Uplift Modeling: Not So Scary After All (Case Study)

January 13, 2015 Nicole Bump

I’m not terrible with numbers. I’d even go so far to say that I’m data-driven in my marketing. But terms like incremental modeling and uplift modeling still sounded a little intimidating when I was assigned to create some marketing materials on this analytics solution. But I put on my big girl pants and got the run-down from one of our analytics experts.

WOW. I suggest you put aside any anxiety you may have with the scary-to-some concept of modeling for a few minutes because this is some impressive stuff.

What is Uplift Modeling?

Actual results from uplift modeling incremental modelingYou can check out this infographic for a great overview on what uplift modeling (aka incremental modeling) is and how it works, but simply put, uplift modeling does more than identify likely purchasers; it also weeds out those consumers who don’t need a promotion from you to make a purchase. This lets you save big marketing dollars by focusing on the group of consumers who need that extra nudge (aka your marketing) to make a purchase.

See It in Action

One of our clients, a leading specialty retailer, was disappointed with the results of their marketing campaigns. Even during major seasonal events, their direct mail promotions were not generating the expected incremental sales. With high costs associated with direct mail, each poorly targeted customer was making a big dent in the bottom line.

How to Fix It: The Game Plan

To optimize the use of this client’s marketing and direct mail dollars, we took a three-step approach:

  1. Perform an exploratory analysis of hundreds of variables to uncover key purchase drivers that could predict incremental spending.
  2. Build an incremental model to establish the existence of  persuadable customers–those who would only shop if provided a call to action from the retailer.
  3. Improve the effectiveness and efficiency of marketing campaigns by focusing outreach efforts only on the persuadables.

This client’s exploratory data analysis uncovered several data points that could predict incremental spend, including:

  • Number of visits the target made to the store
  • Spending in specific departments
  • Number of months from the target’s first purchase

Big Results

  • The incremental model accurately isolated the top 10% of the retailer’s most persuadable customers.
  • By targeting only this top 10% through direct mail, the retailer could experience an uplift of almost 12% and ROI of 287%.

Learn More

That’s not so scary, right? It’s not too shabby either! If you want to learn more about uplift modeling, check out our infographic or drop us a line.

About the Author

Nicole Bump

Nicole Bump, Senior Content Manager, is responsible for developing the Harte Hanks content strategy, bringing this strategy to life through the editorial board, generating much of the company's content and managing the Harte Hanks social presence. A writer at heart, Nicole also enjoys evaluating ways in which new technologies can enable better content production, distribution and measurement.

More Content by Nicole Bump
Previous Article
4 Hacks to Keep Your Company’s Blog Content Flowing
4 Hacks to Keep Your Company’s Blog Content Flowing

As the managing editor of this blog, it’s my job to make sure we have a steady stream of new blog content. ...

Next Article
How to Find Your Most “Persuadable” Targets with Uplift Modeling
How to Find Your Most “Persuadable” Targets with Uplift Modeling

Wouldn’t it be great if we could weed out all of those people that won’t buy, regardless of what we send th...

Don't Miss a Thing.
We'll keep you up to date with our latest content you'll love.

Stay Updated