Procter & Gamble reduces safety stock by 50%

How can supply networks respond to the increasing fluctuations in consumer demand? This is the question Alistair Hill, Director of Western Europe Supply Network Solutions at Procter & Gamble, answered with conviction during his presentation on 9th June at the 3-day SCL Europe congress in Prague.

“The average weekly volatility of shipments of one of our product categories went from 66 per cent in 2005 to 98 per cent in 2007,” commented Hill whilst illustrating the problem at hand. After implementing the supply chain software SAP APO, the accuracy of the forecasting went up only to drop again later. This drop was caused by increased volatility in demand and promotional campaigns.”

Procter & Gamble found the solution when they implemented Demand Sensing software from Terra Technology. This software package comprises a self-learning algorithm that can actually predict shipments after just two or three months. “On average, Demand Planning got its predictions wrong about 50 per cent of the time,” tells Hill. “By using internal data about our actual shipments and outstanding orders, we have reduced the error rate to thirty per cent. This meant that we were able to reduce our safety stock by one third. When we used customer data the prediction errors were reduced to twenty percent, halving our safety stock yet again.”

According to Hill, retailers are surprisingly willing to share their Point of Sale data with them. This most likely has something to do with the increase in the accuracy of the orders, which went up from 97.5 per cent in 2007 to 99.1 per cent in 2009. In the near future, Hill is going to be focusing his efforts at Procter & Gamble on fulfilment.