Marketing Insights Platform

Problem

  • Ingesting marketing data from various data sources
  • Campaign performance and audience segmentation
  • Cross-sell/Upsell, next best action
  • Enabling trend discovery to unlock new revenue streams

Challenges

  • No centralized Data Science platform for Data Scientists
  • Ingesting marketing data from various data sources
  • Visualization of cleaned data in R Studio and Tableau
  • Backfilling of Campaign Manager data into a consolidated data lake

Tools

  • Google Storage

  • Google BigQuery

  • Google Composer

  • Google Dataflow

  • Google Dataprep

  • Google Cloud ML

  • Tableau

  • Shiny - RStudio

Solution

  • Marketing data from multiple data sources were imported into Google Cloud Storage
    • Google Double Click campaign data was transferred to BigQuery using Google BigQuery Transfer Service
    • MOAT Campaign viewability data containing ad impressions and content views were bulk uploaded on a daily basis
    • Audience Segments collected from Oracle Bluekai pixel tags were bulk uploaded on a daily basis
  • Marketing Data Pipeline
    • Data from Marketing data sources were pushed into Google Cloud Storage
    • Source datasets were stored in Google Cloud Coldline Storage to use it only on requirement basis
    • Google Cloud Dataflow transformations were used to enrich the incoming data
    • Google Cloud Dataprep was used to cleanse the data
    • Google Composer was used as a workflow orchestrator to run the marketing data pipelines. Google BigQuery Transfer Service, Google Dataflow jobs and Google Dataprep Jobs constitutes Marketing Data Pipeline and triggered daily by Composer
    • Consolidation of multiple marketing data was done using Google Cloud Dataflow
  • Marketing Datawarehouse
    • Enriched marketing data were stored in Google BigQuery
  • Marketing Data Visualization
    • R Shiny App and Tableau read the enriched data from Google BigQuery for creating marketing insight dashboards