Newsletter 03/2018 - Statistics in Decision Making, Article 3
2frame Analytics' newsletters are back!
This is the third in a series of several articles on Statistics in Decision Making: Answer business questions using Analytics
Imagine the following scenario: you already have a metric that defines what a profitable customer is, you have those customers' data in your database and you want to understand more about your most profitable customers.
The 4 questions below are an example of how questions with the same origin generate different analysis. They can be asked in a row or not, depending on the need of the business, and it is always good to remember: they are summarized, serving only as an example for discussion.
1. Who are my most profitable clients?
If you already have a definition of what a profitable client is for your business, just do a query in your database: fetch the clients' records and sort the data according to your profitability metric.
2. Is there really a difference between profitable clients and the average client?
You already have the list with your clients, with the most profitable ones at the top (question 1). Then the following question arises: are they really "profitable?"
This question is a conjecture, and to test its veracity we will have to use Statistical Hypothesis Testing. It will be necessary to specify the question better. For example, "Is the value of these profitable clients significantly different from those of an average client, with a probability < 5% of being just random?"
3. Who are these clients? How can I characterize them?
You already know who your most profitable clients are (question 1) and whether they are significantly more profitable than average clients (question 2).
But do you know what sets them apart from average clients? What are the characteristics, the patterns these clients show that separate them from the other clients?
Several Analytics techniques are used to solve this question, depending on the type of data, what you want to understand, and what will be done with these characteristics/patterns found.
4. How can I know if a new client will be profitable? How much revenue should I expect from this client?
You already know who your most profitable clients are (question 1), if they are statistically more profitable (question 2), and you already know which characteristics/patterns are specific to that type of client. What else is there?
Well, it would be great to know if a new client is from the profitable group, and what will their behavior be, right? To be able to identify immediately when they enter, treat them differently. Predict the behavior of this new client with your business.
Again, several Analytics techniques are used to solve this question, depending on the type of data, its availability, what you want to understand, among other things.
And in your company, how many questions like those above can be asked with the stored data? How many can be used in decision making?
Read more about the subject of this Newsletter:
1) Data Science for Business, Provost & Fawcett, 2013
Originally posted on 04/15/2018