Weighted Email Attributes

I'll use my headphone hobby as an example.

Weighting (you'll develop your own weighting scheme based on a regression model).

• 0-12 Month Purchases = 100% Weight.
• 13-24 Month Purchases = 60% Weight.
• 25-36 Month Purchases = 35% Weight.
• 37-48 Month Purchases = 20% Weight.
• 49-60 Month Purchases = 12% Weight.
• 61+ Month Purchases = 7% Weight.

Let's say a customer purchased two times.

• October 10, 2024 = $200 on in-ear monitors, $140 on a dac/amp.
• May 1, 2021 = $800 on open-back headphones.

Weighted spend:

• In-Ear Monitors = $200 * 1.00 = $200.
• Dacs/Amps = $140 * 1.00 = $140.
• Open-Back Headphones = $800 * 0.20 = $160.

Weighted percentages:

• In-Ear Monitors = 200/500 = 40%.
• Dacs/Amps = 140/500 = 28%.
• Open-Back Headphones = 160/500 = 32%.

On the surface, the customer spent 800/1140 = 70% of historical spend on open-back headphones.

Because the open-backed headphone purchase happened a long time ago, the purchase isn't weighted as heavily ... so it's "relevance" is 32% of past purchase activity.

You'd market all three categories to the customer, given that weighted history is mostly equal. And yes, you'd correlate historical weighted percentages with future activity to "know" that the weights are relevant.

Regardless, use weighted email attributes to personalize your assortment to each individual email subscriber.