A CASE STUDY ON P&G : Types of Analytics and how P&G implemented it.

Uncertainty and an overwhelming number of alternatives are two key factors that make decision making difficult. Business Analytics approaches can assist by identifying and mitigating uncertainty and by prescribing the best course of action from a very large number of alternatives. In short business analytics can help us make better informed decisions.

There are three categories of analytics: descriptive, predictive and prescriptive. Descriptive Analytics describes what has happened and includes tools such as reports, data visualization, data dashboards, descriptive statistics, and some data mining techniques. Predictive Analytics consists of techniques that use past data to predict future events and includes regression, data mining, forecasting and simulation. Prescriptive Analytics uses input data to determine the best course of action. This class of analytical techniques include simulation, decision analysis, and optimization. 

Consumer goods giant Procter & Gambler (P&G), the maker of such well-known brands as Tide, Olay, Crest, Bounty and Pampers, sell its products in over 180 countries around the world. Supply chain coordination and efficiency are critical to the company’s profitability. After many years of acquisitions and growth, P&G embarked on an effort known as Strengthening Global Effectiveness. A major piece of that effort was the North American Supply Chain Study, who purpose was to make the supply chain in North America as efficient as possible, while ensuring that customer requirement were met.

A team of P&G analysts and managers partnered with a group of analytics faculty at the University of Cincinnati to create a system to help managers redesign the supply effort in North America. The fundamental question to be answered were:

1.      Which plants should make the product families?

2.      Where should the distribution centres be located?

3.      Which plant should which distribution centres?

4.      Which customers should be served as each distribution centre?

The team’s approach utilized all three categories of business analytics: Descriptive, Predictive and Prescriptive.

At the start of the study, data had to be collected from all aspects of the supply chain. These included demand by the product family, fixed and variable production by costs by plant, and freight costs and handling charges at the distribution centres. Data queries and descriptive statistics were utilized to acquire and better understand the current supply chain data.

Data visualization, in the form of a geographic information system, allowed the proposed solutions to be displayed on a map for more intuitive interpretation by management. Because the supply chain had to be redesigned for the future, predictive analytics was used to forecast product family demand by three-digit zip code for ten years into the future. The future demand was then input along with projected freight and other relevant costs, into an interactive optimization model that minimized costs subject to constraints. The suite of analytical models was aggregated into a single system that could be run quickly on a laptop computer.

P&G product category managers made over a thousand runs of the system before reaching consensus on a small set of alternative design. Each proposed design in this selected set was then subjected to a risk analysis using computer simulation, ultimately leading to a single go-forward design.

The chosen redesign of the supply chain was implemented over time and led to a documented savings in excess of $250 million per year in P&G’s North American supply chain. The system of models were used to streamline the supply chains in Europe and Asia, and P&G has become a world leader in the use of analytics in supply chain management.  

To conclude it can be expressed that Descriptive and Predictive analytics can help us better understand the uncertainty and risk associated with our decision alternatives. Predictive and Prescriptive analytics, also often referred to as advanced analytics, can help us make the best decision when facing a myriad of alternatives.    

Author - Kunal Patel

Visit me at - Kunal Patel