Who Audits the Verdict the Machine Produced?
My own Google Ads engine compared a bid step against the wrong window. Google's auto-apply rewrote a target CPA before dawn. Both had a rationale. Neither had evidence.
SEM · 6 min read
Growth · UA · Google Ads · Meta Ads · Data
GrowthMadre is my open notebook on growth, user acquisition, ads and data. I publish analyses, A/B tests and case notes from real accounts. Reading is free; the door is open if you want to work together.
₺304 ₺131
Cost per lead, in 4 weeks, a 57% drop
Diving center in Kaş, Türkiye
₺1,142 ₺148
Cost per appointment request, period averages, a 7.7x improvement
Psychiatry clinic in İzmir, Türkiye
₺910 → 1 inquiry
One keyword's bill, caught and paused
Anonymous account audit
Business names are withheld for client confidentiality; the numbers are real account data.
My own Google Ads engine compared a bid step against the wrong window. Google's auto-apply rewrote a target CPA before dawn. Both had a rationale. Neither had evidence.
SEM · 6 min read
Language targeting is going away too. As the control levers close, what actually gains value? Four real cases: measurement accuracy, causal discipline, and keeping ads aligned with business reality.
SEM · 10 min read
A psychiatry practice's Google Ads account was paying ₺1,142 per appointment request. The search terms report showed 40% of the budget flowing into two classes of searches that never converted. After cleanup and restructuring, the cost dropped to ₺148 — a step-by-step account of how.
SEM · 5 min read
Half of the Display campaign's conversions were not human. How I caught and cut fake clicks with GA4 minute-level forensics, a placement blacklist and session-age gated conversions.
SEM · 9 min read
How I recovered traffic lost to Google's AI content update using a writing style skill file built from Dan Brown's prose — analysis, file structure, pipeline.
SEO · 6 min read
I consult for local service businesses on Google Ads management, measurement and conversion setup, and conversion-focused websites. We fix measurement first; decisions only become right after that.