Managing demand can be the differentiator between success or failure, and there can be a huge number of variables to be considered. Robert Pols looks at how manufacturers face up to the perils of planning
From augurs prodding entrails to fortune tellers studying palms, mankind was always anxious to see into the future. It still is. For manufacturers, survival may depend on reading the signs correctly, as they strive to predict demand and match it with supply. But with no crystal ball to rely on, they face a daunting task.
Demand planning is a tricky business: new products can change the patterns of demand; promotions may succeed in increasing sales, but their impact may be difficult to predict; and customers’ requirements for ever-shorter lead times put pressure on complex supply chains, especially where components are sourced from afar.
In addition, a company may have to handle tensions from within. “Many companies simply don’t have a demand consensus,” observed Möbius consultant Bert Praet. “The sales department produces forecast figures that are different from those of production, and if there’s no consensus, supply and demand become very difficult to balance.” Liam Harrington, managing partner at Oliver Wight, agreed. “Most businesses are behaviourally very poor, and they fail to work from joined-up numbers.” There are even, he added, cases where a single function produces two sets of figures. “Sales may give one set of numbers to finance and another (adjusted) set to the supply chain, because it doesn’t really trust the supply chain to deliver.”
Then there are such imponderables as the weather. A soft drinks manufacturer may plan for a 28° summer and then find a 30° interlude driving up demand by 30 per cent or more. Conversely, a company in the heating business can experience a 25 per cent increase in demand during the one year in five when there’s a particularly cold winter. Faced with these and other uncontrollable factors, it’s little wonder that many companies feel beleaguered. “In fact, some people come to see themselves as victims,” said Harrington. “But the better companies are those that decline to see things this way and seek to take control.”
One major step towards taking control is offered by forecasting software. There’s no shortage of products on the market that use sophisticated statistical algorithms to examine the sales history of a product or product type, interpret existing customer profiles, allow for factors influencing the volatility of demand, and calculate the optimum inventory to carry. Their range and versatility is well illustrated by a cross-sample of customers’ criteria. Weetabix, when choosing demand planning software, looked for a product that could accurately measure the impact of promotions, because around a third of its sales are promotionally led. Sportswear provider Russell Europe was anxious that fashion, seasonality and long lead times on Asian-sourced textiles should be taken into account. Arla Foods, with its 44 per cent share of the UK retail fresh milk market, was concerned that forecasts should be daily rather than weekly in order to reduce finished product wastage. Farmer-owned First Milk needed a package capable of handling the conflicting time-scales that result from an 18 month cheese maturation process and one-day notice for supermarket orders. Canon Europe wanted a product that would weigh up the implications of short product life-cycles, frequent new product introductions and high inventory values.
All professed themselves satisfied with their chosen software, but Canon added that implementation was to be treated as a business transformation rather than an IT project. Such a sense of perspective is just what Liam Harrington advocates. “It’s like the old golfing story of the man who plays once a year, always appears with a new set of clubs, and has the worst swing imaginable. I need a golf club to hit a golf ball, and I need software to handle forecasting. But there are many companies that have good software yet produce poor forecasts.”
The solution, in his view, is sales and operations planning (S&OP), which was invented to help companies change their behaviour patterns. “When properly managed, it’s used to run the business rather than simply as a supply chain process, and it allows an organisation to achieve alignment between its business plan and its volume plan.”
In essence, S&OP enables manufacturers to take a high-end view of all critical data and so determine the best route for balancing supply and demand. “If your forecast is too high,” Bert Praet pointed out, “you’ll manufacture more than you can sell, while underestimated forecasts lead to a shortage of products and unhappy customers. S&OP tries to avoid both situations. In a global company there are several levels of planning, from strategic planning (with, typically, a five-year horizon and buckets of a whole year), through planning at SKU level (which looks three to six months ahead with buckets of weeks and days), and on to the most detailed plan (which thinks a couple of weeks ahead and has buckets of single days or even hours and minutes). But, before the S&OP concept was developed, people complained that they drew up brilliant strategic plans and then had difficulty in carrying them out.”
What was generally missing, he argued, was medium term planning at product family level, where conflicts between broad view and detailed execution can be reconciled. As an example of the S&OP process in action, he cited the system that Möbius worked on with Tate & Lyle Food and Industrial Ingredients Europe (as it was, before the recent sale of its starch business).
Because Tate & Lyle had a multi-site operation, its new medium-term planning activity was divided into two stages – operational group planning and operational local planning – and a planning cycle with three broad phases was instituted. This starts with the preliminary processes, including all the data-crunching and the initial balancing; it moves on to a demand and supply consensus meeting; and it concludes with the translation of the consensus plan into plans for sales, production and inventory.
Increasingly refined forecasts are generated by a rigorous procedure that follows a ten-step monthly cycle. This includes statistical forecasting, two levels of rough-cut capacity planning, adjustments by a product application manager and the creation of a forecast dashboard. The dashboard graphically represents historical demand and forecasts, and it provides forecast accuracy figures at product family or sub-family level. As a result of this new system, Tate & Lyle has reported an improvement in forecast accuracy of around 10 per cent and has been able to downsize its safety mechanisms. Yet, Praet added, this was all done with minimal IT investment. “No huge integration of software tools was involved in this programme. The project was supported by simple adjustments to their existing software.”
Heinz Continental Europe, with help from the Oliver Wight consultancy, has also introduced S&OP as the context in which demand management is handled. Here the major steps are product management review, demand planning review and supply planning review, and these come together in an integrated reconciliation activity that’s concluded by a management business review.
The first and basic stage in their demand planning is to analyse good quality data, maintaining statistical models for all products at account level and reviewing the accuracy of each model every month. Then account managers meet with members of the demand planning team who are aligned to their product categories. The refined forecasts from these meetings are submitted to an integrated reconciliation revue, chaired by regional demand planning manager Ko van Brakel. Here, those involved in the previous steps come together with finance management to examine forecasts from a top-down perspective and assess the implications of trends and changes. Finally, a demand review meeting gives each category team the opportunity to present its latest forecast for management approval.
But the planner’s work isn’t simply a mechanical progress through a series of meetings. “Demand planning is a difficult job that involves working with others across the organisation,” van Brakel insisted, “so you need smart people. This company has invested in a team of high calibre people to work at quite a high level within the organisation. As a result, others appreciate our view and work well with us. That’s very important, if you’re not to become simply a postman who receives information and passes it on.”
Indeed, he concluded, the relationship between demand planning and other functions needs to be carefully developed. “It’s important that the methodology we apply is understood by the people we work with. We try to have a transparent process to avoid demand planning becoming a black box. This strengthens the acceptance of our activity throughout the business and it helps us achieve the necessary consensus in our meetings.”
So, there is no magic crystal ball; but crystal balls are probably overrated. Instead, there is software, which (given high quality data) can do a much better-informed job, but which is only one element of a coherent planning process. As for the other components, van Brakel was in no doubt. “For demand planning to make the effective contribution a company needs, you require a com-bination of good people and good systems.”