P L E A S E  W A I T...

Is Market Basket Analysis Still Valid?

For several decades, Market Basket Analysis (MBA) has been a tool used by companies looking to raise their dollar-per-ticket by promoting associated items. The question is: Is it an outdated concept that has been passed over by more modern ways of examining customer behavior?

MBA is an aspect of data mining, i.e., taking large groups of data and looking for useful correlations. It’s specifically related to customer purchasing, and ways to influence them to buy more by offering products that would logically be purchased together through upselling or cross-selling.

An example would be the ubiquitous, “Would you like fries with that?” which is a staple of the fast-food burger industry. Decades of consumer data says yes, a person purchasing a burger would, indeed, like fries with that, and most likely a drink as well. Recognizing and quantifying these connections, and the likelihood of them arising, is the foundation of MBA.

MBA is a statistical model that has a variety of applications. For instance, to use the fast-food analogy, all places that have hamburgers as their core product know that French fries and a drink are the most likely add-ons to a burger purchase. But, the connection flows in the opposite way as well – someone buying fries might be induced to purchase a burger or a drink.

The statistical likelihood is likely much smaller when reversed but that’s the point of MBA: To break that chance down into usable numbers that companies can use to promote their products and showcase them together on menus.

A different example would be a person purchasing an electronic toy is likely to also purchase batteries or, if it has rechargeable batteries embedded, an extra home charger or one that can be plugged into a car’s 12-volt receptacle. There even a probability that the customer may purchase two, so the new user can play with someone else.

MBA also makes companies aware of connections that they may have never considered when shopping such as fads and trends. For instance, I personally like cocktail sauce with my shrimp – an obvious connection. But ,I rarely like the commercially available cocktail sauce so I make my own from ketchup, minced horseradish, spices, and lemon. I am certain this is statistically insignificant to the point of being nonexistent but MBA could pick up on the new fad of people making their own cocktail sauce by showing an upward tick in sales of those products in conjunction with shrimp. This might make a grocery store to put shrimp on sale and place the other products nearby so that shoppers will associate them.

There are also the possibilities of connections that make no logical sense until carefully considered. There is a business urban legend that, on Friday evenings, beer and diapers are often sold together because young men are picking up beer for the weekend and their wives remind them to buy diapers and the weekend can be enjoyed uninterrupted by a trip to the store. It’s not actually true, but it makes for a great story and points up how such intertwining of product purchases might happen.

All of these processes and examples come down to the fundamentals of human interaction. People are more likely to buy more when products presented to them have an obvious and logical connection. In contrast, they make associations that make no sense at first glance, but have their own underlying logic that makes sense to them. If a large enough group of people follow this, it becomes statistically relevant enough to be useful to companies in marketing.

Because human behavior hasn’t changed for centuries, if not millennia, it’s a fairly good bet that Market Basket Analysis will be around for a very long time, too.

Drive better results by understanding customer data