New Buyer Quality: Logistic Regression

Ok, a geeky detour for today in case you want to do this work yourself.

Every first-time buyer in "Hillstrom's Newbies" purchasing for the first time between a date range is analyzed ... did the customer purchase again within twelve months?

We use a Logistic Regression framework - the methodology allows us to see the impact each variable has on subsequent purchase activity.

Now, the numbers are kinda hard to read here, so don't really focus on the numbers. Focus on the bullet points below.

• I measure the impact of each additional item purchased in a first order.
• I measure the impact of the price of each item purchased in a first order.
• I measure the impact of discounted/promoted items in a first order.
• I measure the impact of new/existing items in a first order.
• I measure the impact of canceled items in a first order.
• I measure the impact of returned items in a first order.
• I measure the impact of the month of acquisition.
• I measure the impact of the share of a first order in each merchandise category.
• I measure the impact of marketing channels in a first order.

There are other attributes worth analyzing, but for the brand being studied, this is sufficient to tell a story ... the story of why a first-time buyer purchases again.

When you get a "Hillstrom's Newbie" project, you get the Logistic Regression analysis as the foundation of your customer behavior. From there I rank-order the new buyers from most likely to purchase again to least likely ... and report to you what I learn.