52 Things We Wish Someone Had Told Us About Customer Analytics by Alex Sherman & Mike Sherman

Author:Alex Sherman & Mike Sherman [Sherman, Alex & Sherman, Mike]
Language: eng
Format: azw3
Tags: This book is for anyone who uses customer information to make business decisions: CMOs, CEOs, marketing directors, product owners, consultants, and the people who provide that information: data scientists, market researchers, business analysts, statisticians, subject matter experts. By tying impact to tools and techniques, through real-life stories, we hope to help decision makers better understand how to practically use customer data while helping data analysis providers to better understand how to create output that end users will value.
Publisher: Amazon.com
Published: 2021-04-01T00:00:00+00:00

1. Look at behavior more than demographics

What people do can be more insightful than who they are.

At a Southeast Asian cable television operation, ethnic packages were created and targeted solely based on demographics: Chinese packages for Chinese families, Bahasa packages for Malay families, Tamil content for Indian families, and so on. Although this was safe and avoided selling irrelevant content, it limited the programing sold. By looking for an iron core, we found the language of the content watched.

Extracting the languages watched from the viewing history suggested that some Malay families were already watching Chinese content, and vice versa. The initial reaction of management was that this finding was wrong, so they wanted to disregard it. By the next day, the response changed, “Well, maybe our demographic data is wrong for these families.” Or maybe the viewer is in a mixed marriage, or is a child studying Chinese by watching TV, or it is the maid who is watching, or ...? You get the point; it could be for many reasons. As a result, we were allowed to test a model that leveraged this finding and the upsell rates increased substantially. This did not surprise us because if someone already watches some Chinese content, that is a pretty good indicator that he or she might be interested in more, regardless of their ethnicity.

2. Transform the data

Transforming data is useful as it can reveal latent variables.

When teaching, we often write a date and name on a whiteboard and ask how many features we just provided. Most students confidently state: two features! We then explain that from the date we now know if it is a holiday, weekend, period close, and much more. Names come in and out of style, so we can use them to infer age and other demographic information. If you told me your name was Alex, chances are much higher that you were born in the 1990s than the 1960s.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.