How Technology Will Change Black Friday 

Black Friday is now a distant memory. But the next one is never less than 356 days away, so pundits are analyzing the results of shopper habits during the last holiday buying season and what they ultimately mean for the future of retail. Here’s one key example: This was the first year online shoppers outnumbered in-store shoppers, according to the National Retail Federation.

Many bottlenecks in the shopping process, such as the home bandwidth shortage that created Cyber Monday, are now a thing of the past. Instead, we now see many new technologies that are fundamentally changing the way retail works. How will these technologies affect Black Friday 2016? Here are some predictions:

1) The rise of “persuasive analytics.” Unless you’ve been stranded on a desert island, you are aware of big data analytics, which are fueled by cloud computing and Open Source platforms such as Hadoop and Spark. By analyzing millions of events per second, retailers can predict a shopper’s next action based on actions of similar shopping behavior. Today, predictive analytics are being superseded by persuasive analytics – retailers can not only predict what you will do next; they can steer you toward a desired outcome, i.e., a purchase decision. The key enabling technology is uplift modelling, a mathematical technique that directly models the incremental impact of a treatment (e.g., being shown a product illustration) on an individual's behavior. This approach was perfected in the political arena - some say it was a key factor in President Obama’s election victory in 2008 - and has now crossed over into the retail domain.

One familiar example of uplift modelling is A/B testing, which can determine which presentation elicits the best result from a given user. Companies also use uplift modelling to reduce churn, by first determining leading indicators for a customer departing, then further analyzing that group to determine which customers can be saved, and finally sending in the hug squad to smother those wavering but salvageable users with attention and personalized targeted discounts. Uplift modeling can also be used to maximize sales of follow-on items after the user has made an initial purchase. For example, suppose a customer has just purchased a cubist poster from an online art store. This customer should immediately be offered a frame and lighting that suits the type of art purchased and the customer’s taste, as inferred from previous online purchases. Uplift modelling capabilities are currently available as part of the R analytic language, as well as in commercial analytic offerings from Dell, IBM and SAS, among others.

If uplift modelling is so powerful, why is it not being used more widely? The main constraint is scale – it takes twice as large a sample group to perform uplift modelling rather than standard statistical modelling, somewhere on the order of 100,000 customers to achieve a statistically reliable result. So, this technique is used today by the larger players who have that volume of customers.

2) Variable pricing: Forget about a single fair price for all consumers all the time. Modern Big Data analytics enables the retailer to dynamically modify the price based on:

Demand: This is known as surge pricing. It has been used most (in)famously by Uber, to charge more for a ride when the demand rises on a rainy day. Retailers can dynamically raise the price based on the number of consumers who want to buy the product right now.

Who you are: Algorithmic commerce companies like FeedVisor and Zilliant can adjust an item’s price to a specific customer in real time, based on factors such as the competitive landscape, pricing trends, user demographics and behavior analysis. This is a retailer’s dream – getting the best possible price every time.

3) Data Science in a box: Today, retailers have a hard time extracting actionable insights from the Big Data they collect. This requires hiring data scientists, who have deep domain knowledge of retail, as well as a thorough understanding of statistical and machine learning algorithms in order to implement new Big Data-based solutions and technologies. The problem is, good data scientists are as rare as unicorns and typically work for only the biggest and most technically advanced retailers – at high wages. But now a number of companies, such as DMWay and C-B4, are seeking to make Big Data insights available to all retailers by simulating the expertise that a great data scientist would bring to bear. Soon all retailers, large and small, will be able to benefit from Big Data analytics.

As technology eliminates more and more bottlenecks in the shopping process, one can imagine a future where Black Friday becomes obsolete. With no need to wait on line or squeeze into a brick-and-mortar store, no need to wait for online access to the item you desire, and no worries about timely delivery, shopping should become a much more relaxed and pressure-free experience. But here’s the reality: even in the digital future, retailers will continue to create events like Black Friday or Amazon Prime Day in order to make users feel pressured to buy now or risk losing out on purchasing the items they want. The timing may be arbitrary, and the psychological pressure may be artificial, but retailers know that consumers buy more in the face of a deadline. For retailers, Black Friday is not a constraint – it’s a choice.

To survive in this digital future, retailers will have to up their game, particularly in areas like customer experience engineering, Big Data analytics and agile development processes. Some companies will build up this expertise in-house, while others will find trusted partners to help them. In an environment where change is constant and the competition is nimble, standing still is simply not an option.

Moshe Kranc is CTO of Ness Software Engineering Services, a custom software development firm headquartered in Teaneck, NJ.