Managing product introductions across the supply chain: findings from a development project

Transcription

1 Managing product introductions across the supply chain: findings from a development project Jan Holmström, Hille Korhonen, Aki Laiho and Helena Lakervi jan.holmstrom@hut.fi Industrial Engineering and Management Helsinki University of Technology POB 9500 FIN HUT Finland 0 / 21

2 Managing product introductions across the supply chain: findings from a development project Abstract An increase in product variety and the number of product introductions make demand-supply balancing processes labour intensive and prone to mistakes. This is especially the case when there are many levels in the supply chain and the lead-times for material supply and production are long. The paper presents a concept developed to improve the productivity of the planning effort for a consumer goods manufacturer. The proposed planning process takes into account that the manufacturer has products at different stages of the product-life and utilise sales and inventory information collected from distributors and retailers. Trials in the case company indicate that supply chain responsiveness could be improved in product launches using the proposed concept. Supply chain efficiency in the maturity phase could also be improved. However, the usefulness and effectiveness of the developed concept depends on the assumption that product mix changes can be modelled and point-of-sales and channel sell-through data is available regularly and reliably. Key words: demand-supply planning, planning efficiency, demand visibility, point-ofsales Introduction Information technology solutions developed to support the material and production planning processes of manufacturing companies typically deal with items in a uniform manner (Euwe and Wortmann, 1997). Difficult phases in the product lifecycle, such as the introduction and ramp-down, are handled using calendar based planning procedures. Expert-based forecasts are used as input in the product introduction phase, while statistical inputs can be used only for products that have a sales history available. The benefit of managing the supply chain across several organizations was first recognized in the early 1960's (Clark and Scarf, 1960; Forrester, 1961). Today, effective supply chain management is of paramount importance for many manufactures. The reason is an increase in product variety and the number of product introductions. This makes planning processes labor intensive and prone to mistakes when there are many levels in the supply chain and the leadtimes are long. 1 / 21

3 The importance of managing demand over the lifecycle is illustrated by the following example. In consumer goods, many product variant specific materials have long lead-times, months rather than days or weeks. In this situation, when the initial forecast is too low, then the cumulative shortages over the material supply lead-time result in significant lost-sales. Similarly, when actual demand is lower than forecasted and this is not identified in time, then the result is a high cost of obsolescence and high inventory holding costs over the lifecycle. Long lead-times, and slow reaction times, are not uncommon in supply chains for complex consumer goods. In fashion goods and seasonal goods the lifecycle in relation to purchasing and production lead-times may be so short that there is no time at all to adjust supply to actual demand (Fisher, 1997). However, at the same time it is increasingly feasible to get channel visibility through collaboration with retailers and distributors. Retailers and distributors have invested in information systems that collect and aggregate sales and inventory information, while at the same time Internet and Extranets make it easier to share the collected information with manufacturers and suppliers. Product data synchronization has also made it increasingly feasible to share demand information from the points of sales with suppliers. For example in the USA synchronization is underway around the UCCNet initiative (UCCNet, 2003). How could manufacturers use the improved demand visibility to improve planning in situations when lead-times and capacity, at least in theory, would allow a manufacturer to react to market demand? Literature review Managing the demand-supply network The concept of inter-company collaboration, especially in the area of supply chain planning and demand forecasting, has received significant attention over the last decade (Lee et al, 1997; Fisher et al., 2001). The 1980 s saw the emergence of initiatives such as Just-in-Time and Quick Response that emphasized information sharing, partnerships and new technologies as ways of improving logistical performance (Ko, Kincade, and Brown 2000; Vokurka and Davis 1996). By developing processes that make it possible to adjust plans and forecasts in a collaborative fashion, supply chain parties aim to make it easier to take into account events, such as promotions, new product introductions or assortment changes, that affect demand throughout the supply chain (Barratt and Oliveira, 2001). 2 / 21

4 According to Lee et al. (1997) improved channel visibility means providing each stage of the supply chain with reliable information about demand and supply. Initiatives such as Vendor Managed Inventory (VMI) and Collaborative Planning Forecasting and Replenishment (CPFR) are usually developed between two stages of the demand-supply network (Lewis et al, 2001). The two-stage information sharing can be generalized to multi-stage information sharing, or n-tier CPFR. Here, the number of demand-supply network stages involved in collaborative information sharing is not limited to two successive levels. The purpose is that one set of numbers for forecasts, inventories, orders, and deliveries is used throughout the demand-supply network to improve both demand and supply flows. Only small-scale pilots of multi-stage collaboration between independent supply-chain parties can be found. The most promising area is in forecasting collaboration. Collaborative forecasting makes it possible to take advantage of the expertise of several, supply chain members. One benefit that is suggested to follow from this is a reduced reliance on historical records (Helms et al., 2000). Through collaborative forecasting, a company or department strives to get access to useful information during product introductions and special situations, such as promotions (Seifert, 2002). Furthermore, working based on one shared and collaborative forecast reduces the problems related to what Mentzer et al. call the "islands of analysis" phenomenon. The problem is that different groups, departments or companies develop their own forecasts independently of each other according to their own specific planning needs. The different organizations risk ending up acting based on contradictory plans, even when they share the common goal of providing the consumer with the products that they require (Helms et al., 2000; Mentzer et al., 1997). Collaborative planning, forecasting and replenishment has been more difficult to implement in practice than expected initially (Småros, 2003a). The practical challenges in developing collaborative planning, forecasting and replenishment indicates, according to AMR Research (Bacon et al, 2002) that the practice of collaboration in industry is very different from the vision of CPFR. Typically collaboration with customers and suppliers are within the functional silos of a supplier's sales and a customer's purchasing, or between two product development departments. There is evidence of very little co-ordination of collaborative practices within the firm, such as using channel visibility for improving production and inventory control. The conclusion made by AMR is that firms need to organize better internally the coordination of collaborative practices between functions to benefit from their development projects with customers and suppliers. For example, channel management in practice ignores 3 / 21

5 product lifecycles (Asgekar and Columbus, 2002). Companies that have invested in developing channel management information systems do not co-ordinate the launch of complex products and services, nor announce pricing changes, nor reposition products, and do not distribute competitive information. Channel management is mostly used to track and escalate sales leads, as well as to track orders in the supply chain. Tools and processes for managing the demand-supply network Channel visibility here can be defined as the manufacturers and suppliers getting access to sales and inventory information collected by retailers and distributors. The issue is what tools and processes are available to make efficient use of this information for managing the demand-supply network? The tools and processes available from technology providers are typically extensions of Enterprise Resource Planning (ERP), such as "available-to-promise" and "sales configurators". The question is, however, how well ERP extensions can provide the needed basis for coordinating complex products and services in a network of several independent organizations? Today, the term advanced planning and scheduling is typically used for software products that extend ERP to the distribution and supply chain area. DeKok and Fransoo (2002) as well as Fransoo, Wouters and dekok (2001) have analyzed these supply chain planning solutions and concluded that the development focus has been on improved optimization virtually to the exclusion of practical implementation challenges. This focus on optimization can be considered problematic. Optimization algorithms typically require accurate and complete data to function correctly. In practice, manufacturing companies can seldom achieve this state of affairs more often than once a month or once a week (Euwe and Wortmann 1997). It is also difficult to include customer or supplier information in the plans since the data available is not accurate enough. In addition, even if accurate information is available from some suppliers or customers, it is not available from all partners, which mean that the completeness requirement is not met. This is also the typical situation in collaboration efforts such as Vendor Managed Inventory. In general, only part of a company's customer base is involved in VMI collaboration, making it difficult to utilize the additional channel visibility available from the partners (Småros et al.,2003). The leading edge in the use of channel visibility can be found in retailing (Fisher et al, 2001) where innovative approaches have been developed to use early sales to improve forecasts, and to optimize the trade-off between lost and obsolescence in the channel. There are also other novel approaches in development, such as the use of "rank and share" to forecast the impact of 4 / 21

6 product assortment change (Småros and Hellström, 2003) and the use of advance demand information in planning project deliveries (vandonselaar et al., 2001). Methodology The results of the literature review demonstrate that research presenting practical planning concepts and examples for using channel visibility is difficult to find. Particularly, solutions that would reduce manual planning work at the introduction stage of the product lifecycle and help streamline purchasing and production scheduling in the maturity phase are absent. This is a gap in the academic body of knowledge that has major implications for managerial practice. There is a need for research on how to streamline supply chain planning using channel sellthrough (or Point-of-Sales) information in the assembly industry supply chain. In particular, better understanding is needed on what information is useful for which type of organization in which stage of the product lifecycle. It is also essential to understand better the details of how the information can be used by different organizations in different stages of the lifecycle. To study this problem a case research approach was adopted. The research problem was first identified in a previous publicly funded research project with strong industry participation ( During this research project it was realized that the primary issue for many original equipment manufacturers was not how to get better visibility to downstream demand, but how to actually benefit from already available information in production and supplier operations. Based on the results and experiences of previous research efforts in the case company it was also clear that the use of channel visibility for production planning and for supplier collaboration is difficult to study. Previous attempts at answering the research question by university affiliated researchers using a conventional case study research process did not succeed. The reason was that the researchers could only with difficulty locate the appropriate persons in the complex demand-supply network. Also, many times the researchers did not recognize that when respondents were asked about current practice the answer instead described intended practice. On the other hand, it was also recognized that people from the internal research and development organization would run into similar problems of distinguishing between actual practices and intended practices when searching for suitable solutions concepts in other industries and environments. For this task the use of university researchers were found to be effective. To study the problem a team was set up that consisted of researchers and practitioners. The practitioners were selected from the manufacturer's research and development organization, 5 / 21

7 from two key component suppliers and the manufacturer's sales and distribution organization. The researchers were selected from a university research institute. The selection criteria were relevant practical experience and academic ambitions in form of Master's and Doctoral Theses. Out of the core team of 10 persons 7 were actively pursuing their own thesis projects. The research was carried out within the extended partner network of a case company. The case company is a manufacturer of a popular line of durable consumer products. The company is increasingly dependent on its ability to build and rejuvenate its product lines according to seasons and consumer fashions. The extended network covers component suppliers, assembly operations, sales and distribution, and retailing. The demand-supply network is complex ( Figure 1). It consists of 80,000 retailers and other points of sales (POS), 1000's of distributors and retail chains, 17 sales units, 3 Original Equipment Manufacturer (OEM) plants, 2 Contract Equipment Manufacturer (CEM) plants and up to 150 component suppliers. Suppliers (~150) OEM plants (3) + CEM plants (2) Sales Units (17) Trade customers (~1.000) POS (~80.000) Distributors Retail central warehouses Figure 1: Demand-supply network structure and number of players The research project was carried out during a six-month period from October 2002 to April The main stages in the research were a benchmark study, analysis of current processes and problem areas, concept creation, piloting and synthesis of results from both an academic and practitioner point of view. The analysis of current processes and problems was carried out through interviews and quantitative analyses that tracked the actual information flow from markets to suppliers. In the piloting phase the created concepts were tested in parallel with 6 / 21

8 current processes using the information available in the extended network. In the pilots a weekly feedback and consolidation loop was created to present the results and progress to all members of the core research team. This made it possible to identify in time and place problems due to data quality, speed of updating and sharing information. In the synthesis of the results the business management from both the manufacturer and supplier were involved through the project's steering group. This way concept elements could be selected for further development based on both the criteria that they are needed to achieve visibility and that they bring tangible business benefits within a reasonable time-frame. Concept development The number of product and variant introductions has increased fourfold over the last 5 years in the OEM's demand-supply network. The increased number of product and variant introductions has had the effect that a lack of coordination between the different levels of the supply chain causes many unnecessary and abrupt changes in production and purchase plans. A lack of coordination between different players in the supply chain also extends the duration and difficulty of product introductions. The most difficult aspect with product introductions is responding to the demand for different variants. The problem is how to quickly update the manufacturer's and key suppliers' plans with accurate information on demand and consumer preferences for different variants (e.g. demand of different color, or casing variants) when the product is being introduced to the market place. For example, products may first be introduced into markets with special variant preferences, but this is not communicated to suppliers, who are taken by surprise when the demand for different variants changes over the lifecycle. Or, some product variants are not introduced until later due to supply constraints, but the supply constraint for a variant is not communicated to all suppliers, which first are disappointed by unexpectedly low demand and later are taken by surprise when demand for a variant suddenly takes-off. Development of best practice in the retail industry was reviewed in the search for potential solutions for more managing the increased number of product introductions and variants. Process innovations such as Fisher et al.'s use of early point-of-sales data to forecast total sales ( Fisher et al., 2000), Småros's (2003a) use of the product lifecycle to select appropriate level of forecasting collaboration, and Salmi et al's use of distributor sell-through to model product mix (2002) were identified as the most relevant practices to be explored further. Based on analysts' reports (Asgekar et al.,2002; Bacon et al. 2002) it was decided at this stage to not explore further the available commercial solutions. This decision was based on the 7 / 21

9 assumption that for managing the product mix and variants across the network there are currently not available any commercial alternatives. Advanced scheduling and planning systems are not seen as an alternative. The company already has such systems in place, but they are not addressing the problem of an increasing number of product introductions and variants. Planning according to the product life cycle: Sense and respond An initial solution concept was formulated based on the use of ramp-up curves to model product introductions, variant mix views to model relationships between variants, and using sell-through and end consumer sales as inputs to streamline forecast and requirements planning. The objectives of the solution concept is to improve planning efficiency and supply chain performance in a situation when the number of product introductions is increasing and to improve supply chain efficiency in terms of reduced lost sales due to fewer supply disruptions. The first phase in the life-cycle, ramp-up of production, is by necessity based on plans and forecasts. This is a necessity because there is no advance information on end customer demand. In this stage the important communications link is between sales and marketing, production and suppliers. The visibility focus is on updating and communicating supply constraints when forecasts and product mix estimates are revised based on feedback from important distributors and retail customers. The transition to the following phase - the market introduction - starts when monitoring of the channel and end customer demand can be started. However, plans cannot immediately be adjusted as information is gathered from the channel and the points-of-sales. Care needs to be taken to avoid unnecessary changes that only induce nervousness in the chain and prolong the market introduction phase. Figure 2 illustrates sensing the actual market demand using channel visibility in the market introduction phase. How quickly adjustment of plans is possible depends on the quality of and type of information on end customer demand. To facilitate the adjustment process information can either be directly collected from the point of sales or indirectly by monitoring channel inventories. The access to point-of-sales is potentially more useful because it is not distorted by ordering and inventory management decisions. However, the collection is complicated by the larger number of point of sales locations that need to be involved and monitored. 8 / 21

10 Cumulative volume for product (or variant) History Future Supply plan Demand forecast Channel visibility Rampup Market introduction Time Figure 2: Ramp-up production and monitor the success on the market Figure 3 illustrates the adjustment of supply plans and forecasts to market information. In the situation illustrated by the figure, demand visible through the channel is significantly lower than the demand forecast. Both the supply plan and the forecast are now adjusted according to the visible end customer demand. The supply plan is temporarily postponed, and the forecast adjusted downwards. The objective with such an adjustment is to maintain a high service level at a low risk of obsolescence, and to enable an early switch to the next stage in the planning cycle. Cumulative volume for product (or variant) Sense and respond Adjust supply plan and forecast based on channel and POS visibility Switch to streamlined planning Supply plan Demand forecast based on channel visibility Switch to end-of-life planning Market introduction Rampup Business-asusual Rampdown Figure 3: Respond to the success on the market The next stage, that is business as usual, is reached when supply and demand for the new product is in balance. Now information on demand from the channel can be directly used to drive supply. When supply capability is in balance with the pull of end customer demand, it becomes possible to streamline and simplify the planning process by using replenishment 9 / 21

11 logic for production and material control. The goal is to ensure availability in the points of sales, and reduce inventory levels in the channel. The final phase - ramp down - starts when production and materials supply plans are phased out to prepare for the entry of a new product and to minimize obsolescence costs. The objective is to only produce and purchase those materials that are needed to complete the products for which there already exist expensive long lead-time materials. Product introduction: Variant mix planning and updating The developed demand-supply planning concept proposes that sales units switch from variant level planning to product level planning. While sales units concentrate on total demand for the introduced products a centralized supply chain planning and control function takes over the responsibility for variant planning. Even though sales units would plan on product level they would keep ordering on variant level. After the change, the variant mix planning would be done closer to production and the suppliers based on demand information collected from the distribution channel. In addition to reducing planning work the concept is expected to improve the quality of plans to the supplier. This is an expected result of monitoring the variant split in sales to the consumer instead of aggregating plans made by key accounts and sales representatives in the sales units. The goal is to use the improved planning efficiency in the supply chain to focus sales management attention on updating product level plans in the market introduction phase. For example, improved planning can be translated into improved supply chain efficiency in terms of reduced lost sales if faster updates lead to fewer supply disruptions when the product is launched. Improved accuracy also reduces misallocation of available product, costs due to the more efficient use of resources, and improves the timing of production and distribution decisions in the demand-supply network. The variant mix model plays a critical role in the developed concept. Planning and forecasting based on a centralized model of the variant mix is possible under the following conditions: the variant mix on the market remains stable independently from total demand for the product and, changes in the variant mix can be predicted based on new variant introductions and ramp-downs Figure 4 illustrates a situation where variant mix and total demand changes independently (Småros and Hellström, 2003). In the supply chain studied by Småros the end-consumer 10 / 21

12 demand varied dramatically. The product was a pick-and-mix shelf with up-to 160 different variants of sweets. As can be seen in the top chart consumer demand varied between a minimum of 30 units a week and a maximum of 250 units a week. However, at the same time the distribution of demand between product variants remained stable. The bottom chart shows the relative share of total demand for the 160 product variants. The developed planning concept attempts to leverage a stable or slowly changing variant-mix profiles to centralize the variant planning in the case demand-supply network. The principle is to focus planning efforts in the sales companies on the product level, which is highly variable and depends on the lifecycle phase. Variant specific demand is then managed through modeling and monitoring variant mix profiles centrally. 300 Indexed Sales [kg] Total sales / week Selected weeks _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _2002 Weeks 3,00 2,50 2,00 1,50 1,00 0,50 0, Share of sales (%) Week 001_2002 Week 007_2002 Week 013_2002 Week 015_2002 Week 020_2002 Week 043_2002 Week 049_2002 Products ranked in ascending order Figure 4: Variable total demand and stable mix in an example supply chain (Småros and Hellström, 2003) In the case demand-supply network this simplification of planning has a most dramatic effect on the sales company level. For example, instead of planning more than 200 variants every week the sales organizations can plan 50 products and leave the monitoring and adjustment of the variant level to the central supply chain planning and control function. The increase in 11 / 21

13 efficiency is multiplied by the number of sales organizations. For a network with 10 sales organizations the potential reduction in planning is from 200 weekly plans in 10 organizations to 50 plans in 10 organizations. In total this is a reduction from 2000 weekly plans to 500 for the sales units. On the supply chain level the aggregation work for producing the 200 weekly variant plans needed by suppliers can also be simplified, as fewer plans means fewer problems with data quality. To test the feasibility of this concept in ramp-up and product-introduction phases two products being introduced on the market were selected for detailed analysis. The product variants analyzed were color variants. The analysis was based on early-order, delivery, channel inventory, and channel sell-through information. The concept can be regarded as feasible for further development if the variant-mix converges to a stable split between variants, and this split can be identified using demand information from the channel. The assumptions of the proposed concept were then tested through the following analysis and pilot: past ramp-ups and early order information was analyze to estimate the variant mix before product introduction inventory-sell through was monitored to establish the variant mix during market introduction Analyze past ramp-ups to estimate the variant mix before market introduction Before the market introduction phase, for the ramp-up, the OEM needs to forecast the variant mix between colors for the new product. This forecast is needed 1-5 months before the product is introduced to the consumer depending on product type. The historical color split and early orders from a type of retailer with special promotional requirements are the only available sources of information before the product is introduced on the market. An analysis of this information was performed for the two example products. For example product A, the information is presented in Figure 5. The color options for the example product are yellow, blue and gray. Before the analysis of early orders and the search for reference products, marketing had estimated the color split to be yellow 20 percent - blue 50 percent - gray 30 percent. 12 / 21

14 There were no past products with exactly the same color options. In the second step of the analysis in figure 5 a product with yellow, blue and light blue colors was chosen as a historical reference. In addition, early order information was analyzed from one distribution channel. Early orders of a specialized retail channel indicate that gray will be more popular than blue. Early orders also indicates that yellow has only marginal demand. However, this is not reliable information as each retailer usually only orders one color variant in advance. Here, the reference product is a better indication. To sum up, the analysis indicates that gray will sell better than blue, and that the estimate for yellow is slightly too optimistic. The color split based on early orders and the reference product indicate that the estimate should be yellow 15 percent - blue 35 percent - gray 50 percent. 2. Color split of a product with similar colors 100% 1. 80% 55% 3. 60% 100% 80% Originlal estimate 30% 40% 20% 0% 29% 16% 100% 80% Updated estimate 50% 60% Yellow Light Blue Blue 60% 40% 50% Color split of pilot product A in early orders 40% 35% 20% 0% 20% 100% 80% 20% 0% 15% Yellow Blue Grey 60% 70% Yellow Blue Grey 40% 20% 26% 0% Yellow Blue Grey Figure 5: Original estimate, reference products and early orders, and updated estimate for example product A For example product B there were no historical products with same type of colors. Color split information was only gathered from the early orders database from the group of retailers with special promotional needs. The split in the database showed that 99 percent of the ordered variants were blue. In addition to blue, the product is made also in plum. Based on this information, it was estimated that the blue will be distinctly the dominant color. Marketing had estimated the split to be blue 70 percent - plum 30 percent. 13 / 21

15 Monitor inventory-sell through to establish the variant mix during market introduction In order to simplify planning through variant mix monitoring it would be desirable that the sales units could plan on product level, and that the variant mix planning could be done separately based on centralized variant mix models and monitoring the variant mix of sales to consumers. In practice, variant mix planning could be done in connection with production and material requirements planning if the variant mix can be reliably modeled and monitored. In theory, the color split could be monitored on three levels of the network simultaneously: the OEM s shipments to the distribution channels, the sell-through from distribution inventory and from the point of sale. The sell-through from distribution inventory was gathered weekly for the two example products during the product ramp-up and introduction to the market. For product A the variant-mix observed from the distribution inventory sell-through started out with equal shares for the three colors (Figure 6). Then, blue gained share from yellow and the split was approximately the same as had been forecast by marketing. However, when the large retail customers, with their promoted sales, started ordering the split changed again. The gray gained market share from blue. This development was as forecast based on the analysis of early orders. Seven weeks after introduction the variant mix is approaching the variant mix estimated in the ramp-up phase, with yellow variants moving slower than blue, and blue moving slower than gray. In this studied product introduction it seems that estimating the variant mix based on previous similar introductions and early order information, and actively monitoring the variant mix development may be helpful to move the demand-supply network more quickly into the maturity phase. The accuracy reached by estimation and active monitoring is also better than the original estimate made by marketing and the accuracy previously reached by the sales units through the weekly planning process on the variant level. 14 / 21

16 Trade customer sales to POSs Color split share 100% 90% 80% 70% 60% 50% 40% 30% 20% 27% 32% 40% 20% 29% 50% 12% 11% 6% 3% 21% 28% 49% 51% 67% 61% 45% 45% Yellow Grey Blue 10% 0% +1 wk +2 wks +3 wks +4 wks +5 wks +6 wks Weeks after start of sales Figure 6: Weekly variant-mix observed in inventory sell-through for product A during introduction phase. The example product B has had a difficult product introduction. The first orders from the dominant retail and distribution channels were not confirmed until the fifth week after sales start. Due to these difficulties the total sales volume has been significantly below the forecast and the impact of promoted sales in the dominant channel was not visible in the monitored variant mix at the end of the seven-week test period. Reaching maturity is taking significantly more time for this product than for the first example product. The variant mix graphs presented here were during the seven-week test period analyzed weekly on Wednesdays together with the launch managers. A practical problem that was identified was the slow reporting speed of inventory figures from retail and distribution customers. On Wednesday, the coverage of the previous week s reports was so low that it was difficult to draw conclusions. It was also noticed that for market introductions the coverage is in general worse than for the coverage for mature products. After several weeks of shipments to the channel, the reported sell-through was only per cent of shipped volumes. The practical testing thus also revealed the importance of improving speed of reporting and coverage with those customers that start sales first. Business as usual: Streamline planning using demand visibility A further improvement in supply chain efficiency can be gained when end-consumer demand can be used to control material supply, i.e. control supply by demand pull. 15 / 21

17 This is possible only after the product and variants have been successfully introduced on the market. In other words, a switch from a push-mode to a pull-mode can be made in planning after the market introduction if there is reliable channel visibility. The basis of the developed pull-concept is to actively use the inventory buffer that has accumulated in the supply chain by the maturity phase to simplify planning processes. Instead of planning purchases and the assembly of modules, these activities are performed according to the sell-through downstream in the chain. As long as the lead-time for delivering components and modules is short compared to the days of supply in the chain, then forward planning can be replaced by monitoring consumption downstream and replenishing OEM and CEM component buffers. To test the feasibility of using down-stream demand to control material replenishments an analysis using inventory sell-through collected from retailers and distributors was performed. The analysis included three levels of the supply chain: retail, distribution and production. Country specific channel inventory Quantity Inventory Repl. quantities Sell-through Weeks Figure 7: Streamline material planning by replenishing to sell-through The results of the analysis is illustrated in figure 7. The figure shows the sell-through as monitored on the distributor level. The sell-through is the out-flow of products from the distribution warehouses to the retail outlets. The replenishment quantities are the material supplier replenishments to the OEM and CEM assembly operations, and at the same time the replenishment quantity to the distribution level warehouses. The inventory in figure 7 is the inventory on the distributor level. The replenishment lead-time from suppliers is three weeks. 16 / 21

18 The analysis showed that the replenishment concept could be effectively used in the maturity phase of the product lifecycle with the example product. When the channel sales are moderately stable, the replenishment solution does not result in stock-outs and the needed channel inventory levels are the same or lower than current ones. In this phase of the product lifecycle this automatic replenishment model could lower inventory carrying costs, forecasting costs, and ordering costs. However, if a peak in demand occurs, the model does not work well. Because the peaks in the future cannot be anticipated from the current sales information, these must be managed using input from sales units or customers. In the analysis a peak in demand in week 33 resulted in a severe stock out. The peak was due to a promotion, which would have been necessary to anticipate by the sales unit personnel. Conclusions The results from the development project indicate that the quality of variant forecasting can be improved with access to channel visibility especially in the market introduction phase using the variant mix modeling. The supply chain responsiveness to mix changes is improved in product launches by removing manual planning steps from the sales units and instead model and monitor the variant mix on the total market level. Supply chain efficiency in the maturity phase can also be improved through a shared market overview and the simultaneous use of available demand information at several levels of the chain. However, the usefulness and effectiveness of the developed solution concept depends on the following assumptions being true in a manufacturer's demand-supply network: Ramp-up curves of similar products and historical product mix changes can be used as input to variant-mix planning during ramp-up Point-of-sales and channel sell-through data can be used to update sales volume targets during ramp-up and as input to variant-mix planning. Suppliers can benefit from improved visibility to sell-through of mature products in streamlining requirements planning In the market introduction phase the monitoring of the market mix needs to involve personnel from both customers, sales units, production units and key suppliers. In the development project a weekly review cycle was used and the involvement of experts was essential to spot and interpret the results from monitoring. 17 / 21

19 When entering the maturity phase, the monitoring of variant mix need not be as intensive as during market introduction. The variant mix stabilizes after the large retail and distributor customers have started their sales. It may even be adequate to monitor the variant mix on a monthly level and include the detailed analysis and decision making into the monthly demand-supply planning process. The variant mix information could also be added to the monthly plans that are currently provided to suppliers only on the aggregate product level. There may be situations in which the color split changes significantly during the month for mature products. Further analysis is needed to evaluate the need for monitoring of the variant mix for mature products. For example, how to deal with customer-specific campaigns and supply constraints. On a general level, the case illustrates the complexity involved in the co-ordination of collaborative practices and for using channel visibility for improving production and inventory control. External collaboration initiatives such as Vendor Managed Inventory (VMI) and Collaborative Planning Forecasting and Replenishment (CPFR) are not sufficient on their own to produce improved efficiency and responsiveness. Firms also need to actively co-ordinate internal collaborative practices between functions to benefit from their development projects with customers and suppliers. The adjustment of internal process when better external information becomes available is an area requiring more research. For example, how does the issue of adequate coverage of external demand information sources depend on the distribution channel structure? Or, how could the monitoring of different demand sources be automated for production and inventory control purposes? References Asgekar, V and Columbus, L. (2002) "Channel Management Software Best Practices: It's All About Orders", AMR Research, September 2002, Report Bacon, A., Lapide, L., and Suleski, J. (2002), " Supply Chain Collaboration Today: It s a tactic, not a strategy", AMR Research, September 2002, Report Barratt, Mark and Alexander Oliveira (2001), Exploring the Experiences of Collaborative Planning Initiatives, International Journal of Physical Distribution & Logistics Management, Vol. 31, No. 4, pp Clark, A.J. and Scarf, H. (1960), Optimal policies for a multi-echelon inventory problem, Management Science, Vol. 6, No. 4, pp / 21

20 DeKok, T., and Fransoo, J. (2002), "Planning Supply Chain Operations: Definition and Comparison of Planning Concepts", in Handbook of Operations Research: Supply Chain Management, Chapter 14, Manuscript September 9, 2002 Euwe, M.J. and Wortmann, H. (1997), Planning systems in the next century (I), Computers in Industry, Vol. 34, Iss. XX, pp Fisher, M.L. (1997), ``What is the right supply chain for your product?, Harvard Business Review, Vol. 75 No. 2, March/April. Fisher, M.L., A. Raman, and A.S. McClelland, Rocket Science Retailing Is Almost Here Are You Ready?, Harvard Business Review, Volume 78, Issue 4, pp Forrester, J. (1961), Industrial dynamics, Cambridge MA, MIT Press Fransoo, J., Wouters, M.J.F., and dekok, T. (2001), Multi-echelon multi-company inventory planning with limited information exchange, Journal of the Operational Research Society, Vol 52, No. XX, pp Helms, Marilyn M., Lawrence P. Ettkin, and Sharon Chapman (2000), Supply Chain Forecasting Collaborative Forecasting Supports Supply Chain Management, Business Process Management Journal, Vol. 6, No. 5, pp Ko, Eunju, Doris Kincade, and James R. Brown (2000), Impact of business type upon the adoption of quick response technologies The apparel industry experience, International Journal of Operations & Production Management, Vol. 20, No. 9, pp Lee, H.L., V. Padmanabhan, and S. Whang, Information Distortion in a Supply Chain, Management Science, Volume 43, Issue 4, pp Lewis, C., L. Roth, and A. White, Collaborative Planning, Forecasting and Replenishment n-tier CPFR, Global Commerce Initiative, Interim Report Update #4, URL: ( ), 55 p. Mentzer, John T., Mark A. Moon, John L. Kent, and Carlo D. Smith (1997), The need for a forecasting champion, Journal of Business Forecasting Methods & Systems, Vol. 16, No. 3, pp Salmi, L., J. Småros, and J. Holmström, A solution for monitoring new product introductions with sell-through data from channel partners, Working paper, Helsinki University of Technology, 9 p. 19 / 21

21 Seifert, Dirk (2002), Collaborative Planning Forecasting and Replenishment - How to create a Supply Chain Advantage, Galileo Press GmbH, Kevelaer, Preprint Edition. Simchi-Levi, D., and E. Simchi-Levi, The Effect of e-business on Supply Chain Strategy, MIT Forum for Supply Chain Innovation, URL: ( ), 24 p. Småros, J. (2003a) Collaborative Forecasting: A Selection of Practical Approaches, forthcoming in the International Journal of Logistics: Research and Applications, 17 p. Småros, J. and Hellström, M. (2003), Using the assortment forecasting method to enable sales force involvement in forecasting: A case study, forthcoming in the International Journal of Physical Distribution & Logistics Management Småros, J., Lehtonen, J-M., Appelqvist, P. and Holmström, J. (2003), "The impact of increasing demand visibility on production and inventory control efficiency", International Journal of Physical Distribution & Logistics Management Vol. 33, No. 4 UCCNet, 2003, on September 23 rd, 2003 VanDonselaar, Karel, Rock Kopczak, Laura, Wouters, Mark, The use of advance demand information in a project-based supply chain. European Journal of Operational Research 130, Vokurka, Robert J. and Robert A. Davis (1996), Just-in-Time: The Evolution of a Philosophy, Production and Inventory Management Journal, Vol. 37, No. 2, pp / 21