A SaaS in the Aesthetic Medical Practice space

Cut cost per qualified lead 56% in a three-month test, by rebuilding the account around one clean conversion signal.

What was broken

The account was already profitable on Google. That was the problem. The account made money, but it would not scale, and the client suspected there was room they were not reaching. They hired us as the more expensive option to a marketing team that had taken the account as far as it could, and they gave us three months to prove a higher fee could earn its keep.

Under the hood, the account was working against itself. Targeting was layered so tightly that it strangled volume. The Search Partner Network was on, spending into placements that did not convert. Every keyword ran on exact or phrase match, capturing only the searches we already knew about. And the conversion tracking was a mess: more than six core conversion events firing into different campaigns, some of them duplicates, some firing in the wrong place. The algorithm was being handed contradictory signals and pushed budget around chasing them.

What we did

Step 01Removed the over-targeting and the Search Partner Network.

We stripped the targeting layers back and turned off the Search Partner Network in the first weeks. Leads started pouring in almost immediately. What we did not fully expect: the extra signal volume also pulled cost per lead down hard. More leads and cheaper leads, from the same spend.

Step 02Tested into expansion with broad match.

The account was built entirely on exact and phrase match. We knew broad match would open volume we could not see, we just did not know how much. We launched a broad campaign and started scaling budget by percentage as each change proved out. Broad runs wide at first, so we fed it negatives continuously to teach it what a real qualified lead looks like.

Step 03Reset the conversion signal to one direction.

In the middle of warming up the broad campaign, the data started swinging: some days zero conversions, some days far too many. Something was off. We audited the entire conversion structure and found the six-plus events were pulling the algorithm in different directions, some duplicated, some firing in the wrong place. We made the call to reset everything to a single conversion goal. It caused a short dip, and then the best month on record for qualified leads, cost per qualified lead, and revenue.

Early on, Google was reporting four to six times the leads the client's CRM actually received. The algorithm was optimizing toward a number that was mostly noise. As we consolidated to one clean conversion event, the reported count fell back toward the truth. By May the two lines nearly met: 65 reported, 47 real.

When we removed the duplicate event firings, results looked at first like they fell. They had not. The inflated numbers simply dropped to match what the backend was actually recording. After a two to three week recovery period, the account posted a record month for leads, conversion rate, and revenue. Stable signal is the infrastructure that carries a bigger budget without the account falling apart.

Three months in, the search terms are aligned with the business, and the new broad campaign is holding only about 23% search impression share, roughly half what the old exact campaign carried. That gap is not a problem, it is runway. The client is now scaling budgets deliberately, on infrastructure that can carry them, and posting record months.

What happened

Across the three-month test, cost per lead fell 42% and cost per qualified lead fell 56%, with qualified leads running about 3.8 times the prior pace. Because our fee sits inside the ad spend, that efficiency is measured against a higher cost base than the team before us, not a lower one. The qualified pipeline carried through to the backend: May was the record month, with 47 leads and 9 closed deals, the most of the engagement.