How Machine Learning Keeps Retailers Ahead of Trends

As consumers increasingly reveal their shopping habits online, retailers can access social media, purchase history, consumer demand and market trends to better understand their customers, maximize spending and encourage repeat purchases. Retailers are considered early adopters of big data technology, integrating it into every imaginable business process to achieve a deeper understanding of consumers and associated buying trends.

In order to gain an in-depth understanding about the consumer, retailers need to access and analyze all available pertinent information. And while there's an unprecedented amount of data that retailers collect regarding consumer patterns, the ability to manage and mine information from this data presents an overwhelming challenge.

Retailers are implementing technologies like Hadoop to build this big data solution, and quickly realizing that's only the start. They also need a solution that can make sense of the data in real time and provide insights that translate into tangible results, such as repeat purchasing. Machine learning technology intelligently processes massive amounts of data and automates the analysis all the way through the supply chain to make this lofty goal possible.

In the hands of forward-thinking retailers, the possibilities for advanced machine learning are limitless, from sourcing, buying and supply chain all the way to marketing, merchandising and customer experience, retailers can make significant improvements by deploying a machine learning solution. Take the example of a company trying to predict what consumers will be buying next winter. Machine learning algorithms can determine availability of materials from outside vendors, incorporate predicted weather conditions that would affect transportation or create an increased need for outerwear, and recommend the quantity, price, shelf placement and marketing channel that would best reach the target consumer in a particular area. They can even incorporate volume-based or margin-based metrics to optimize sales based on individual store or corporate objectives.

Up until now, the process of forecasting trends has been an intelligent guessing game. Retailers were forced to try to predict what would be in demand next season based on sample data of sales history, consumer demand and market trends. Today however, retailers can accurately collect and analyze structured sales history with unstructured data, including iterations of samples, lots and trial sizes to determine the optimal blend of style, color and size that would best fit their customers’ needs. Collecting and dissecting sentiment from consumers about past designs informs future planning and forecasting activities. The improved accuracy of these results delivers a category manager’s ultimate goal — accurate forecasts, increased sales and improved consumer loyalty.

Big data solutions with machine learning provide an opportunity to blend in unstructured data like social networks and call-center data with traditional cues used to design profiles, colors and trends. Supervised machine learning techniques taking all available historical information can predict trends and effects of seasonality for planning and forecasting. In addition, natural language processing (NLP) can analyze consumer sentiments and shifts in fashion trends and apply those as inputs to the planning and forecasting process.

Forecasting and planning are just some of the many ways that retailers can gain insight into their customers through machine learning and truly realize improved performance and profitability from their data. By automating complex tasks such as NLP on social data, retailers can finally get value out of the massive data sets they collect and turn it into actionable intelligence. Retailers have already begun integrating this technology into their processes to better understand their customers. The insights they get from machine learning can catalyze business and operational changes that will not only improve business-critical functions today, but also build and sustain customers season after season.

Alexander Gray, Ph.D., is chief technology officer, Skytree, an enterprise machine learning platform. Eric Thorsen is general manager, consumer products and retail, Hortonworks, a business computer software company.