Will your organization recognize and respond to disruptions faster than players in the retail industry?

By Lei Tong and Katherine Hampton Tong

2017 is poised to be the worst year for store closings in retail history. There has been an alarming increase in the rate of retail bankruptcies, up 32% since last year, with store closings triple that of last year.  Several large retailers have come to dominate the industry over the last year largely because of innovation and technological advances. Most notably Amazon has contributed to disruption in the retail market, essentially blurring the lines between a retail company and a technology company.  Its advanced technology and delivery platform has facilitated the success of smaller companies that otherwise could not compete with their larger competitors, and in so doing forever changed the industry by democratizing e-commerce.  This past year, the retail giant has rightfully earned the nickname of ‘ultimate disrupter’ as it jumps to fifth most valuable US-listed company and third-largest retailer in the world. For the rest of the retail industry this has created an imperative to innovate and rapidly in order to secure their survival.

Are there similar signs of imminent disruption within your industry?

How did Amazon accomplish such an incredible feat of retail disruption?  As Artemis Berry, vice president of Shop.org and the National Retail Federation, puts it, “Amazon is a case study in ceaseless innovation and interminable disruption.”  Innovations from Amazon include more efficient, cost reducing and free shipment methods; 1-click shopping; Amazon Prime member loyalty program; digital book purchasing and on-demand production; breadth of assortment; unconventional but consumer delighting return policies;  dynamic pricing; Amazon devices; targeted automated emailing’s; reviews and ratings; recommendation engines; and soon delivery drones and physical stores.  These innovations are helping to drive execution of the strategic plan by ensuring a solid grounding in innovation and investing in long-term customer satisfaction/loyalty.  Many such advancements were developed based on insights through Amazon’s strong foundation in analytics.  The supply chain and logistics optimization alone is projected to reduce costs by 10% to 40%.

How is innovation incorporated within your organization’s long term strategic planning?

In addition to Amazon, companies such as Alibaba, Wal-Mart, CVS, Target, The Home Depot, Nike and The North Face have utilized analytical algorithms to stay in the game.  For example, Nike and The North Face have taken product recommendations to another level even beyond what Amazon has accomplished.  These companies are using Artificial Intelligence (AI) to help online customers more easily find and customize products of choice.  North Face has partnered with Fluid to use IBM Watson to replicate the personal shopper experience online.  Considering the huge consumer shift to digital personal assistants, The North Face is capitalizing on this trend by creating a Digital Personal Shopper who will ask individual consumers targeted questions about what they want, and for what application just like a store associate would do.  For every question the Digital Personal Shopper asks, the system calculates a score for every product in a catalogue, uses previous answers to determine the next most likely question and finally presents high, medium, and low product matches for the consumer.  Customers can simply speak in their own language to express their needs and allow analytics to guide their selection process.

Is your organization developing “code capital” to create competitive advantage?


How to get started incorporating predictive analytics into your organization’s growth strategy

Although the analytical and technological rigor of these companies might seem unattainable for many, all organizations no matter the size can take small steps to implement advanced analytical techniques that drive increased profits.  For example, predictive analytics can help companies identify useful patterns in its data by grouping customers into segments, anticipating the behavior of those customers, and generating business insights that support or even suggest strategic actions.  Companies can also use its data to identify and quantify the drivers of customer purchasing behavior, as well as their levels of satisfaction and likelihood of attrition.  Furthermore, deployment of these tools will be vital in establishing an effective CRM strategy to support sustainable organic growth.  The outputs of the models developed will enable individual companies to offer more personalized recommendations for products to their customers throughout their life cycles, thus increasing average spend, growing lifetime value, improving retention, and strengthening brand loyalty.

When it comes to incorporating analytics into business management and decision support, one of the biggest barriers for leaders is knowing how to structure the work required to extract the necessary insights and value from data. Without a structured approach, organizations often lose sight of the end goal and allow lack of priorities to become obstacles to achieving success. Every organization will have to deal with some combination of these challenges, including lack of access to information, poorly structured data sets, static and outdated analyses, and a shortage of advanced analytical capabilities.  Keep in mind that the end goal is not to have perfectly clean data sets or to have an in-house team of top data scientists. Rather the goal is to leverage data you already have to uncover actionable insights that drive profitable growth and to institutionalize a data driven foundation for decision making.  Much of the technical work along the way can be contracted to outside help as long as the desired inputs and outputs are well understood.

Viewing your organization and customers through these lenses often lead to changes in strategic direction.  It can help to build a deeper understanding of your current performance challenges, uncover new pockets of growth, and expose sources of competitive advantage. If what is happening to retail is an indication of what’s to come for other industries can you afford to stand idle and wait to see what happens next?

A Structured Approach to Help Get Things Started 


Customer Segmentation Analysis

Since the early 1960s, marketing researchers began using K-means and hierarchical clustering algorithms to model market segments; these techniques are considered the more traditional approaches, but still have applicability today. Since then many more advanced techniques have been applied to segmentation such as evolutionary algorithm, kernel methods, rough set, Taguchi method, multidimensional scaling, random forest, RFM analysis, bagged clustering, etc.  These techniques work better with consumer data with complex patterns.  Prior to model input, the data may need to be pre-processed to fit the model of choice.  Generally, multiple modeling approaches are used and may even be combined into an ensemble model; the results are then evaluated for performance and the best model is selected.

Results of the clustering method will produce more meaningful segments, and possibly sub-segments, with homogenous characteristics of consumers.  These homogeneous consumer characteristics could include demographic, geographic, psychographic, behavioral, cultural, generational similarities. In addition to defining homogenous characteristics related to the needs and preferences of each segment, business metrics such as total sales, profitability, return rates, and customer defection will be associated with the consumer groups.


In line with characterizing the consumer segments is understanding their behavior.  Part of this analysis involves churn modeling with the segmentation groups as an input variable; it’s purpose is to identify customers at increased risk of voluntary churn and identify early churn signals.  Churn prediction is useful in retail since acquiring a new customer costs 5 times higher than the cost of retaining a current customer.

Models that are used to predict and analyze customer churn include decision trees, neural networks, linked Bayesian networks to structural equation models, logistic regression, Particle Swarm Optimization (PSO), survival analysis.  The churn models that are built in the analysis are evaluated for performance metrics and the best performing model is selected.  It is important to evaluate the models’ performance for pre-leave so the company can deploy retention efforts prior to the customer dropping.


It is important to understand which products each customer segment purchased, which products are frequently purchased together, and in which sequence the products were bought either on an individual item level or product category level.

Techniques such as frequent pattern mining, association rule mining, and sequential pattern mining algorithms can accomplish this for cross-sell and up-sell efforts. A seasonality analysis is usually also important to determine at what times products were purchased most frequently in each segment.  Insights from this analysis can provide the company with customer-centric recommendations that will enable the business to create marketing strategies tailored to the different segments.

The example in the right illustrates how one consumer segment purchasing behavior may look.  The letters (A,B,C) represent different product options.  In the upper left column a single market basket purchase is represented as A,B,C (1), such as there is a strong association with product A, product B, and product C all being brought together in a single transaction.

The columns to the right represent the propensity of sequential transactions the purchaser of A,B,C (1) may make.  The information can be used to either make cross-sell or up-sell recommendations within a single transaction, or may be used to send targeted communications to the consumer for future purchase recommendations or promotions.


We Always Want to Hear from You

Let's Start a Conversation.

Contact us