About
Uplift Modeling is an algorithmic technique that classifies email campaigns into four quadrants based on shoppers' performance metrics upon receiving an email.
Why
In eCommerce email marketing, you encounter many aspects beyond your control, such as demographics or time since the last customer purchase. You must rely on educated guesses regarding aspects you can influence, such as email design, value propositions, product recommendations, or coupons.
You might wonder:
How much will this customer buy if we send an email without any incentives?
How much will they buy if we offer them free shipping?
How much will they buy if we give them a $10 coupon?
The answer is provided by something akin to a 'crystal ball': analytics that use uplift modeling to analyze attributes you can't control, such as demographics and time since the last purchase, and estimate the impact of actions within your control, like sending a discount or offering free shipping.
How
The goal of Uplift Modeling is to predict the incremental change in behavior caused by a marketing action, such as targeting a specific audience with an email offer.
The simplest approach to uplift modeling involves sending an offer to a subset of an email list (the ‘test’ group) while withholding it from others (the ‘control’ group). These two groups should be as similar as possible in all other respects.
After conducting your test, you can use email contacts’ attributes and the uplift measurements to categorize your contacts into four groups, as shown below:
Source: Towards Data Science
WHAT
The insights are displayed as two-dimensional click-through bubble charts, allowing you to see the big picture and quickly dive into the details of individual campaigns and the impact of products on the campaigns' scores.
Uplift Model: A chart that categorizes all email campaigns into four groups: Sure Things, Persuadables, Sleeping Dogs, and Lost Causes.
Group Results: Accessible by clicking on a bubble or a table element, this feature displays the individual results of all campaigns belonging to a group.
Product Results: Accessible by clicking on an individual campaign bubble or a campaign table row, this feature displays how individual products or landing pages are impacting campaign results.