CHICAGO– A high response rate does not always equal marketing campaign success.
That was the message credit union analytics expert Karan Bhalla used to kick off his talk before an audience of Predictive Analytics World conference attendees this week. Marketers must instead be focused on a particular campaign’s ability to produce incremental sales, Bhalla said. It’s a task he explained is made simpler by using a sophisticated, yet accessible, technique – uplift modeling.
In a session titled “Evaluate a Marketing Campaign's Success Using Uplift Modeling,” Bhalla introduced the predictive data analytics technique, explained its value and also shared a case study.
Uplift modeling, Bhalla explained, directly models the incremental impact of a particular campaign on a particular individual’s behavior. Essentially, the method goes beyond identifying which consumers in a given population are most likely to respond; it identifies which will respond in a way that results in incremental sales for the business.
“In other words, the technique allows you to identify those individuals likely to buy from you anyway -- even if they had not received your offer,” said Bhalla. “We’ve all been there. You’re just about to book your flight or buy that pair of shoes when you randomly come across a coupon. It’s awesome. Well, awesome for us, but maybe not for the company. We want to help financial institutions avoid the wasted marketing expense of persuading people who have already been persuaded.”
Rather, the goal is to target those consumers open to persuasion.
“These are the people who will not buy if you leave them alone, but who will buy if you connect with them,” said Bhalla.
One Top 20 FI's Story
Arguably just as important are the individuals for whom contact, marketing or offers may have a negative impact. Retention campaigns, in particular, have been known to trigger both telecommunications and financial services consumers to end relationships by canceling subscriptions or contracts.
“There are people who may stick with your financial institution if left alone but who may have a negative or adverse reaction to your marketing,” explained Bhalla. Uplift modeling can identify these individuals, as well, so marketers can exclude them from a particular campaign or alter the content of their marketing efforts.
Bhalla shared the story of a top-20 U.S. financial institution that used uplift modeling to improve the results of a HELOC cross-selling campaign it had performed annually for several years. The results of two campaigns that relied on uplift modeling produced incremental revenue more than 300 percent higher than similar campaigns executed in the past. What’s more, the financial institution’s mailing volumes were reduced by nearly 40% for each campaign.
Targeted marketing strategies, especially those that rely on uplift modeling, are in philosophical alignment with community financial institutions because they are consumer-centric, Bhalla added.
“As with any new technique, trend or technology that has people talking, uplift modeling is going by a few different names these days,” he said. “You’ll also hear people refer to it as incremental modeling, true lift modeling or net modeling. We like ‘uplift’ because it describes how we want the consumer to feel after receiving a truly valuable, personalized offer that will in some way improve their financial lives.”