Case study: Unilever Crosses the Data Streams

Computerworld |

Unilever's demand planning system is pushing the state of the art. Two and a half years ago, three business units were merged. Each used software from Rockville, Md.-based Manugistics Group Inc., but each implemented it differently. The current system, which also uses Manugistics software, blends historical shipment data with promotional data and allows information sharing and collaboration with key customers.

Statistical information is "completely ineffective for the part of the forecast driven by events, promotions, rollouts and special packages," says Raz Caciula, director of best practices at the consumer products maker's U.S. headquarters in Greenwich, Conn. The company has designed a process through which planners augment statistical forecasts with sales systems that plan promotions. For each promotion, the sales planning system predicts the "lift," or projected increase in sales, and routes it to the demand planning system, which applies it to appropriate stock-keeping units and distribution centers each week. Planners review those forecasts and make adjustments. Unilever also uses external market research in sales planning but not in the demand planning system. The company is experimenting with point-of-sale (POS) data as well. "We compare POS data and forecasts that come directly from customers with our own forecasts," Caciula says. That data helps drive the final forecast but isn't incorporated into the model yet.

Unilever collaborates on statistical and market promotion forecasts for key products with a few large customers, using a collaborative system from Waltham, Mass.-based Syncra Systems Inc. Final numbers are negotiated in weekly meetings and fed into the demand planning system.

The benefits have been worthwhile, but the process was hard. "We had a dedicated team experienced with Manugistics, and this was their life for a year," Caciula says, adding that expensive tools are only a small part of the total cost. He says the system has reduced inventory and improved customer service, but he gave no specific numbers. Caciula hopes to see dramatic improvements through more customer collaboration and use of POS data.

JUST THE FACTS

How Unilever Did It

Unilever’s system forecasting is built on the foundation of shipment history and current order information.

Internal sales projections, retail customer promotions and external market research are analyzed and combined before being fed into the demand planning system.

Retail POS forecasts and actual sales data will improve accuracy and reduce inventory lead times, but they are available only for a few large customers.

Integrating promotions data was an expensive, yearlong effort, but the resulting inventory reductions made the return on investment worthwhile.

Click image to download a printable PDF

Missing Links

Stories in this report:

ERP Systems

Robert L. Mitchell writes on a wide range of topics, including analytics, emerging technologies, green IT and data centers.

Bing’s AI chatbot came to work for me. I had to fire it.