nBrain

Ads Intelligence Platform

AI-Powered Marketing Analytics with Industry Context Integration

Production Live Version 1.0 October 21, 2025

Table of Contents

Executive Summary

Unified Marketing Analytics with AI Intelligence

Ads Intelligence Platform is an enterprise-grade analytics and intelligence system that combines Google Analytics 4 (GA4), Meta Ads, and Google Ads data with real-time industry insights. The platform provides AI-powered marketing intelligence, enabling marketing teams to understand their digital performance in the context of broader market trends with actionable recommendations powered by GPT-4.

3
Data Sources Unified
GPT-4
AI Analysis
RAG
Industry Context
Multi
Property Support

Key Value Propositions

DA

Unified Dashboard

Single interface for GA4, Meta Ads, and Google Ads data

AI

AI-Powered Insights

GPT-4-driven analysis connecting metrics to causation

IC

Industry Context

RAG-powered correlation with market trends

Platform Overview

Intelligent Marketing Analytics & Industry Insights

What It Does

  • ✓ Unified Analytics - GA4, Meta Ads, Google Ads in one dashboard
  • ✓ Natural Language Queries - Ask questions in plain English
  • ✓ Deep-Dive Analysis - Automatic correlation of related metrics
  • ✓ Industry Context - RAG-powered news correlation
  • ✓ Multi-Property Management - Portfolio view across properties
  • ✓ Streaming Responses - Real-time AI analysis delivery

Intent Classification

GPT-4 automatically classifies and routes queries:

Analytics Only
"Show me my users last month"
Industry Only
"What's new in the market?"
Hybrid
"Why did my traffic drop?"

Complete Technology Stack

Python Flask with AI & Vector Search

Backend

Python 3.11.9
Flask Web Framework
SQLAlchemy ORM
PostgreSQL 14
Gunicorn WSGI Server

AI & Data

GPT-4 Turbo Analysis
text-embed-3-small Vectors
Pinecone Vector DB
AWS S3 Data Lake
Pandas Analysis

Integrations

Google Analytics 4 API
Meta Ads CSV Import
Google Ads CSV Import
MCP Server Protocol
RSS Feeds Industry Data

System Architecture

Multi-Layer Architecture with RAG Integration

┌────────────────────────────────────────────────────┐ │ Flask Application Layer │ │ ┌────────────┐ ┌────────────┐ ┌────────────┐ │ │ │ Routes │ │ Agents │ │ Analysis │ │ │ │ (main/) │ │ (AI Logic) │ │ (insights) │ │ │ └────────────┘ └────────────┘ └────────────┘ │ └────────────────────┬───────────────────────────────┘ │ ┌────────────┴────────────┐ ▼ ▼ ┌──────────────────────┐ ┌──────────────────────┐ │ Data Integration │ │ AI Intelligence │ │ ┌────────┐ ┌──────┐│ │ ┌──────┐ ┌────────┐│ │ │ MCP │ │ S3 ││ │ │GPT-4 │ │Pinecone││ │ │ Server │ │Client││ │ │Turbo │ │ Vector ││ │ └────────┘ └──────┘│ │ └──────┘ └────────┘│ │ ┌────────┐ ┌──────┐│ │ RAG Server │ │ │ Athena │ │ DB ││ │ (Industry News) │ │ └────────┘ └──────┘│ └──────────────────────┘ └──────────────────────┘ │ ▼ ┌──────────────────────────────────────────────────┐ │ Data Sources │ │ ┌──────┐ ┌──────┐ ┌────────┐ ┌────────────┐ │ │ │ GA4 │ │ Meta │ │ Google │ │ Industry │ │ │ │ API │ │ Ads │ │ Ads │ │ RSS │ │ │ └──────┘ └──────┘ └────────┘ └────────────┘ │ └──────────────────────────────────────────────────┘

Data Flow

1. User Query
Natural language question
2. GPT-4 Parser
Extract metrics, dates, intent
3. Data Retrieval
MCP/S3/Athena/PostgreSQL fallback
4. Industry Context (Optional)
Pinecone vector search for news
5. AI Synthesis
GPT-4 generates insights & recommendations
6. Streaming Response
Server-Sent Events to client

Core Features

Comprehensive Marketing Intelligence Tools

Unified Analytics Dashboard

Single interface combining Google Analytics 4, Meta Ads, and Google Ads performance data.

  • GA4 metrics (users, sessions, conversions)
  • Meta Ads (spend, clicks, CTR, CPM)
  • Google Ads (spend, leads, impressions)
  • Cross-platform comparison
  • Multi-property support
Production Live

Natural Language Queries

Ask questions in plain English and get structured insights with AI-powered analysis.

  • GPT-4 query parsing
  • Metric synonym mapping
  • Relative date handling
  • Intent classification
  • Streaming responses
Active

Deep-Dive Analysis

Automatic correlation of related metrics to explain performance changes with root cause analysis.

  • Related metrics mapping
  • Ad campaign correlation
  • Efficiency calculations
  • Business impact assessment
  • Actionable recommendations
Operational

Industry Context Integration

RAG-powered system correlating performance data with real-time market trends and news.

  • RSS feed monitoring
  • Pinecone vector search
  • Semantic article matching
  • Trend correlation analysis
  • Market context synthesis
Live

Multi-Property Management

Manage and analyze multiple properties across all platforms with unified interface.

  • Property selector
  • Cross-property comparison
  • Portfolio analytics
  • Performance ranking
  • Aggregated views
Active

Data Lake Architecture

Multiple data access patterns with MCP, S3 CSV, Athena, and PostgreSQL fallback.

  • MCP Server integration
  • S3 CSV direct queries
  • AWS Athena support
  • PostgreSQL fallback
  • Flexible data routing
Operational

AI & RAG System

GPT-4 Analysis with Industry Knowledge Base

GPT-4 Turbo Integration

Advanced AI for query parsing and insight generation

Capabilities:
  • ✓ Natural language parsing
  • ✓ Intent classification
  • ✓ Metric extraction
  • ✓ Date range inference
  • ✓ Insight synthesis
  • ✓ Recommendation generation
User Query ↓ GPT-4 Parser • Extract metrics • Determine dates • Classify intent ↓ Data Retrieval (4 fallback layers) ↓ GPT-4 Synthesis • Calculate insights • Root cause analysis • Recommendations ↓ Streaming Response (SSE)

RAG Industry Knowledge Base

Pinecone-powered semantic search for market trends

Configuration:
• Index: adsintell-industry-vector
• Dimensions: 1536
• Metric: Cosine similarity
• Top-K: 2 matches
Data Sources:
  • ✓ Industry RSS feeds
  • ✓ Market research articles
  • ✓ News embeddings
  • ✓ Trend analysis

Supported Metrics

GA4 Metrics

  • • totalUsers
  • • sessions
  • • engagementRate
  • • conversions
  • • bounceRate
  • • avgSessionDuration

Meta Ads Metrics

  • • meta_ad_spend
  • • meta_ad_clicks
  • • meta_ad_impressions
  • • meta_ad_ctr
  • • meta_ad_cpm
  • • meta_ad_conversions

Google Ads Metrics

  • • google_ad_spend
  • • google_ad_clicks
  • • google_ad_impressions
  • • google_ad_ctr
  • • google_ad_cpm
  • • google_ad_leads

Database Schema

PostgreSQL with SQLAlchemy ORM

ga4_data - Analytics Metrics

Google Analytics 4 performance data

• property_id, date
• metric_name, metric_value
• Indexed on (property_id, date)

meta_campaign_data - Meta Ads

Facebook/Instagram advertising data

• campaign_name, account_id
• cost, clicks, impressions, conversions
• ctr, cpm, cost_per_conversion

google_ads_data - Google Ads

Google advertising campaign data

• customer_id, date
• cost, link_clicks, impressions
• website_leads, ctr, cpm, cpc

chat_history - Conversations

User query and response history

• user_id, session_id
• query, response
• timestamp tracking

Security & Authentication

Enterprise-Grade Security

Authentication

  • ✓ Flask-Login sessions
  • ✓ Bcrypt password hashing
  • ✓ Secure session cookies
  • ✓ Persistent login support

Content Security

  • ✓ CSP headers
  • ✓ XSS protection
  • ✓ Code injection prevention
  • ✓ Sanitized outputs

Data Security

  • ✓ HTTPS/TLS enforced
  • ✓ Encrypted at rest
  • ✓ SQL injection prevention
  • ✓ User data isolation

Deployment Infrastructure

Cloud Platform with Managed PostgreSQL

Web Service

Live Python Flask
Runtime: Python 3.11.9 Framework: Flask + Gunicorn Build: pip install -r requirements.txt Migrate: flask db upgrade Start: gunicorn --timeout 120 Region: Oregon (us-west)

Database

Live Managed PostgreSQL
Database: PostgreSQL 14 ORM: SQLAlchemy Backups: Daily automatic Recovery: 7-day point-in-time Connections: 25 max concurrent SSL: TLS 1.3 encrypted

Data Access Layers

MCP Server
Optimized protocol
S3 CSV Direct
No DB overhead
AWS Athena
Serverless SQL
PostgreSQL
Reliable fallback

Real-World Use Cases

Solving Marketing Intelligence Challenges

Use Case: Performance Review

Scenario: Monthly marketing performance analysis

Query Submission
"Show me all metrics for last month vs prior"
GPT-4 Parsing
Extract metrics, dates, classify intent
Data Retrieval
Query GA4, Meta, Google Ads data
Comparison Analysis
Calculate percentage changes
AI Synthesis
Generate insights table with recommendations
Sample Results:
• Users: +10.9% (market growth)
• Sessions: +10.6% (aligned with users)
• Conversions: -5.4% (quality issue detected)
• Ad Spend: +7.1% (budget increase)

Use Case: Deep Dive Analysis

Scenario: Understanding conversion drop root cause

Analysis Process:
✓ Click "Deep Dive" on conversions metric
✓ System queries related metrics (bounce, engagement)
✓ Analyzes ad campaign changes
✓ GPT-4 synthesizes root cause
✓ Generates specific recommendations
Example Insight:
"Traffic increased 10.9% but conversions dropped 5.4% - indicating a quality issue. Bounce rate up 15% and engagement down 18% suggest new traffic from Meta campaigns is less qualified."

Use Case: Industry Trend Correlation

Scenario: Understanding market-driven performance changes

Hybrid Query
"Why did my traffic drop last week?"
Analytics Data
Total users decreased 22%
Vector Search
Pinecone finds industry news (mortgage rates)
Correlation Analysis
Links 22% drop to market slowdown
Recommendations
Adjust messaging, increase retargeting

Use Case: Multi-Property Portfolio

Scenario: Managing 15 properties across platforms

Workflow:
✓ Select "All Properties" aggregated view
✓ Ask: "Which properties had best ROI?"
✓ System calculates cost per conversion
✓ Ranks properties by efficiency
✓ Identifies underperformers
✓ Provides specific improvement actions
15
Properties Managed
Instant
Portfolio Analysis

Platform Metrics

Performance & Capabilities

Performance

Query Response <500ms
Database Query ~12ms avg
Vector Search ~200ms
Uptime 99.9%

Data Sources

  • Google Analytics 4 API
  • Meta Ads (CSV/API)
  • Google Ads (CSV/API)
  • Industry RSS Feeds

Capabilities

  • Multi-property support
  • Natural language queries
  • Streaming AI responses
  • 4-layer data fallback

Platform Statistics

Metric Value
Database Tables7 core tables with indexes
Supported Metrics18+ metrics across 3 platforms
Query Methods4 data access patterns (MCP, S3, Athena, PostgreSQL)
Vector IndexPinecone 1536-dim embeddings
AI ModelsGPT-4 Turbo + text-embedding-3-small
Response DeliveryServer-Sent Events (streaming)
SecurityFlask-Login + Bcrypt + CSP headers

Key Takeaways

What Makes This Platform Unique

Technology Highlights

  • ✓ GPT-4 Turbo for natural language queries
  • ✓ RAG architecture with Pinecone vectors
  • ✓ Multi-layer data access (MCP/S3/Athena/DB)
  • ✓ Flask + SQLAlchemy Python backend
  • ✓ Streaming responses with SSE
  • ✓ Intent classification for smart routing

Business Value

  • ✓ Unified view of GA4, Meta, Google Ads
  • ✓ Natural language interface (no SQL needed)
  • ✓ Deep-dive automatic correlation analysis
  • ✓ Industry context for market-driven insights
  • ✓ Multi-property portfolio management
  • ✓ Actionable AI-generated recommendations

Platform Status

Production Live Version 1.0 October 21, 2025

Project Details

Platform: Ads Intelligence Platform
Industry: Marketing Analytics
Type: Enterprise SaaS
Project Scope: AI-powered marketing intelligence platform
Deployment: Cloud-hosted Flask application
Technologies: Python, Flask, GPT-4, Pinecone, PostgreSQL, GA4/Meta/Google Ads
Ads Intelligence Platform
Unified Marketing Analytics with AI-Powered Insights
Last Updated: October 21, 2025 | Version 1.0