May 20, 2026

Case Study: A Marketing Plan

I ran a logistic regression equation to understand how likely multi-category buyers are to purchase again in the next year ... after controlling for purchase frequency and AOV ... analyzing twelve-month buyers. Here's the results of the Logistic Regression equation ... nerdy stuff you can ignore (though I do want to show my work).



The variable "multicat" has an Exp(B) of 1.361 ... this means that any customer that historically purchased from more than one category was 36% more likely to buy again in the future AFTER controlling for purchase frequency and AOV.

Separately, I was able to demonstrate that "multiple" is important ... whether the number is two categories or seven categories is much less important.

In a prior analysis, we demonstrated that customers are responsive for a few months following a first purchase, then are dormant, then come back at months 11/12/13 and again at 23/24/25 following a first purchase.

A marketing plan becomes obvious.

  1. Separate email streams for first-time buyers within three months of a first purchase, attempting to generate a quick second order (preferably) within a second category.
  2. Personalize the merchandise in email campaigns for first-time buyers to increase the probability of a second purchase (preferably) within a second category.
  3. When a prospect visits via search, the goal is to convert the prospect to a first order and hopefully cross-shop the customer into a second item and/or second category.
  4. When a customer reaches recency = 11/12/13/23/24/25, we ramp-up messaging and frequency.
  5. Personalize the home page and/or landing pages to show customers what they "need" to see to maximize their future value.

Now, this requires "Beans: The Internet's Only Variety Store" to do things differently. They love to "batch and blast" their email campaigns, they send 12 catalogs per year to customers, they under-invest in social. In other words, they're like any other brand. And when you point out that somebody needs to do things differently and they are simultaneously like any other brand ... well ... the message isn't always well-received.

Which brings us to next week's post ...


No comments:

Post a Comment

Note: Only a member of this blog may post a comment.

It's Time!

Once again, we've made it to the next run of the MineThatData Elite Program! In our study of " Beans: The Internet's Only Varie...