Powered by AI - Cello Forecasts the Future

How Cello Helps Sales Forecast for a Retailer?

1. Importance of AI-based Big Data Analysis

“Technologies behind the Fourth Industrial Revolution are shaking up companies across industries,” Klaus Schwab, Executive President of the World Economic Forum, said in January 2016. “Since innovative companies have access to research, development, marketing, sales and retail through digital platform, they will be able to outperform the companies currently leading the market,” he argued. As Mr. Schwab explained, the Fourth Industrial Revolution will completely reshape a business paradigm much greater than the previous three industrial revolutions combined. Today, this Fourth Industrial Revolution is mostly led by information and communication technologies such as Artificial Intelligence (AI), Internet of Things (IoT), big data analysis, self-driving vehicles, robots, shared economy and 3D printers.

AI, in particular, is the key to aggregating and analyzing high volumes of data. It also creates innovative business values by utilizing new data from IoT, self-driving cars or smart factories, etc. That is why AI-based big data analysis stands out the most among technologies driving the Fourth Industrial Revolution.

The rise of the Fourth Industrial Revolution has brought an era of convergence in which industry boundaries are increasingly disappearing. In the logistics industry, too, convergence1) is well underway, with logistics combined with information communication technologies. Keeping up with this trend, global companies have started adopting this new convergence logistics service to increase market share and competitiveness, which raises the need for Korean companies to come up with ways to respond.

Samsung SDS, too, has launched a demand forecasting service, Cello Demand Sensing. It provides forecast by adopting AI-based big data analysis. In this paper, we’d like to introduce Cello Demand Sensing and how it’s actually been used to forecast sales of a German retailer as many manufacturers and retailers have shown a keen interest.

2. Cello Demand Sensing

What is Demand Sensing?

Demand sensing is a forecasting methodology that leverages new mathematical techniques and analyzes near real-time information on sales history or prior promotional activities to create an accurate demand forecast.

Cello Demand Sensing is a demand forecasting system that provides weekly sales forecast for customers by combining Samsung SDS’ total logistics platform Cello together with its SCM solution technologies and Brightics, Samsung SDS’ own big data analytics engine.

Brightics supports 126 functions as well as both machine learning2) algorithm and deep learning3) algorithm.

Project Case: Sales Forecast for a German Electronics Retailer

In this project, we analyzed sales history and previous promotional activities of a German retailer and suggested the most optimized demand forecast to make its entire supply chain more efficient.

To do this, we first analyzed three-year sales and promotion history of the retailer’s 19 products in 428 stores. We also looked at the correlation between sales and 30 parameters including customer benefits, weather, holidays, population and level of income and how sales changes in response to 9 types of promotions such as TV ads, voucher, cashback, etc. Through the process, we let our forecast engine to analyze sales using machine learning. We also let the engine to predict sales volume by itself through decision tree6) algorithms, sort out sales ratio product by product and figure out consumer purchase patterns. These were all reflected to create a final forecast.

Then, by comparing a set of sales forecasts with previous actual sales figures, we selected what was the closest to the actual sales as our final forecast.

How Accurate was the Forecast?

In Germany, superdeal promotions were held for two weeks from the last week of November to the second week of December in 2016 (47th week ~ 49th week). During this superdeal period, we carried out a pilot demand forecasting for 19 types of televisions in 428 stores of the German retailer, using Cello Demand Sensing.

It turned out that while the forecast by the sales person in charge was 53% accurate, our Cello Demand Sensing showed 72% accuracy.

Since the pilot forecasting last year, Samsung SDS has been applying forecasting on a cloud basis to other electronic appliances such as refrigerators and washing machines. To ensure a consistent and reasonable forecast, our Cello operation team has a weekly meeting with the client in Germany. With our forecast, the German retailer plans to adjust inventory to be sold in stores, while manufacturers use the forecast as their sales strategy in connection with promotions of its marketing team.

What Benefits Does it Offer?

For more details, you can download a White Paper from Samsung SDS Insight Report.



▶   The contents are protected by copyrights laws and the copyrights are owned by the creator and Samsung SDS.
▶   Re-use or reproduction as well as commercial use of the contents without prior consent is strictly prohibited.