Every account below was already “working” when it landed on our desk. The problem was never obvious — it was buried in the tracking, the structure, or a setting someone changed six weeks ago.
The client believed the account was healthy — 318 conversions at roughly $10 CPA.
96% of those “conversions” were phone-number button taps, not connected calls. Real traceable leads were 12 over sixteen months — a true CPA closer to $263. The algorithm was being trained on a fake signal.
Demoted the button-tap to a secondary signal, repaired Enhanced Conversions, and re-pointed the account toward Search — the only channel actually producing form fills.
Bidding re-anchored to real leads instead of accidental taps.
New account, no live history — but the prior account had wasted budget splitting into competing campaigns and spending evenly across the week.
The old data showed weekends converted at 13.6% vs 5.4% on weekdays, at roughly $293 per lead vs $507 midweek. Same money, nearly double the return on the right days.
Built one consolidated Search campaign weighted toward weekends, structured around emergency-intent terms, with a full negative list carried over on day one.
Launched clean — budget pointed at the hours that convert; burst-pipe leads at about $33 each.
The client complained of low-quality, irrelevant leads.
Core keywords had been paused, forcing Dynamic Search Ads to carry 66% of Search spend with no bid control — firing on competitor names, the brand, and out-of-area suburbs. CPA drifted from $79 to $125.
Unpaused core keywords, tightened DSA URL scope, layered negatives, and corrected the conversion hierarchy so bidding chased real leads.
DSA reined in, spend refocused on in-area intent.
A 64% conversion drop month-on-month, blamed on demand.
Two causes — every sitelink had been paused during a “clean-up,” and the primary booking event was firing on a third-party domain the account could not see. That platform was already sending real revenue ($3,017 in tracked bookings, ~$131 avg) Google Ads never saw.
Re-enabled sitelinks, imported the real revenue event as the primary conversion, and set up cross-domain measurement so attribution stopped breaking.
Conversion tracking and attribution restored in about eighteen days.
Account spending with almost nothing to show — 1.16× blended ROAS and zero conversions across the final two weeks.
The Shopping campaign had spent $469 and generated 38 add-to-carts but zero purchases. A live audit surfaced nine checkout blockers. The profitable Brand campaign ran at 11× ROAS but was capped at $2/day.
Documented the checkout fixes for their dev team, reallocated budget to the 11× brand campaign, added mobile bid adjustments, and rebuilt the product feed with proper Google taxonomy.
Shopping impression share recovered within the first week after the feed was resubmitted; the real blocker was isolated and handed off.
The client wanted more phone calls; spend “looked fine.”
$3,166 burned across 403 zero-conversion search terms in thirty days, 36.5% of impressions lost to Ad Rank (not budget), and Performance Max eating about 20% of budget to produce a single call.
Built the negative-keyword layer, promoted the converting terms to exact match, fixed call-conversion tracking, and added mobile bid adjustments where 83% of demand sat.
Spend redirected from junk terms to intent that converts.