboard cannot see the channel’s best work.
• What we refuse to do: an early read on this account showed a headline 6.9x return. It did not reproduce the following month, so we retired it. The honest number is closer to break-even at current margins — and knowing that is worth more than a flattering slide.

Outcome

An account where the spend is defensible line by line, the waste got named out loud, and the scale decision waits for order-level evidence instead of a conversion column.