Serverless Architecture for Personalised Marketing App

This is an architecture of Personalised Marketing Application which enables users to offer an individualised experience for every current or prospective customer.

Courtesy: Microsoft Azure

Platform: Microsoft Azure


Twitter: @Azure

Architecture Overview

  • Events Hub ingests raw click-stream data from Functions and passes it to the Stream Analytics.
  • Stream Analytics aggregates clicks in near real-time by product, offer and user to write to Azure Cosmos DB and also archives raw click-stream data to Azure Storage.
  • Azure Storage stores archived raw click-stream data from Stream Analytics.
  • Azure Functions takes in user click-stream data from the website and reads existing user history from Azure Cosmos DB. These data then run through the Machine Learning web service or used along with the cold-start data in Redis Cache to obtain product-affinity scores. Product-affinity scores are used with the personalised-offer logic to determine the most relevant offer to present to the user.
  • Machine Learning helps you easily design, test, operationalise and manage predictive analytics solutions in the cloud.
  • Redis Cache stores precomputed cold-start product affinity scores for users without history.
  • Power BI visualises user activity data as well as offers presented by reading in data from Cosmos DB.