You Are What You Buy

We’ve all heard the old adage, “You are what you eat.” And while a person’s physical and mental well-being certainly are influenced by what he or she consumes on a regular basis — a point made abundantly clear in the recent fast-food documentary film “Super Size Me” — the saying is hardly a universal truth. A far more complete portrait of who a person really is in terms of likes, dislikes, interests, attitudes and lifestyle might be better stated as “You are what you buy.”

The information to be found in credit card and checking account transaction records can speak volumes about any given individual or household. Transaction analytics — the process of mining data for meaningful patterns and actionable insights through the use of advanced tools and modeling capabilities — can shed light well beyond the obvious, such as income sources and mortgage payments.

Records of itemized transactions, especially as they relate to discretionary spending, can almost be viewed as entries in a personal diary. By analyzing a person’s cumulative purchase decisions over time, one can begin to discern his or her psychographic makeup along myriad dimensions — who they are, where they’re going and what they may wish to purchase next.

Consider that transaction analytics can reveal a person’s sports interests, even identifying a favorite team. Similarly, analytics can uncover their musical tastes, including which specific concerts they attend, as well as their culinary preferences — for example, instantly distinguishing a sushi lover from a Big Mac aficionado.

Transaction analytics can bring to light a person’s favorite hobbies, their fashion sensibilities and their appetite, if any, for luxury goods and services. Where and how often do they vacation? Are they adventurous? Are they athletic? How do they pamper themselves? What magazines do they read? What clubs do they belong to?

The answers to these and countless other questions all reside in transaction data, and ultimately allow companies to send the right offers and messages to the right consumers at the right time. But arriving at the correct answers is hardly a simple matter.

The key to this process is pattern recognition. After all, the moment a pattern changes is often the right moment to take action.

Indeed, a deviation from a person’s normal purchase behavior can serve as an excellent early indicator of an impending event in his or her life. The detection and recognition of the deviation can then be used to trigger a specific marketing treatment through the use of rules-based decision technologies.

For example, in anticipation of the arrival of a new baby, parents-to-be naturally tend to make a significant number of baby-store purchases. Given the ability to detect and recognize this pattern, a financial services company would be in the enviable position of being able to put forward specific products geared to the new parent, such as a life insurance policy or an educational savings account plan.

Or consider a purchase pattern that suggests a teenager in the family is about to leave for college, in which case the empty nesters may be faced with a very different set of financial circumstances. A company may be able to address this new set of circumstances very effectively, and in a context-sensitive manner — again, by presenting the right people with the right offer at the right time.

Are baby-store purchases and college tuition payments readily evident in a transaction record? Not always. In fact, a fundamental challenge with transaction analytics revolves around merchant identification, since this information often is completely masked or poorly described.

In general, MCC and SIC codes, which originally were generated by the Visa and MasterCard networks with a very different purpose in mind, provide a limited and often ambiguous view of the nature of a merchant’s business. At times, there’s no telling so much as the industry in which a merchant operates. The key is to remove the non-relevant data from the descriptive information, and then augment the relevant information with ancillary data to create a reliable match.

The process employs sophisticated methodologies, including natural language processing and latent semantic indexing, which essentially means creating a lexicon to describe the nature of each transaction using the text-based information available in the transaction and merchant records. Having created this lexicon, it then becomes easier to understand the relationships that exist among the different transactions.

Transaction analytics presents many opportunities to drive revenue growth for any multiproduct enterprise dealing in demand-deposit accounts. Today, most financial services companies continue to numb their customers’ senses with redundant and irrelevant offers that are solicited based on stale demographic data. In contrast, transaction analytics can empower companies to understand their customers’ wants, needs and situations at a granular level and in an up-to-the-moment fashion, and to respond accordingly.

For leading financial services companies, it may well be the pathway to greater shareholder value.

Gordon Cameron is a vice president and Will Ferguson and Jeff Zabin are directors at Fair Isaac Corp.