AI for data · Paid media
Campaign monitoring with an AI analyst
Code decides whether there's a problem; the AI only explains and prioritizes, and the alert lands on WhatsApp.
- Status
- Template replicated across clients
- Role
- Template design and development
- Stack
- n8n
- Google Ads API
- Meta Graph API
- OpenAI
- Google Sheets
- Evolution API
Results
- clients
- 10+
- steps
- 6
- step where healthy campaigns stop
- 3
Problem
Checking the indicators of dozens of campaigns every week is repetitive, and problems slip by.
Solution
It pulls the last 7 days of metrics, applies rules per campaign type against targets calculated from the account's own history, and only calls the AI when there's an alert. The analyst uses a calculator as a tool and sends the diagnosis on WhatsApp.
How it works
01Trigger
Weekly trigger
Pulls the last 7 days of each account.
02Process
Campaign data
Google Ads API (and Meta Graph API in the Meta variant).
03Process
Rules by type
Search, PMax and the rest against targets from the account's history.
04Output
Healthy
Campaigns with no alert stop at step 3: no noise.
05AI
AI analyst
Explains and prioritizes using a calculator tool and a strict script.
06Output
WhatsApp alert
Diagnosis for the team; history in Google Sheets.
Inside the system
Monitoring template (fictitious data)
Technical decisions
- Code decides, AI explains. Rules determine whether there's a problem; the AI only explains and prioritizes.
- The account's own targets. Thresholds come from each account's history, per campaign type (Search, PMax and the rest).
- Calculated numbers. The analyst uses a calculator tool: no number is estimated by the model.
- No noise. The AI only runs when there's an alert; healthy campaigns stop at step 3.
- Meta Ads variant. Catches active campaigns with no delivery.
Result
The team only gets what needs action, with calculated numbers and a weekly history to compare trends.