• Dimitar Antov
  • Linda Deeken
Jun 27, 2017

Trimming the waste line: How the Uplift framework lifts up vague marketing strategy

John Wanamaker—who pioneered the department store in his native Philadelphia in 1861—once famously said, “I know half the money I spend on advertising is wasted, but I can never find out which half.” Fast forward from the 19th to 21st Century and financial services firms must have a clear answer to that riddle, correct? Or so it would seem.

Measured ad spending for financial services firms hit $5.3 billion in 2007 and jumped to $7.6 billion in 2013: a compounded annual growth rate of more than 6 percent, based on data gleaned from Kantar Media and Nielsen. This is substantial, but can we honestly say we occupy any better position today than Wanamaker did?

Perhaps we don’t even know the half of it.   

Yet it’s certainly not for lack of trying or appreciation of the fundamental problem. Unfortunately, retail banks today and their analytical approaches have failed to keep pace with marketplace shifts. They live in a world where survival depends on strong levers to “pull” customers (innovative products, differentiated brands) and/or stellar “push” strategies (direct marketing engines) to maximize marketing effectiveness and efficiency.

One or the other would float the boat, but too often retail banks have neither and find themselves cast adrift. But with the help of certain analytical approaches, banks can enhance reach and precision across all channels.

The fundamental problem: conversion confusion

Banks too often analyze the wrong outcome metric. Instead of a customer focus where the actions of the bank can impact change, they focus on those who’d convert without intervention and/or those who will not, regardless the amount of intervention. This creates the worst of all worlds.

The framework of a solution: “Uplift”

The solution is to shift the paradigm.

That is: Don’t identify the likely users. Instead, separate the “influenceable” from those who are not; go beyond consumer behavior to “influenceable” consumer behavior where your actions can impact change and deliver ROI.

Uplift Modeling Framework

Banks need to deploy analytical techniques that improve precision and enhance customer reach by splitting the possible population into four distinct groups:

  • Sure things: those who don’t need incentives. They will purchase even without any discount. Money spent to reach them via marketing and discounts is wasted—in fact double wasted.
  • “Persuadables”: those who would make the purchase if offered an incentive and a compelling message.
  • Sleeping dogs: customers alienated by aggressive pursuit with offers. Think of the “pushy sales person” hawking a discounted warranty on the laptop you plan to purchase at the electronic store. The aggressive sales tactic makes you leave the store without the laptop.
  • Lost causes: those who would not purchase regardless of any incentive.

This framework offers three distinct benefits:

  • It identifies where not to spend
  • It quantifies how much to spend , and
  • It indicates how much effort to exert in the right areas.

These benefits are relevant across all stages of customer’s relationship —acquisition, development and attrition. With acquisition, for example, firms shouldn’t focus on “lost causes” who will never buy. Instead, they must find innovative ways to enhance acquisition among “persuadables.”

Efficiency (doing things right) and effectiveness (doing the right thing)

Uplift, however, is not just about cutting costs. It also boosts precision and effectiveness by sidestepping the “wrong” consumers. What’s more, these approaches achieve higher response rates with lower levels of action. While a substantial number of successful companies in outside sectors rely on uplift modeling, only a handful in the financial services sphere leverage this technique today. A major U.S. bank reports that it prioritizes 30 percent of its current customer base for home equity lines of credit (HELOCs) identified through uplift models and achieves close to 90 percent of the incremental sales it would get from attempting to engage the entire population.

We have seen results of similar magnitudes in retail and telecommunication.

Five steps to step up your game 

So how do you transition from current state to a lean, productive marketing engine?  Five simple steps will help your organization fundamentally re-think the approach.

1) Identify the relevant incentives to test.  It’s likely not just limited-duration promotional rates. Rather, these will include new product features, tiered rates, bundling options, fee levels, forgiveness details and access to information and support.  Think holistically about the solution you offer consumers and don’t box yourself into a commodity, price-only position.

2) Identify the population against which to test these metrics.  As you leverage your understanding of the true drivers of demand, it’s time to grasp the segments of the population against which to test each offer and message. Consider the various price sensitivities, current product ownership, motivation for benefits and propensity for loyalty, for example. 

3) Test and learn. Split your priority population into two subgroups with sufficient size to carry out econometric modeling. The test group will receive the incentive you identify in step 1, while the control group won’t. After carrying out the campaign, collect information on the outcomes for each subgroup and compile a modeling dataset.

 4) Create econometric models. Base these on the collected data to predict the outcomes of each group. A spectrum of data elements could be used for predictive purposes—from observable behavior and transaction characteristics in your customer relationship management database to third-party syndicated data fields.

We prefer to rely on data mining techniques such as classification and regression trees to sift through all available data and identify fields with predictive power in dividing the population in the observed subgroups. The models created can categorize each customer into one of the four uplift framework quadrants.  

5) Optimize the campaign based on modeling results. Apply the models to each candidate who would receive an offer. Continue to refine and track; learn from the successes, adjust for the losses and drive towards meaningful engagement with potential consumers.

Putting it all together: A transformation of differentiation

While not difficult to implement, this approach is profoundly transformative. In effect, it creates an insights-driven customer differentiation vs. the traditional data-driven product segmentations. It’s also more hypothesis-driven and customer-centric as well; utilizes existing data more effectively; and most importantly, increases marketing ROI.

Wanamaker, who was also a marketing trailblazer, built his department store concept on this principle: “One price and goods returnable.” It was revolutionary at the time. If only financial services marketing were so simple today. But amended for the playbooks of companies eager to close the gap on that “wasted other half,” you could well adapt Wanamaker’s wisdom this way: “Uplift: One prize and good returns.”

Linda Deeken is chief marketing officer of The Cambridge Group.  She has been published in Harvard Business Review and is a key contributor to several books recently released by The Cambridge Group.  Previously she served as a Cambridge consultant.

Dr. Dimitar Antov is an analytics principal with The Cambridge Group and a member of the Economic Center of Excellence. He has worked on a wide spectrum of engagements involving strategic resource allocation, precision targeting, marketing and sales execution, customer engagement and penetration. 

BAI Banking Strategies

Thank you for visiting BAI Banking Strategies. To view more, please Subscribe or Login.

Dismiss