Audience Target Analysis: In-App Marketing

Introduction

Challenge:

A marketing team at a client has a large daily volume of promotions that have specific audience targets. Promotions are sent via their apps as push notifications to users from a pool of millions across the globe. Each is specifically tailored to a unique group of users based on age, country, state, city, and browsing and purchase behavior on their application. The client has requested a team to analyze user data to deliver approximately 20 curated audience targets daily in the form of UserID lists for them to send their promotions

Objective:

Continuously analyze and output user demographics and behavior for high volume output of ad hoc requested target audiences for promotions on the client's four applications used worldwide.

Execution

Databricks Analysis for Audience Targets:

Every day our team of four would receive about 20 new audience requests. The application data lived on Amazon S3 Storage and I used Databricks as my interface to locate folders/tables where relevant data lived and to write and execute complex Spark SQL queries to output a large list of users catered to who the marketing team wanted to receive their promotions. I often needed to query big datasets from recordings of user actions on the apps, browsing history, running distance, workout frequency. This enabled the marketing team to send it's promotions to people who were most likely to act on them. 

The stakeholders sending audience requests worked from many different countries - The Netherlands, Brazil, Mexico, China, Canada and the US. Time zone differences presented challenges with communication so I had to be very clear with my questions and explanations in emails that would be read by marketing team members overnight.

Impact

User Engagement:

My efforts resulted in high level user engagement through mass marketing strengthening brand loyalty and increasing sales. I received positive feedback about my ability to turn around new requests within hours and for writing complex queries to analyze user behavior and demographics for audience delivery.

The Bottom Line:

The engagement was a fast paced and high volume customer facing real-time analysis  and delivery. On a daily basis I had to think quick on my feet to deliver the right audiences for promotions requiring a turnaround within hours. Our team was able to keep up with pace and succeeded in executing for our client's needs.