Window on the future

"Since implementing demand management software across our production plants we’ve cut overall stock levels by 10%,” says Michiel Tonino, logistics manager at window coverings manufacturer Hunter-Douglas. “That means we’ve achieved payback within two years.” But getting there hasn’t been plain sailing. The story starts back in 1998 when the company decided it needed to investigate the need for a supply chain management (SCM) system. Why? First, the window coverings industry is a very challenging one. Fashion and changing customer tastes for products mean Hunter Douglas has to cope with unpredictable demand and a constantly changing product line. And second, the sheer size and geographic diversity of the business mean the firm operates in an environment where accurate demand forecasting and management are very difficult. Until the end of 1999, the business was generating its forecasts twice yearly using spreadsheets completed by the various sales and product managers. Apart from this method being too slow for effective management decision-making, it was also prone to human error. According to Tonino, “This inaccurate forecasting led to poor customer service on occasions… The problem of inventory overstocking was almost as frequent as stock shortages.” And downstream, manufacturing tended to compensate for inaccurate forecasts by over-producing to avoid stock-outs. This just added to the inventory problems though and Tonino says it also created higher levels of obsolete stock. So, back in 1998, the company decided that the first phase of the SCM project would be to implement demand management software across the business to try to rectify the problems. “We needed to simplify our forecasting process, improve forecast accuracy, forecast more frequently and at different levels, and make it a collaborative process throughout the organisation,” says Tonino. “There was room for significant process improvement.” The grand plan was to make forecasting information available to individuals and departments across the entire business. This would help create a single, central, accurate forecast that would drive the European operations. “We wanted to improve our knowledge of which products would sell best so that ‘slow movers’ and ‘no movers’ could be managed out of the supply chain,” he adds. Real time collaboration “The challenge was to introduce a demand management environment that could accommodate input from many levels and sources: customers, brands, product lines and accessories… The software would be used by different personnel in multiple locations, and it would have to allow collaboration in real time – almost certainly via the Internet.” It wasn’t going to be easy. The Hunter-Douglas group makes a wide range of products: window frames, flexible aluminium blinds, ceiling coverage and a range of window treatment products. 95% of products sold are custom-made, the rest made-to-stock. Headquartered in the Netherlands, the firm has 62 manufacturing plants and 83 assembly sites world-wide, plus hundreds of franchised assembling distributors. There are 14,300 employees and sales turnover last year was £1,100 million. So it’s big business. At the end of 1998, having evaluated a number of different software systems from various suppliers, Hunter-Douglas chose to implement Demantra’s demand management system, Demand Planner. But before this could happen, some integration work was necessary. The company’s existing enterprise (ERP) software, SAP R/3, had to be fully integrated with the demand management software so that sales and customer information stored in the ERP system could be transferred across once a month to the Oracle database used by Demand Planner. Another objective of the integration work was to provide a ‘data route’ for customer feedback – the assumption being, the better and more timely the information from the customer base, the more accurate the forecasting would be. The software was implemented across four manufacturing sites in Holland and took about 10 months to complete (of which two months was spent on ERP integration). Most of the work was carried out by Hunter-Douglas staff who had experience with the ERP system, but there was some support required here from Demantra too. The software went live at the end of 1999. There are currently around 20 Demantra users who tend to be either product managers or logistics staff. Tonino says the implementation as a whole was, “uncomplicated and largely problem-free.” And whilst he doesn’t reveal actual investment figures, he does state that, “Payback was achieved one year after going live on the software, or two years after the implementation work began. And there were no major network upgrades required, only a separate server to run the software on.” So two years after switching on the new software, how is forecasting done now? The integration with SAP R/3 means the company can extract information from the ERP system once a month rather than manipulating data from numerous spreadsheets twice a year to create forecasts. Sales and customer data are imported to the demand management system after a set of built-in approvals are signed off by key staff. Anyone, regardless of geographical location, can now access this forecast information over the Internet using a web browser. More accurate forecasts And the software itself allows multiple forecasts to be consolidated into a single plan. It can then simultaneously apply different forecasting models to analyse market and product data. It basically amounts to using a set of statistical methods and exponential smoothing tools (it uses Bayesian forecasting methods) to optimise the accuracy of monthly forecasts and to try to predict future demand patterns – based on historical data going back three years. And the result. “Forecast errors [measured internally using the firm’s own performance indicator calculation] have dropped by well over 10% and are still falling, resulting in overall stock reductions,” says Tonino. “And production planning now get a more accurate forecast from us each month, so they can plan things like changeover times much more effectively. “We also outsource some of our manufacturing work to suppliers, so we now share our MPS [master production schedule] plans with them… And the business can now anticipate customer orders and see demand patterns as they unfold. Using this knowledge to drive replenishment and configure logistics is the next step for us.” The business benefits have now stabilised due to people issues. But Tonino reckons there’s still more benefits to come, “once individuals using the software begin to learn to collaborate and communicate better and planners become more expert in using the software.” He cites more training and education as key factors here. “It’s these types of issues that we may have underestimated.” he warns.