Asking Predictive Questions
What are the kind of things we can do with predictive analytics?
Let's take one consumer and see what data we can get from her.
We want to be making some predictions about the future, and that's the hard part. These may include:
We want to make predictions about a fixed period in the future. And these can be of 2 types:
Type 1: how many purchases or returns or whatever people are doing
- Predict what customers are going to buy next.
- Predict how many purchases they are going to make.
- Predict things about customer churn - how many of our customers will stay with us?
Let's take one consumer and see what data we can get from her.
- Imagine that the consumer is seeing some ads over sites X,Y,Z.
- Maybe she's getting some direct mail, or maybe email from the company.
- She's getting messages and offers at A,B,C.
- Maybe she's making purchases at X, Y, Z.
- Maybe she returns one of these products.
- Maybe she has contacts with customer service.
- Maybe she's engaged in social media and she's posting ratings and reviews or interacting with other friends about this product or service.
- How much are my sales?
- What is my market share?
- How many promotions did I run?
- How many things did I sell during those promotions?
We want to be making some predictions about the future, and that's the hard part. These may include:
- How many purchases this person is going to make in the future?
- Is this person going to churn?
- Are they going to drop their subscription with us or not?
We want to make predictions about a fixed period in the future. And these can be of 2 types:
Type 1: how many purchases or returns or whatever people are doing
- In the next month or in the next year, how many purchases will this person make?
- How many products will they evaluate, or how many products will they return?
- Will they make a purchase?
- Will they drop their contract?
