An AI-powered system that connects to a live Google Ads account, pulls weekly campaign data, generates a Claude-written performance analysis, and emails a client-ready report — automatically every Monday morning. Built and tested on a real furniture brand’s live ad account in Hyderabad.
See It In Action
The Solution
An automated Python script connects to the Google Ads API via MCP (Model Context Protocol) inside Claude Code. Every Monday at 8am it runs end-to-end without any human input.
Pulls last 7 days of campaign data — impressions, clicks, CTR, CPC, conversions, ROAS — and compares it to the previous 7 days
Sends the comparison to Claude API with a senior analyst system prompt — Claude identifies the top 3 wins, top 3 problems, and writes 3 specific recommendations for the coming week
Formats the analysis as a professional HTML email with the data table and written summary, and delivers it via Gmail SMTP automatically
The Problem
Every agency and in-house marketing team running Google Ads spends 2–3 hours every week doing the same thing — exporting CSVs, building comparison tables, writing performance summaries, and formatting reports for clients or leadership. It's analytical work disguised as administrative work.
At 10 clients, that's one full-time person doing nothing but reporting. At a single brand, it's half a day every week that could be spent on actual optimisation. The data is already there — the bottleneck is turning it into insight, fast.
The Impact
Project Visual
Tools & Stack