Predictive algorithms are watching your shopping cart

Back in the times when it was an online bookseller and not today’s ecommerce giant, Amazon used to rely upon human-written book reviews for suggesting the next book purchase to its customers. However, when a predictive algorithm was experimentally introduced, Amazon quickly found out that the algorithm-generated recommendations led to more purchases than human-generated ones, at a fraction of the price.

Predictive analytics is the name for a set of algorithms and techniques that can extrapolate the future trend of some variable of interest based on the recognition of recurring patterns in a population of data. In the case of Amazon’s recommendations, the algorithm suggests an item that has a high likelihood of being appreciated by the customer, given the customer past purchase record and the preference patterns of other users with a similar purchase profile.

Amazon’s patented recommendation system is probably the most widely known example, but the use of predictive techniques is widespread, both off- and online, as proposing personalized purchase suggestions is proven to be a highly effective marketing strategy. Today’s availability of large and complex data sets makes the results of such techniques disturbingly precise.

Predictive shopping

The natural evolution of customized shopping recommendations is a system that, based on the records of your past purchases and on your registered preferences, can identify with accuracy what you will want to buy and takes care of having it purchased and delivered to you before you actually choose or order the thing.

This system, generally referred to as predictive shopping, is already close to becoming a reality. Indeed, to some extent it is just an extension of already existing subscription plans or auto-bill options for some goods. Upsetting as it might appear at first sight, predictive shopping also holds a concrete promise of liberation from tedious everyday tasks.

Think for instance about routine goods like soap or toothpaste for which shopping is a distraction. Or think food. Maybe you are one of those people whose meals are strictly immutable from day to day, or maybe you are the kind who like to treat themselves to something different every day; but even in this latter case you experience an amount of repetitive weekly shopping such as for milk or coffee or pasta.

Viewed in this light, subscription to a predictive shopping plan could spare you from having to deal with repetitive tasks or distractions and grant you some precious free time. Going a step further, it is easy enough to imagine predictive shopping plans for clothing, books or music based on the suggestion systems of Amazon or Last.fm. Exciting, but this liberation might come at a price.

Freedom is slavery?

Some of the issues with predictive shopping already fly you in the face when you think about it. The idea that a retailer can bill you for goods that you haven’t ordered or even chosen, based on the mere expectation that you will want or need to buy them, is disturbing enough. A second, not less worrying issue is with businesses having access to a profile of your habits and preferences so precise that it could actually pose serious threats to your privacy.

A Forbes.com contribution expresses an even more extreme worry. The experience of choosing for ourselves has a value and a purpose, even though at times it can be frustrating or annoying or it could appear as a waste of time. By denying us this experience, predictive shopping makes our lives easier, but at the price of reducing our capacity to plan ahead and think about the future.

These worries may appear farfetched in a scenario where predictive shopping is offered as a service, because in this case only people who feel comfortable with outsourcing their shopping decisions to the algorithm would choose to do so. A small survey conducted by Law professor and philosopher Cass Sunstein revealed that people having a favorable attitude towards predictive shopping represent a minority, even though an unsettling large one.

It should be considered, however, that predictive shopping is extremely advantageous for both producers and retailers, since it enables an almost perfect planning of production and supply. Thus it is to be expected that, once predictive shopping is introduced, the backing for this option will be strong. Customers will be encouraged to subscribe to predictive shopping plans, and such plans might even become the default option for the purchase of some goods, similarly to the way ready-to-wear has became the standard in the clothing industry while made-to-measure and customization are luxury features.

In light of this scenario, even the preoccupation expressed in the Forbes.com article sounds a bit less extreme and a bit more worth thinking about.

Leave a Reply

wpDiscuz