Case studyAnonymised client
The result
−72%
Cost per lead
+54%
Conversion rate
+306%
Conversions
Rebuilding a Google Ads account that couldn’t tell students from customers
Atom Digital rebuilt a grooming education provider’s Google Ads account to separate genuine course enquiries from a flood of broad-match traffic searching for grooming services, cutting cost per lead while conversions tripled.
The outcome
Cost per lead fell 72%, conversion rate rose 54%, and total conversions were up 306% over the period. The turnaround is visible from around April 2025 onward.
Evidence
Cost per conversion fell as conversions climbed
Google Ads — conversions and cost per conversion
Weekly conversions (red, left axis) against cost per conversion (amber, right axis), 30 December 2024 to 28 July 2025. Cost per conversion spikes early in the period, then falls away and stays low, while weekly conversions climb and hold. Across the full period: 905 conversions at £19.20 each from £17.4K of spend. The turn is visible from around April 2025. View at full size.
What the rebuild involved
- Data forensics pass through the search term reports
- Negative keyword architecture to exclude service-seekers
- Intent-based audience layering to isolate course-seekers
- Landing page rebuilt to match the targeting
- Conversion path rebuilt to reduce friction
How it was done
The situation
The business trains people to become professional dog groomers, filling classroom-based courses through paid search. Demand for the courses was strong, but the cost of turning that demand into enrolled students kept climbing.
Two searchers, almost identical language
Broad match targeting can’t distinguish between two searchers who use almost identical language: someone looking to have their dog groomed, and someone looking to become a dog groomer. Both search close to “dog grooming”, and a meaningful share of the account’s budget was going to the wrong one: pet owners after a grooming service, not a career.
Diagnosis and decisions
A data forensics pass through the search term reports surfaced the leak directly: a consistent share of spend was landing on service-seeker queries that would never convert into a student. The decision was to stop treating this as a keyword-tuning problem and rebuild the targeting architecture around intent, not just relevance.
What Atom Digital built
Atom Digital built a negative keyword architecture and intent-based audience layering to isolate genuine course-seekers, then rebuilt the landing page and conversion path to match — reducing friction and aligning the message with what a prospective student actually needed to know.