E-Commerce
AI & Personalisation
2025
7 min read
AI-Powered Customer Engagement Analytics: 22% Conversion Increase with Power BI & OpenAI GPT
Industry
E-Commerce
Challenge
Descriptive Analytics · Manual Campaigns
Platform
Power BI · Azure ML · OpenAI GPT
Result
22% Conversion Increase
A major e-commerce company had rich transactional data and a mature Power BI implementation, but all insights were backward-looking, and marketing teams were creating campaigns manually. Numlytics built an AI-powered customer engagement analytics platform combining Azure Machine Learning for churn prediction, Power BI integration, and OpenAI GPT for automated personalised content, delivering a 22% conversion increase and 15% churn reduction.
The Challenge: Descriptive Analytics, Manual Campaigns
Despite mature Power BI reporting, the client's analytics capability was entirely retrospective. There was no personalised marketing automation Power BI layer, no predictive churn model, and no mechanism to personalise content at scale. Every campaign was produced manually.
-
No predictive capability: Power BI dashboards showed past behaviour only, no purchase intent prediction, no churn prediction Power BI Azure ML model existed
-
Manual campaign creation: Marketing teams hand-crafted product recommendations and email campaigns, slow, generic, and ineffective at individual-level personalisation
-
No personalisation at scale: Customer segmentation existed, but was too broad for meaningful personalisation; the same message was sent to millions of customers
-
Reactive churn management: Customers were only flagged as at-risk after they had already disengaged, too late for effective intervention.
The Numlytics Solution:
Numlytics built a six-component architecture connecting Power BI's visualisation strengths with Azure ML's predictive capability and OpenAI GPT's GPT marketing content generation, creating a system that both predicts customer behaviour and automatically acts on those predictions.
-
Power BI Customer Data Layer
Purchase history, web traffic, and engagement data are ingested from CRM, Shopify, and Google Analytics. Custom DAX metrics for customer segmentation, lifetime value, and purchase patterns - the data foundation for all AI-powered customer engagement analytics.
-
Azure Synapse Data Integration
ETL pipelines extract data from all systems into a structured format. Azure Data Lake handles both structured and unstructured data at scale, feeding both Power BI and Azure ML models.
-
Azure ML Predictive Models
Purchase intent and churn prediction Power BI Azure ML models built on historical customer data. Models integrated directly into Power BI, predictive scores visible alongside existing behaviour analytics in the same dashboard.
-
Power BI OpenAI GPT Integration
Numlytics built a Power BI OpenAI integration connecting customer segments from Power BI to OpenAI GPT via a custom API - enabling GPT to generate personalised product recommendations, email subject lines, and marketing copy at scale.
-
Azure Cognitive Search Enrichment
Customer data indexed for improved search and categorisation - improving the relevance of GPT marketing content generation by linking it to product catalogues and individual purchase history.
-
Azure Logic Apps Automation
Content distribution automated: GPT-generated campaigns sent via Mailchimp and Salesforce without manual intervention - enabling personalised marketing automation, Power BI at scale.
The Results
22%
Conversion Rate ↑
AI-targeted personalised product recommendations vs generic campaigns
18%
Click-Through Rate ↑
GPT-generated personalised email content vs manual campaigns
15%
Churn Reduction
Azure ML predictions enabled proactive at-risk customer engagement
40%
Campaign Time ↓
Automated GPT content generation replaced manual campaign production.