Customer Behavioral Analytics

Dave Wentzel

Philadelphia Microsoft Technology Center

Contact Information: LinkedIn
Presentation

What is this?

This is a short course/primer on how I do Customer Analytics using a data-driven approach. This is not meant to be all-inclusive. This is meant to demonstrate an approach to performing Prescriptive Analytics using standard data analytics tools like:

• SQL (every business analyst should understand a modicum of SQL).
  • We will use Synapse SQL Serverless but we could use any SQL tool that supports a SQL abstraction over files in a data lake. Databricks is an excellent choice.
• python/notebooks/Spark (data professionals use these tools but they are within the grasp of any business analyst)
  • We will use the tooling built into Synapse

But the technology is much less interesting than understanding a simple repeatable process to perform analytics that will actually work for your people and their capabilities.

You do NOT need to be a python expert or understand advanced SQL. What I will present is merely the thought processes that will help you on your analytics journey. I switch between python and SQL a lot -- that is by design -- I choose the best tool for me to accomplish a task, but you could use whatever YOU are comfortable with. The MTC is here to help.

I consider customer analytics to be part of what I call Prescriptive Analytics. This means that the analytics we do are far more advanced than just predictive or descriptive (machine learning and basic BI reporting). We want to provide our business with ideas regarding What do we do next?.

Customer Analytics

Without customers, your company is out-of-business. The most successful companies are using customer behavior data to make key business decisions. Many companies struggle with things like:

• customer segmentation
• loyalty
• understanding Customer Lifetime Value
• "optimizing" churn.

I’ll bet your company has the data to tell which customers have churned and which might, but most struggle with “ok, now what do I do?” As data professionals, we are uncomfortable making opinionated recommendations on what we should do next based on what the data is telling us. In this session I’ll show you how to use data and analytics processes to understand customer analytics issues and how to help your business leaders interpret their data to answer the question: “What should we do next?”.

Customer Analytics deals with using data and being data-driven to:

• attracting the right customers (promotions)
• satisfying and meeting the needs of your customers
• retaining customers
• set the correct price point to maximize revenue (known as price optimization)

It's been proven (you can go find the links yourself) that if your business gets good at customer analytics it will definitely provide lift to your business, in the form of higher profits and reduced costs.

But even more importantly, if you master the Design Thinking and Rapid Prototyping process that I use, and tailor it to YOUR business, these techniques will help you solve ANY analytics problem.

So what is it?

Customer analytics is a process where we use data from customer behaviors to help make key business decisions. We use the outputs to determine how to do better direct marketing, store and warehouse site selection, and customer relationship management. We want to predict customer behavior and determine how we can shape it.

Some use cases:

• marketing
  • increase response rates
  • improve campaign ROI by contacting the right customers with timely and relevant offers
• improve customer loyalty
• decrease attrition/churn by predicting the most likely churners and developing proactive campaigns to retain them.
• segment customers more effectively to understand their behaviors.
• come up with a calculation for Customer Lifetime Value (usually abbreviated CLV or LTV) that we can use for tons of different business use cases.

Today's Use Cases

Here's what we'll look at today:

• Customer Segmentation Analytics
  • There are a lot of uses for grouping your customers into "segments" or cohorts.
• Customer Attrition Modeling
  • Businesses want to know why are my customers leaving and what do I do next?
• Customer Churn Analytics
  • This is a simple bank dataset where we predict churn. There is no commentary in this notebook, just the code.
• Customer Lifetime Value