The following article is written by Peter Zajonc, Senior Director, TS and TSP Analytical and Modeling Services, Epsilon, and the Membership Chair of the DMA Analytics Council. a member community advancing the interests and knowledge of marketing analytics professionals. Peter recently led a Town Hall conversation on this topic, and is working with other members of the Analytics Council to develop a new language around predictive analytics. If you are interested in joining the Analytics Council to work on this or other projects, please sign up here.
Selling a car is different from financing the new car. Often shoppers see the product (car) and the means to acquire the product (financing) bundled as one transaction but the marketing for the product and the means to determine eligibility for financing the product are different transactions, using different data for different purposes.
Along each path to purchase – from cars to catalog items – consumers are presented with a bundle of options, hard to distinguish and discern, but all enticing them to ask: Is that my color, style, brand? Is it in my price range? And how will I pay? Shouldn’t I be waiting?
For the marketer’s part, the data-driven marketing analysts build models to predict each of these consumer propensities and behaviors. Given these likely consumer propensities, our clients, successful advertisers and marketers, seek to drive targeted audiences to make the purchases that resonate, and that fit their tastes and needs.
Well-modeled audiences, using appropriate marketing data, such as age, income, lifestyles and purchase history, help make tough marketing odds — 1 in 100 or, even, 1 in 1000 — profitable. Targeted offers urge consumers to respond to ads in their mail box, on TV, on their computer screen, or mobile phone. Propensity models may predict behaviors along the path to purchase, such as, conversion of a confirmed lead to buyer status, likelihood to re-subscribe or lapse, and propensity to be in the market for ancillary products. Through harnessing the power in data, consumers benefit from timely and accurate marketing for products they are interested in and marketers benefit by improving the ROI for their stakeholders.
For large, non-discretionary purchases, easing the pinch on the pocketbook through convenient financing is essential to completing a deal, as well. So merchants and marketers try hard to make financing a seamless part of the package. And, while it appears seamless, the types of data used are quite different.
The financing itself is often marketed, but when doing the analytics we maintain two types of data, and two distinct purposes.
- First, the consumer may receive a marketing invitation to apply for financing, and marketing data is used to find that audience. Who is interested in the product? Who may need financing?
- Second, once the consumer applies for the financing, what data is used to determine eligibility? When credit eligibility comes into play, the financial institutions use their own data to determine credit eligibility. That data, such as credit scores, is subject to the safeguards of the Fair Credit Reporting Act. So, marketers use their data to find the parties interested in their products; the financial institutions use their data to determine credit eligibility. It’s that simple.
As data-driven marketing analysts, it is important that we take an active role in re-defining the distinction between these two disciplines—marketing data analytics and financial data analytics used for eligibility purposes. The similarity between statistical techniques, and, in particular, the language and words used to describe them, creates ambiguities between marketing propensities and credit scores. As similar as they are from a mathematical and statistical view, the data serves different purposes and has different protections and origins. To that end, marketers need to speak a new language to better maintain this distinction.
The table below may be a start of a better vernacular, or lingo, that might help us in our day-to-day jobs to reduce confusion over the long run.
The downside of using terms borrowed from our credit score modeling brethren puts too much of our livelihood – and the livelihood of other data-driven marketers — at an unnecessary risk of being misunderstood. Some terms may not need to change. Terms such as “selects” and “pre-selects” do not have credit connotations. And, while “rank” and “decile” do suggest scoring models, they are not found in credit modeling and might not need synonyms.Table. Credit v. Marketing — Parallel terms serving different purposes
It’s no small ask. Words are our friends. They represent concepts we’ve lived and worked with for years. Some have deep meaning for us, and signal immediate information to our co-workers.
While making a change in our lingo is unnatural, it can and should be done. In fact, this is a chance for the modeling and analytics community to show industry leadership and make a difference. So, with that next project you present, remember you are in a great position to explain the propensity audiences your models target using more useful, data-driven marketing analytics vocabulary.
Summary. Marketers use advertising to compete and win market share. Identifying the best audience for their products and services helps get their message into the minds of consumers most likely to buy. As we move into an increasingly complex data-driven marketing age, it is increasingly important, as well, to use the words that best describe the data analytics part of the marketing business.
What do you think? Please add comments below.