April 15, 2019

Hillstrom's Targeting: Weighting Variables

The secret to my targeting strategy is in the weighting of data, specifically, prior purchase data.

Now, you might have your own weighting strategy, and if so, that's fine, go with it. I like to discount older transactions.
• 0-12 Month Transactions =   100% Weight.
• 13-24 Month Transactions =   60% Weight.
• 25-36 Month Transactions =   35% Weight.
• 37-48 Month Transactions =   20% Weight.
• 49+ Month Transactions =      12% Weight.
What does this mean?

Let's look at a sample customer:
• 0-12 Month Spend = \$100.
• 37-48 Month Spend = \$100.
• Weighted Spend = \$100*1.00 + \$100*0.20 = \$120 Weighted Dollars.
Here's another sample customer.
• 0-12 Month Spend = \$25.
• 13-24 Month Spend = \$25.
• 25-36 Month Spend = \$100.
• 37-48 Month Spend = \$100.
• 49-60 Month Spend = \$100.
• Weighted Spend = \$25*1.00 + \$25*0.60 + \$100*0.35 + \$100*0.20 + \$100*0.12 = \$107 Weighted Dollars.
The first customer - even though the first customer spent just \$200 historically ... the first customer has more "weighted value" than the second customer.

In individual projects, I use a regression methodology to assign the weights. On average, the weights end up being similar to what is described above.

Tomorrow, I'll show you that the weights "matter", ok? We're in the process of building a targeting strategy to implement Welcome Programs, Anniversary Programs, and Optimization Programs.