The current issue and full text archive of this journal is available at www.emeraldinsight.com/1741-0401.

htm

REFLECTIVE PRACTICE

Achieving customer service excellence using Lean Pull Replenishment

Sameer Kumar
Opus College of Business, University of St. Thomas, St. Paul, Minnesota, USA, and

Customer service excellence

85

David Choe and Shiv Venkataramani

TE Connectivity, Shakopee, Minnesota, USA
Abstract
Purpose The purpose of this study is to highlight a key strategic initiative within the former ADC Company (now part of TE Connectivity) called Lean Pull Replenishment, designed and implemented to achieve Six Sigma customer service excellence. This case study would also help facilitate problem-based learning pedagogy. Design/methodology/approach The study showcases implementation of the Lean Pull Replenishment approach using the define, measure, analyze, improve and control (DMAIC) framework. Key input variables were analyzed that contributed to historically inconsistent and unsatisfactory customer delivery performance. Analysis resulted in improving the allocative efficiency of critical input variables through pilot programs on strategic value streams by deploying dozens of kaizen events, and sustaining the gains through leveraging best practices and effective change management principles. Findings The study presents a strong case for the team work and the cultural transformation that occurred during the course of implementing this initiative across ADC supply chain. The paper also summarizes the improvement in customer service metrics and financials of the company. Originality/value Through this study, it has been established that with consistency of purpose, using the right tools for solving problems and through teaching Lean principles, remarkable results can be achieved, which can be sustained for the long-term and become a self-sustaining business philosophy. Keywords Six Sigma, DMAIC, Lean Pull Replenishment, Process capability, Process improvement, Quality assurance, Telecom, Customer service, Customer service management Paper type Case study

Introduction Business organizations continue to be involved in Lean and Six Sigma implementation to improve their operations. Real life industrial case studies providing detailed implementations of Lean Six Sigma using define, measure, analyze, improve, and control (DMAIC) methodology would markedly facilitate problem-based learning pedagogy (e.g. see Johnson et al., 2006; Rasis et al., 2002a, b). There are, however, few comprehensive real-world case studies available on this topic (Allen, 2010). Six Sigma is a multi-method approach and has been described as a data-driven method for process improvement (De Mast, 2007; Linderman et al., 2003). The study described in this paper shows how a team of highly engaged employees executed a large business transformation christened Lean Sigma Pull Replenishment in a leading US global telecommunication products company from the bottom-up. By reaching across functional, geographic, and organizational boundaries; focussing on a

International Journal of Productivity and Performance Management Vol. 62 No. 1, 2013 pp. 85-109 r Emerald Group Publishing Limited 1741-0401 DOI 10.1108/17410401311285318

IJPPM 62,1

86

common vision; and working together, they overcame years of stagnation and cynicism and demonstrated a real cultural change in less than one year. This is also a story of how the same employees integrated Lean and Six Sigma methodology very quickly and applied these to every step of critical supply chain, order management, factory floor execution, and logistics operations to drastically improve inventory turns, cost performance, and more importantly reduce lead time to meet the customers demand, while reducing overall inventory at the same time. In particular, this case study provides an example of major implementation efforts to streamline business operations and supply chain of a company with a focus on improving quality of delivery performance to customers. Literature review The purpose of this literature review is to provide a brief overview of Lean thinking, Six Sigma, and Lean Pull Replenishment methodology to identify the major challenges that will face any business organization in implementing these practices, and examine their use in similar organizations. Lean thinking Lean thinking originated within the Japanese automobile industry following Second World War and is principally based on the Toyota Production System (TPS), which was developed by a production executive named Taiichi Ohno and was used to improve the quality and productivity within Toyota Motor Company (Ohno, 1988). Lean thinking later increased in popularity in the 1990s, after the publication of the bestselling book, The Machine that Changed the World: The Story of Lean Production, which chronicled how organizations could transform their operations by adopting the Lean approach developed at Toyota (Womack et al., 1991). Lean has since been widely accepted and adopted across every industry ranging from automobiles to electronics and in the recent years, is being increasingly applied to a wide range of service organizations, including health insurance companies, hospitals, clinics, retail stores, etc. Lean is an integrated system of principles, practices, tools, and techniques that are focussed on reducing waste, synchronizing work flows, and managing production flows (de Koning et al., 2006). The reduction of waste is the fundamental philosophy of the Lean approach. In Lean, waste is also referred to as non-value-added activities. Value-added activities are those for which the customer values and is willing to pay. Other activities are considered non-value-added activities and should be eliminated. The elimination of these non-value-added activities reduces cycle time and costs, which results in more competitive, agile, and customer-responsive organizations (Alukal, 2003). The level of competition to capture customers in both domestic and international markets demands that organizations be quick, agile, and flexible to compete effectively (LaLonde, 1997; Fliedner and Vokurka, 1997; Shenchuk and Colin, 2000; Wang et al., 2005). Six Sigma Six Sigma was developed in the early 1980s at the Motorola Corporation and was popularized in the late 1990s by former General Electric CEO, Jack Welch (Furterer and Elshennawy, 2005; de Koning and de Mast, 2006). Six Sigmas foundation was in the statistical analysis of data, and this is reflected in its name, which refers to a statistical measure of process performance (Maguad, 2006). Besides Motorola and

General Electric, other major corporations have embraced Six Sigma including AlliedSignal, DuPont, Honeywell, Lockheed Martin, Polaroid, 3M, Samsung, and Texas Instruments (Hahn et al., 1999; Yang et al., 2007). The reported advantages to implementing Six Sigma include increased market share and higher profit margins (Harry, 1998). While Six Sigma originated within manufacturing in the electronics industry, it has since been adopted across many other industries and has spread into the service sector, including a few government organizations. Six Sigma refers to the philosophy, tools, and methods used to seek, find, and eliminate the causes of defects or mistakes in business processes by focussing on the outputs that are important to the customers (Anthony and Banuelas, 2004). Six Sigma represents a highly disciplined and statistically based approach to quality (Hahn et al., 1999). Six Sigma methodically analyzes underlying data and identifies the root causes of problems as opposed to using subjective opinions. Since every step in a process represents an opportunity for a defect to occur, Six Sigma seeks to reduce the variation in these steps, which results in the occurrence of fewer defects and the production of higher quality goods and services. By controlling this variation, Six Sigma prevents defects from occurring rather than simply detecting and correcting them. Pull Replenishment methodology The Lean Pull Replenishment (or kanban) system is a compromise between the ideal of one-piece flow and the traditional large-batch push business (Liker, 2004). A kanban system is a means to achieve just-in-time production. It works on the basis that each process on a production line pulls just the number and type of components that process requires, at just the right time. The mechanism used is a kanban card. This is usually a physical card but other devices can be used. In the 1940s, Toyota began studying US supermarkets to understand and analyze how they managed to anticipate, plan, stock, and replenish goods based on customer demand. Supermarkets, fresh meat, and produce in particular, are great case studies for Lean management methods. A profitable supermarket only stocks what it can sell and supermarket customers only buy what they need when they need it because they are confident in an uninterrupted supply of goods and produce (Liker, 2004). In the 1950s, Toyota began to create a type of supermarket for parts and materials within its factories. Toyota purchased a strategic inventory of parts and supplies based on its calculations of Takt rate or customer demand. Employees at each workstation withdrew the parts and supplies they required as needed on-demand to maintain continuity and efficiency throughout the vehicle assembly process. In a Pull/Replenishment system a very small strategic level of inventory is built and maintained in bins at selected points in the production process. When a downstream customer takes away specific items they are replenished. If a customer does not use an item, it sits in a bin but is not replenished. When the strategic inventory is depleted, the downstream customer uses a simple card or kanban as a signal to order the upstream supplier to refill the bin with a specific number of parts or send back a card with detailed information regarding the part and its location (Martin, 2007). As this system of consumption-driven replenishment proves its applicability across tiers of distribution and supplier networks, it is becoming an increasingly compelling alternative to more traditional material requirements planning (MRP) or push style systems, which rely on forecasts to determine what and how much to produce (Lakham, 2008). Although forecasts have proven useful in predicting overall demand, they can be poor indicators of exactly which products will be needed and

Customer service excellence

87

IJPPM 62,1

88

when. Multi-tiered supply chains push manufacturers further and further from their customers, and forecasts further and further from reality. As a result, information is diluted by each layer in the distribution system, and excessive inventory and costly last-minute changes orders ripple through the supply chain. In addition to these issues, manufacturers often complain that they incur large inventory-carrying costs with an MRP system, yet can still run out of key parts. These stock-outs stop production, delay customer shipments, increase premium freight charges, and disrupt plant operations by forcing unnecessary and expensive changeovers. In the current MRP world there is often no clear record of how many times stock-outs occur or which parts repeatedly stock-out. This could be due to configuration of ERP systems where the records of the previous day could be over-written the next day and thereby the system loses track of historical stock-outs, needed for corrective actions. Another drawback of this system is re-configuration could be very expensive and time consuming. Stock-outs also can lead to an overreaction of parts buying, followed by substantial excess inventory, which is often carried for months afterwards. Keeping expensive inventory is a waste of resources, including working capital, storage space, and the manpower needed for additional handling. Unlike MRP forecast-driven replenishment, a kanban system re-orders based on actual consumption at the point of use. However, it is important to note that the telecommunication products company uses collaborative forecasting processes such as sales and operations planning that drives internal company collaboration and collaborative planning forecasting and replenishment that involves customers and suppliers. The simplest version of this is the two-bin method. In this case, an operator has two bins of material, one being consumed and the other, full. When the first bin is empty, the operator continues working using the second bin. The empty bin is sent to the producing station, an obvious signal to replenish. The amount of material per bin is set so a full one returns before the operator runs out. Another example of the kanban process uses the manual kanban card, which travels with its inventory and contains information such as the description of the item or part number, and its location. Each card has a number and is used to trigger an order for replenishment when an item is consumed. However, manual kanbans benefits are severely limited when an external supplier enters the supply chain. In an electronic kanban system, the card information is translated into a barcode that is scanned and electronically communicated at each stage of the replenishment cycle (consumption, shipping, receiving, etc.). In this way, electronic kanban dramatically increases the effectiveness of kanban throughout the supply chain. While MRP systems push material throughout the supply chain, pull-based manufacturing strives to synchronize production with consumption in real time, which increases on-time delivery performance, reduces stock-outs, and cuts down on costly last-minute changes to orders. As orders arrive, material is pulled from the end of the final assembly line, which instantly sends an order to final assembly to produce more (Liker, 2004). Implementation of Pull Replenishment methodology at a leading US telecommunication products company The company provides the connections for wireline, wireless, cable, broadcast, and enterprise networks around the world. The companys innovative network infrastructure equipment and professional services enable high-speed internet, data, video, and voice services to residential, business, and mobile subscribers. With sales more than 130 countries, the companys innovative, high-performance fiber connectivity and wireless coverage and capacity solutions support a broad range of

network applications that help its customers provide new and advanced services to meet their own customers unrelenting demand for bandwidth. In 2008, this company was a $1.5 B company, yet it sold more stock keeping units (SKUs) than target corporation, a $60 B giant. The SKU proliferation created a literal gridlock in its supply chain. Since the company was operating under a build-to-forecast model, no matter how brilliant the SAP software system expert master schedulers were, they invariably incorrectly guessed product mix. It may be noted the company has been using SAP enterprise software to manage business operations and customer relations. This software is developed by SAP AG, a market leader in enterprise resource planning. When the real orders came in, operations and sales were at odds just to ship even basic regularly ordered products. The operations literally looked like the I-405 in Los Angeles, 12 lanes of gridlock. There was so much congestion that the company started inventing sales alerts, phantom orders that might turn into real orders. If it could just start earlier, it might ship on time. It did not work (Choe, 2011). After four years of false starts and over $10 M spent on consultants, software, and FTEs, every transformational initiative the company tried failed. Cynicism grew rampant about the flavor of the month initiative. Employees felt that management was totally disconnected from reality. However, the consultants were not wrong. The underlying operating system was falling apart. Without massive change the system would crash, meaning, the company would be unable to profitably serve its customers and compete in the twenty-first century. Something had to change (Choe, 2011). A new operations management team with deep and broad expertise in Lean, Six Sigma, and Lean Pull Replenishment methodology, was assembled from other Fortune 500 companies. Its purpose was to create efficient operation, eliminating waste and reducing variation, superior customer service with minimal inventory. This team brought to bear Lean and Six Sigma methodologies, including Lean Pull Replenishment and process capability thinking. Process capability thinking was important since the companys sales force was setting delivery expectations with the customers without understanding what the capability of operational and associated transactional processes. Through the application of the time-tested DMAIC methodology, was able to energize the companys team to transform the fundamental approach to manufacturing: Push- based MRP planning to a Pull- based replenishment model. In most companies, the traditional manner of implementing Lean and Six Sigma would be through extensive Lean and Six Sigma tool training, black belts, and kaizen senseis converging with metrics like: the number of green belts and practitioners trained, number of kaizens facilitated and so on. Also, these kaizens and improvement activities are focussed on the shop floor of the manufacturing or assembly plant for a couple of reasons: (1) (2) Lean is primarily associated with manufacturing due to its association with the TPS; and one can touch and feel the products flowing through the production floor, as well as, easily observe the physical flow of material, machine workers, etc.

Customer service excellence

89

This approach results in improvement in production efficiencies, labor and material productivity, etc, but in customer service there is no visible improvement. The fundamental reason is that teams and initiatives only focus on what they touch and feel

IJPPM 62,1

and end up optimizing parts and not the whole process end-to-end. The companys operations management team recognized this due to their experiences from other companies and organizations. The team decided to investigate the entire customer order-to-delivery process for a few part numbers, instead of taking on the entire product portfolio of 60,000 finished goods part numbers. Step 1: define the problem The team decided to map out what it took to book and ship an order. In the history of the company, this had never been done. After four days and hundreds of post-it notes, the research team was able to show all the hand-offs and steps on a 12 8 ft mural. It was so large that it could not be taped on the wall. The centerpiece of this process was what was dubbed the circle of death. The circle of death was the product of the companys build-to-forecast mess. Since customers could not get products on time, they started giving the request dates of today and even January 1, 1900 in the hopes that their order would get bumped forward in the queue. With over 30,000 orders per month, this phenomenon turned into white noise. The only orders being worked on was where the salesperson and customers were yelling the loudest. Back and forth the e-mails and phone calls would go. The process was essentially a negotiation. The customer would ask for delivery today on products where we had no forecast visibility. Operations would immediately reply with a 90-day lead time. Customer service would hit the expedite button. Operations would then reply with a 50-day lead time. The sales person would yell and scream. Operations would then start looking at capacity, scheduling, and materials and then commit to a 45-day lead time. This process would go back and forth for a week until operations gave their best and final: 40-days. The exhausting circle-of-death aggravated customers, infuriated sales people, and wore down everyone from customer service through supply chain. But, a ship-to-promise metric was created that stated order completions were 97 percent on time to that final commit date! This was also validated by voice of the customer (VOC) data that the company had obtained and following are some of the quotes from our customers and salespeople:
The first confirm date is ridiculous. Its taking me so long just to get a delivery date. Why am I spending so much time chasing down orders? Our lead-times arent competitive. Its difficult to do business with the company. Ship-promise does not reflect customer satisfaction. Our inconsistency is costing our customers money. People dont trust our ability to ship product. We are all over the place in lead-times.

90

At this point in time it was concluded that the MRP based-push system was a failure in this company and the company would have to implement a Lean Pull Replenishment system instead of a build-to-forecast model, but the task seemed impossible. Step 2: measure Using the MRP-based push system, the company was trying to cram too much demand into the supply chain without truly understanding lead time requirements of customers and cycle times of not only the manufacturing processes, but also the transactional processes which created chaos in the system. In order to get ahead, sales, product management, and other functional groups designed and created tools like expedites, sales alerts, hot orders, and the like. These functional groups had no confidence in the companys processes and were setting false expectations with customers. Sales and product management were signing contracts with customers with unrealistic lead times. In short, no one in the company understood what the true process capability was of the order creation to delivery process of different product lines and segments. When the newly created Lean Pull Replenishment team, led by the authors, tried to measure process capability of the end-to-end process of one of the product lines, it was impossible to do so, because the true process capability was masked by the above mentioned transactional noises such as expedites and sales alerts, etc. Why the company cannot use inventory as a buffer against forecast errors? When the fallacy of this MRP-based approach to other stakeholders like sales and product management was discussed, they said that the company could achieve superior lead times and customer service with finished goods inventory as a buffer. This intuitively made sense if a company sells a few SKUs, but it was known that it would not work for 60,000 active SKUs. To prove our hypothesis, we did a regression analysis of lead time vs finished goods inventory of 35 part numbers (all high runners) sold to this companys number 1 customer for fiscal year 2008 (see Figure 1).

Customer service excellence

91

90 80 70 Lead time (days) 60 50 40 30 20 10 0 0 100

S R2 R 2 (adjusted)

9.78084 0.0% 0.0%

200 Safety stock (quantity)

300

400

Figure 1. Correlation plot of lead time (in days) to safety stock inventory for 35 SKUs in the ODF portfolio sold to one customer

IJPPM 62,1

92

Figure 1 shows that for the 35 part numbers, there is no correlation (R2 value 0 percent) between quantity in stock and lead times. In other words, if we had a quantity of 100 at any given time, it could take us anywhere from two days (if we happen to have the SKU in stock) to 90 days (if even one of the component is in short supply and not available for use). Imagine the scenario for 60,000 finished SKUs, each consisting of hundreds of components and sub-assemblies in its bill of material (BOM structure). We have performed similar analysis on numerous parts for numerous customers and the resulting analysis does not look very different than the above chart. This type of analysis conclusively proves that in this line of business (telecom), it is impossible to forecast every SKU accurately and maintain an inventory buffer. A group forecasting approach was tried and led to higher inventory levels of components resulting in greater excess and obsolete inventory. This approach was abandoned in favor of a Pull system (Replenishment, sequential, and hybrid Pull). In the telecom business, there has been a rapid evolution of technologies due to the advent of smart phones, smart TVs, data centers, social networking sites, etc., that consume tremendous amount of bandwidth. The technologies become obsolete very quickly and therefore finished goods that we keep in inventory, could become obsolete within months leading to excessive dead and excess inventory, and a drain on the companys financials and infrastructure resources. This problem alone behooves us to look at models such as Make-to-Order and Lean Pull Replenishment, where no finished goods are manufactured or assembled unless a purchase order (PO) is received from the customer. In some cases, Pull systems maintain a small amount of finished goods inventory that gets replenished only on consumption and not to a forecast. The second reason why it makes stocking finished goods to forecast impossible is SKU proliferation, which commonly happens in many organizations and the telecommunications company was no exception. Figure 2 shows the Pareto chart of the companys revenue in North America by SKU. Figure 2 speaks for itself in how skewed the Pareto looks in terms of the 80/20 rule. In fact 7 percent of SKUs make up 80 percent of the revenue. The tail as we call it is amazingly skewed. A total of 40 percent of the revenue in the tail generates only 0.5 percent of the overall revenue. So, as a team, it was decided not to put efforts into

60 50 40 30 20

Ship order

Figure 2. Telecommunications company revenue Pareto of all SKUs in Americas

10 0 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 Dollar revenue (FY 2009)

implementing Lean Pull Replenishment on the tail SKUs. Instead, the team focussed only on the top 7 percent in waves 1 and 2 before considering the middle section of the Pareto chart. The skewness of the Pareto was not the only problem. The issue was also that there was little differentiation in pricing and lead-time between high-runners and lowrunners. There was no incentive for customers to make choices so they chose more customization. In other words, at this telecommunications company, if one ordered a custom cabinet, you got the same lead time as an off the shelf cabinet. Why would the customer not opt for the custom cabinet most of the time? All of the above issues resulted in varying lead times for the customers for the same product. In other words, customers could not rely on this telecommunications company to give them the same product at the same time every time. The team analyzed historical delivery times for several products and most of them had the profile as shown in Figure 3 (actual delivery times for a typical high-turning SKU). Figure 3 is an individual chart (X-Y plot) that shows all lead times for the sales order for one product SKU. In some cases, the product was delivered in one day and in some cases it took 200-250 days to deliver the same product. This variation in lead time was also independent of delivered quantity. Due to the inconsistency, customers could not rely on the telecommunications company to deliver products on schedule to the project site, be it a carrier installation or an enterprise data center. In this business, if a component or a product does not get delivered to the installation site and the installation contractors are already on-site, the customer (carriers/installation contractors) still have to pay the contractors for their time, and this creates a bad experience for the customer. Therefore, some customers hedge their bets by having the telecommunications company sales people expedite orders, placing sales alerts (phantom orders) in the system creating false demands on the system, driving material and taking away manufacturing capacity meant for actual orders. All of the above resulted in an overall customer service level of 65 percent, which means only 65 percent of the line items were shipped on time. However, the company reported a metric called ship-to-promise and showed an on time shipment of 98 percent (Figure 4). Ship-to-promise was an internal company metric that basically said, we will deliver the goods when we promise them, which of course was hardly the first promised date. In fact, most of the time, it was the delivery date that we promised the customer after several expedites, phone calls, e-mails, and the like with several people involved sales, product management, master schedulers, planners, buyers, and so on. However,

Customer service excellence

93

300 250 Lead time (days) 200 150 100 50 0 Sales orders for LSX-9000 during FY 2009

Figure 3. Actual delivery lead times for product LSX-9000 during FY 2009

IJPPM 62,1
Ship-to-promise %

100 95 90 85 80 75 70 65 60 55 50 Month-FY 2009

94
Figure 4. Ship-to-promise percentage for LSX-9000 product line during FY 2009

as the voice of the customer was analyzed, this was not the case at all. Customers were measuring the company by how reliable it is with the lead times written in the contracts (ship-to-target) or when they requested the product (ship-to-request). These metrics were more meaningful than ship-to-promise. The value stream and process capability approach In the measure phase, true process capability was needed to be measured using a well-known tool in the Six Sigma tool kit process capability analysis. From the VOC data and interviewing customers as to what would be a reasonable, predictable lead time they could accept, the answer invariably came back as two to three calendar weeks if the company can predictably deliver the products to their distribution centers or installation sites (barring emergencies, of course). In the customers world, when parts are ordered to service new installation sites or turning up or upgrading customers sites, customers typically use their own technicians or outside contractors. If the products are not delivered to the site at the agreed upon dates, the contractors cannot finish the installation and still have to be paid expensive rates, not to mention disappointing the end consumer or business person. Armed with this information, the telecommunications company decided to measure process capability of its product streams (value stream performance) using process capability tools (Cpk and Ppk). For example, for one of the strategic, high running product families (DS-3), Ppk was measured as 0.04 for the past 12 months performance to 14 days as the upper specification limit (please refer to Figure 5). In this case, Ppk was measured as the team wanted to understand long-term process performance vs short-term process performance, usually measured as Cpk. This shows that the entire value stream (order entry to delivery) is not even capable to perform at 14 days, yet the sales people and product managers were giving customers arbitrary or SAP generated artificial lead times that we were only met by taking material meant for different orders, expensive expediting, air shipment from suppliers, or simply, the times were not met at all. It took an army of people to scramble and hope to meet the customers dates. It was not surprising to find out that other product streams performed similarly or even worse, especially if they had a few longlead time components. When the research team mapped the order management process, it was obvious that the process was very complicated it would take

USL
Process data LSL * * Target 13 USL Sample mean 15.9595 666 Sample n 1.79622 Shape Scale 17.9587 Observed performance PPM < LSL * PPM > USL 578,078.08 PPM total 578,078.08 Overall capability * Pp * PPL PPU 0.04 Ppk 0.04 Exp. overall performance * PPM < LSL PPM > USL 571,396.50 PPM total 571,396.50

Customer service excellence

95
Figure 5. Process capability analysis of total process lead time (sales order create to finish process) for DS-3 product line before Pull Replenishment implementation

16

24

32

40

48

56

anywhere between two and 21 days, just to process an order. The high-level process map summary is shown in Figure 6. Moreover, it was very difficult to measure true manufacturing process capability because of the existence of the expedite tool within SAP. In other words, if a production order is released to floor and the planner has started to build to that order, an expedite generated by a product manager, master scheduler, or any other person up and down the chain, could steal materials meant for the first order, thereby stopping the order mid-way and re-starting only when material becomes available again. Since 40 percent of the line items were expedited or escalated, it was very difficult to understand true manufacturing process capability. Another culprit for inconsistent delivery performance is the tool called sales alerts. This was a glorified forecast, where the sales person submitted a sales alert in the system without a PO from the customer, believing that the customer would send the PO soon. Our SAP system immediately converted the sales alert into a production

Quote 7-21 days

Receive order

First date @ PTF confirm 60 days

Expedite

Wait for response

Confirm 45 days

EX1

Ops/SC scramble to improve date

Wait for response

Confirm 40 days

EX2

Confirm 36 days

Wait for T-5 in MRP to release order

Create production order

Build order

Ship order

Figure 6. Current state process map of the order management process before Lean Pull Replenishment was implemented

IJPPM 62,1

order (phantom order) and built inventory of finished goods, hoping that the customer would send in the PO shortly. In 40 percent of the instances, customers would cancel the sales alert or postpone the PO, which results in finished goods sitting in the warehouse, leading to excess and obsolete inventory. Step 3: analyze After analyzing the delivery lead times against lead times specified in customer contracts, and brainstorming all the reasons as to why the company ships 40 percent of the line items late, it was obvious that the main problem was not in the manufacturing operations. As shown in Figure 7, for DS1 product line, the median end-to-end lead time was 27 days. As shown, it takes on an average, six days of transactional activities (mostly non-value added) from the time a sales order is created in the system to when the production order is created on the factory floor. This represents approximately 25 percent of the total lead time. Also, as can be seen, it takes 13 days from the time a production order is created to when it is actually released to factory floor. The reason for this long lead time is that the manufacturing line is waiting for a raw material or a sub-component from a supplier or an internal plant. Once the production order is released to the factory floor, it takes only three days on an average for the product to be manufactured in the factory. Similar process lead time profiles exist for most of the products made in the company. In most instances, the customer demands a lead time of one to two weeks. This shows that the process is not capable of meeting the customers lead time specifications. Based on the above data, the team decided to focus on the following three critical input factors as shown in Figure 8: (1) (2) (3) raw material and sub-assembly availability; transactional processes like expedites; and sales alerts.

96

Figure 7. Breakdown of total process lead time (sales order created to ship from distribution center) for DS-1 product line

1 day*

6 days*

13 days median

3 days median

LT end-to-end 27 days median

Pr od u Sh cti ip on pe o d rde to r D C Sh ip pe d fro m D C

or de r

C us to So me cr r ea te d

Pr od uc tio n

Pr od uc tio n

or de r

1 day

3 days

One-piece flow Set-up reduction

Portfolio selection Cell layout

Capacity

Customer service excellence

BOM simplification Component availability Sales alerts Expedites

97

Figure 8. Funnel showing the various input variables and the critical input variables (Xs)

To analyze true process capability and true lead times, we decided to do an experiment with 36 part numbers out of the value stream. The BOM of each of the 36 part numbers was exploded and, based on historical demands and future forecast, an SAPs kanban calculator was used to fill up the bins with raw material components by working with both external suppliers and internal plants. This took two months, and in the meantime, sales and product management shut off expedites and sales alerts in the system for 45 days to help the company understand true process capability. Sales and product management responded by saying that they do not wish to expedite since they spend 40 percent of their time expediting orders internally and not closing deals with customers. The expedite button was grayed out in SAP for the 36 part numbers and called them optimized part numbers. This was the indicator that the particular part number was on a Lean Pull Replenishment mode. Step 4: improve: implementing Pull Replenishment The key to making Pull Replenishment work is by establishing supermarkets. A manufacturing supermarket (or market location) is, for a factory process, what a retail supermarket is for the customer. The customers draw products from the shelves as needed and this can be detected by the supplier who then initiates a replenishment of that item. It was the observation that this way of working could be transferred from retail to manufacturing which is one of the cornerstones of the TPS (Liker, 2004). In a manufacturing supermarket, processes consume what they need when they need it. Since the system is self-service, the materials management is reduced. The shelves are refilled as parts are withdrawn on the assumption that what has been consumed will be consumed again, which makes it easy to see how much has been used and to avoid overstocking. The most important feature of a supermarket system is that stocking is triggered by actual demand, as shown in Figure 9, which is a standard two-bin system utilized in Pull manufacturing. Usually manufacturing supermarkets are in clearly marked bins right on the shop floor with easy access to the manufacturing cell. Another key ingredient for a Lean Pull Replenishment system to work efficiently is the role of water spider or Mizusumahi. Water spider is a person who manages all

IJPPM 62,1

Supplier

Telecommunications company

Warehouse

Supermarket

98

External

Shakopee

P.O. sent to supplier upon at reorder point

Reorder point

Water spider

Figure 9. Schematic showing the mechanics of Pull Replenishment where the entire supply chain is aligned to customer demand

Manufacturing cell

the logistical work of bringing components, raw materials, etc., in small quantities to workstations to minimize work-in-process inventories. This allows machines to be placed closer together, and spares the operator from having to interrupt his/her cycle time, thus minimizing transportation muda (waste). Water spiders usually are experienced workers. They know where needed parts or raw materials are stored, and serve several workstations. The supermarkets on the shop floor are generally a set of bins, with each bin containing a finite set of parts. For example, for a certain raw material or sub-assembly, we could have two bins of that part, each containing 1,000 pieces. When one bin is emptied due to consumption and the re-order point is sent to 1,000, the system or the water spider will signal the need for another bin, either from an external supplier or from an internal plant or a different process. During the lead time it takes to fill the empty bin, the process is consuming from the second bin. By the time the second bin is empty, the first bin is stocked with parts. Generally, the re-order point is established based on manufacturing lead time of the part, transportation lead times, etc. SAP and other ERP systems have re-order point calculation algorithms based on demand patterns of the particular part. Supermarkets work very efficiently when there are robust signaling systems between the supermarkets and their suppliers. Most commonly used signals, or kanbans, are faxes, ERP signaling, visual index cards (kanban cards), web cams, PO, etc. When the company combines the availability of raw materials and elimination of expedites with classic Lean tools on the production floor (like 5S, Takt time analysis, standard work, cell design, operator balancing and line balancing, visual management, cross-training of operators, etc., all implemented using a bottom-up kaizen approach) there was also immediate improvement in the delivery performance. Delivery

performance (ship-to-target) increased from an average of 76-100 percent for these 36 part numbers over a period of 45 days. In other words, the telecommunications company delivered the 36 SKUs to its customers (hundreds of line items) at or before 14 days. In summary, here are the steps in the Pull Replenishment implantation process: (1) (2) (3) (4) (5) (6) choose a select number of SKUs representing high volume (80-20 rule); maintain 100 percent availability of components through a kanban-bin system; actual orders only trigger a production order; production orders deplete kanban bins; when component inventory falls below a reorder point, it automatically triggers replenishment in the supply-chain; and suppliers are aligned to a 14-day replenishment lead time.

Customer service excellence

99

The results from this experiment gave the company confidence and it extended the experiment ( July experiment) to 250 part numbers over three value streams, as shown in Figure 10. When an order for any of the 250 SKUs came in from a customer, the order was immediately acknowledged with a delivery estimate of 14 days (predictability). Through the elimination of expedites and sales alerts and having a kanbanreplenishment approach instead of a guesstimating-forecast approach, and optimizing the manufacturing floor through kaizens and continuous improvement, the 100 percent ship-to-target results were extended to the 250 SKUs that were in scope over hundreds of line items and over a period of a month. Figure 11 shows the comparison of actual delivery time (in days) for the 250 part numbers that were placed on Pull during the month of July to that of the same
105%
Average of ship-target Average of target %

100%

95%

90%

85%

Ship to target %

80%

75%

70% 5-Mar-10 9-Mar-10 11-Mar-10 15-Mar-10 17-Mar-10 19-Mar-10 23-Mar-10 25-Mar-10 29-Mar-10 31-Mar-10 2-Apr-10 6-Apr-10 8-Apr-10 12-Apr-10 14-Apr-10 16-Apr-10 20-Apr-10 22-Apr-10 25-Apr-10 27-Apr-10 29-Apr-10 3-May-10 5-May-10 7-May-10 11-May-10 13-May-10 17-May-10 19-May-10 21-May-10 25-May-10 27-May-10 30-May-10 1-Jun-10 3-Jun-10 7-Jun-10 9-Jun-10 11-Jun-10 14-Jun-10 16-Jun-10 18-Jun-10 22-Jun-10 24-Jun-10 28-Jun-10 30-Jun-10 ul-10 2-Jul-10 ul-10 6-Jul-10 ul-10 8-Jul-10 ul-10 12-Jul-10 ul-10 14-Jul-10 ul-10 16-Jul-10

Figure 10. Showing ship-target percentage improvement after Pull Replenishment implementation on 250 finished goods SKUs

LT for 250 SKUs (after pull)

IJPPM 62,1

100

250 part numbers prior to July (three months data). As can be seen, the Pull Replenishment methodology ( July experiment) is very consistent with a mean of five days and 95 percent confidence interval (CI) of 11 days. This means that 95 percent of the time, the company will deliver the products within 11 days. Using the MRP methodology (pre-July), there was too much variation (anywhere between one day and 150 days) with a mean of 18 days and 95 percent CI of 50 days. These results were truly astounding and customers started to take notice and asked the telecommunications company to accelerate the Lean Pull Replenishment methodology to more SKUs and by October, 750 more SKUs (20 percent of Americas revenue and line items) were added to the Pull systems. The results were even more impressive as shown in Figure 12. As shown in Figure 12, ship-to-target performance was still 495 percent compared to 65-70 percent prior to implementing Lean Pull Replenishment. This clearly proved that the methodology is highly scalable and extendable to at least 75-80 percent of the revenue or line items shipped. If the telecommunications company was able achieve this kind of results, the company could end up delighting the customer and as a company, and could win on superior and predictable delivery times vs win on price, a strategy very difficult to sustain as a large public corporation. Manufacturing process improvements Once the transactional process improvements were made and sustained, the team turned its attention to improving the manufacturing and assembly processes using traditional Lean and Six Sigma tools such as 5S, operator balance charts, cell design, material replenishment, kaizens, standard work and QCPC/turnbacks, visual management, etc. In one of the manufacturing lines, for example, the company used operator and line balancing using Takt time analysis (Takt time time available for operations/ customer demand). Before any improvements were done, the line was not balanced and had too many operators as shown in Figure 13. Figure 13 shows the cycle time of each operator to perform their respective tasks with respect to a Takt time (red line) of approximately 165 seconds (Takt time is defined as the time available for operations divided by customer demand). The assembly line in this case was a straight line and it was difficult for the operators to carry out multiple tasks in order for the line to be balanced. It is obvious that the line
350 300 Calendar days 250 200 150 100 50 0 Pull lead times Non-pull lead times

Figure 11. Box-plot showing lead time variation reduction after Lean Pull Replenishment implementation on 250 SKUs

Ship-target % 100% 95% 90% Operator cycle time (OCT), in secs 85% 80% 75% 70% 65% 60% Before 55% 50%
-A 2- ugS 10 4- epS 10 6- epSe 10 8- p10 Sep 10 - 12 Sep 10 - 14 Sep 10 -S -1 16 ep 0 - 18 Sep 10 - 20 Sep 10 - 22 Sep 10 -S 24 ep 10 - 26 Se 10 -S p-1 28 ep 0 - 30 Sep 10 -S -1 e 0 2- p-1 O 0 4- ctO 1 6- ct-1 0 O 0 8- ctO 10 10 ct - -1 12 Oc 0 - t14 Oct 10 -O -1 16 ct 0 -O -1 18 ct 0 -O -10 20 ct - -1 22 Oct 0 - -1 24 Oct 0 -O -10 26 ct -O -10 ct -1 0

Customer service excellence

101

After

Figure 12. Chart showing shipto-target percentage performance improvement after 1,000 SKUs were put on Pull Replenishment methodology

31

Takt time = 165 sec/tray

Takt time = 165 sec/tray 160 Operator cycle time (OCT), in secs 140 120 100 80 60 40 20 0 40 16 12 14 13 45 30 20 50 35 36 33 34 25 46 28 48 45 49 35 37 29 25 50 41 35 25 30 70 40
Operations cycle time Takt time

Timing for each operator is for one tray

Figure 13. Chart showing operator cycle time for RZX-3 product line (time taken by each operator in an assembly line to do their respective repeatable tasks) with respect to Takt time before line balancing (one of the Lean tools used in Pull Replenishment)

could be balanced better using a cell, in this case a U-shaped cell, that would allow the operators to move within the cell to execute multiple repetitive tasks. Upon re-balancing the line and the operators, the cycle times for the operators are shown in Figure 14.

Stripping of Press CTU Subassemb Bushing Assembly Assembly Press the Body Assemble Assembly Assembly Assembly Inspection Press the Clip, arm, Central Clip and Componen Assembly Press the Assembly Assembly Componen Screw the Insertion Resistance Continuity Resistance High Hi pot test, Operators

IJPPM 62,1

Figure 14 shows that operations 1-5 have been combined and allocated to one operator and operations 6-8 have been allocated to the second operator and so on. Each bar represents an operator and thus we can see that ten operators can perform the tasks of 30 operators and still meet Takt time of 165 seconds (red line) and thus meet customer demand. This means that the efficiency of the production line has increased three times for the same required output. Step 5: control: sustaining the gains The key to sustaining any process improvements is the control phase. To make sure that the gains made during improve phase are sustained and further improved to achieve and exceed the customer delivery performance goal of 95 percent every month, the initiative used many of the tools in the Lean/Six Sigma tool kit. The key metrics used to monitor the performance of the end-to-end value streams are ship-to-target and process capability (Ppk) to 14 days (market lead time specified by our customers). Also used are the CI concepts to statistically show to sales people and customers what the process is capable of and what would be the lead time (in days) for the service level to be 95 percent. For example, Figures 15 and 16 show the
Operator cycle time (OCT), in secs Takt time = 165 sec/tray 160 140 120 100 80 60 40 20
... 20 ...

102

121 95 95 92

122

120 91

125 105 70

ps

ps

ps

ps

ps

ps

ps

Figure 14. Chart showing operator cycle time for RZX-3 product line (time taken by each operator in an assembly line to do their respective repeatable tasks) with respect to Takt time after line balancing

0
36 48 35 34 25 35 ps 0, 5, 8, 0, 5, 9, 1, ,3 O ,3 ,3 ,2 ,4 ,2 ,2 ,4 25 45 50 46 45 33 37 49

0,

16

24

32

40

48

56

Figure 15. Histogram and statistical summary of delivery lead times (sales order creation to completion) before Lean Pull Replenishment

95% confidence intervals Mean Median 14.0 14.5 15.0 15.5 16.0 16.5 17.0

ps

Anderson-darling normality test A-squared 7.08 0.005 p-value < 15.959 Mean 9.247 SD 85.504 Variance 0.888794 Skewness 0.974757 Kurtosis 666 n 1.000 Minimum 9.000 1st Quartile 15.000 Median 21.000 3rd Quartile 58.000 Maximum 95% confidence interval for mean 15.256 16.663 95% confidence interval for median 14.000 15.000 95% confidence interval for SD 8.775 9.772

ps

70

12

15

95% confidence intervals Mean Median 4 5 6 7 8

Anderson-Darling normality test A-squared 1.24 p - value < 0.005 Mean 6.8163 SD 4.1012 Variance 16.8197 Skewness 0.492507 Kurtosis 0.912261 n 49 Minimum 1.0000 3.5000 1st Quartile 6.0000 Median 10.0000 3rd Quartile 15.0000 Maximum 95% confidence interval for mean 5.6383 7.9943 95% confidence interval for median 4.0000 8.0000 95% confidence interval for SD 3.4201 5.1236

Customer service excellence

103

Figure 16. Histogram and statistical summary of delivery lead times (sales order creation to completion) after Lean Pull Replenishment

histograms of delivery performance and key statistical measures before and after the value stream transformation using Lean Pull Replenishment methodology. Operations lead times before Lean Pull Replenishment implementation averaged 16 days. The SD between these points was 9.2 days. The CI was 31.2 days. This means that 95 percent of the time, the product will be built in o31.2 calendar days. Operations lead times after Lean Pull Replenishment transformation had a mean (average) of 4.8 days, with the median being four days. The SD between lead times was 3.9 days, with a 95 percent CI of nine days. This means that 95 percent of the time the product will be built in less than nine calendar days. Process capability analysis Figures 17 and 18 show the process capability analysis of the value stream before and after Lean Pull Replenishment deployment in the value stream.
USL
Process data LSL * Target * USL 13 Sample mean 15.9595 666 Sample N 1.79622 Shape 17.9587 Scale Observed performance PPM < LSL * PPM > USL 578,078.08 PPM total 578,078.08 Overall capability Pp * PPL * PPU 0.04 Ppk 0.04 Exp. overall performance PPM < LSL * PPM > USL 571,396.50 571,396.50 PPM total

16

24

32

40

48

56

Figure 17. Process capability of delivery lead times (sales order creation to completion) before Lean Pull Replenishment showing a Ppk of 0.04

IJPPM 62,1

USL
Process data * LSL * Target 13 USL Sample mean 2.27701 361 Sample n 1.94188 Shape 2.58509 Scale Observed performance PPM < LSL * PPM > USL 0.00 PPM total 0.00 Overall capability Pp * PPL * PPU 2.31 Ppk 2.31 Exp. overall performance PPM < LSL * PPM > USL 0.00 PPM total 0.00

104
Figure 18. Process capability of delivery lead times (sales order creation to completion). Lean Pull Replenishment showing a Ppk of 2.31

10

12

Actual manufacturing build time in average is 2.43 days. That means when the waste and non-value-added time were taken out and optimized the manufacturing line, value stream performance has attained performance (Ppk 2.31) is Six Sigma or world class level! The above is an example of just one value stream that has been transformed from no process capability to 14 days to one that performs at Six Sigma level. Figure 19 shows ship-to-target performance of the value stream before and after the transformation. As can be seen in Figure 19 chart, ship-to-target performance of this value stream has steadily shown improvements and has been sustained for Six Sigma delivery performance of 14 days and is in control over six months after improvements have been carried out in the transactional processes and in the manufacturing line. The average ship-to-target performance has increased from 64 to 89 percent. The telecommunications companys consistency has improved so much that the 95 percent CI is ten days. It has been communicated to the sales force that the company can deliver

Before Pull Replenishment 100.0% Ship to target performance 90.0% 80.0% 70.0% 60.0% 50.0% 40.0%

After Pull Replenishment UCL=98.5% X=90.0% LCL=81.4%

Figure 19. Ship-to-target percentage performance of DS-3 product line before and after Lean Pull Replenishment

Ja n1 Fe 0 bM 10 ar -1 Ap 0 r-1 M 0 ay -1 Ju 0 n1 Ju 0 l-1 Au 0 gSe 1 0 p1 O 0 ct -1 N 0 ov D 10 ec -1 Ja 0 n1 Fe 1 b11

Month

this product in ten days if it is able to gain more market share. Consistent use of this methodology across the strategic value streams in the company can help gain market share, defend price position with customers, improve inventory turns, and maintain a competitive advantage. Many cross-functional teams are working on multiple value streams and we expect to achieve similar results like the above model value stream. The new current state process map of the order management process as a result of all the improvements is illustrated in Figure 20. With the implementation of the new Lean Pull Replenishment system, customer service levels (ship-to-target) increased from 64 to 89 percent and operations lead time reduced from an average of 16 to 5 days. The telecommunications company transformed from MRP-Push system to Lean Pull Replenishment system using visual management tools. This included transforming the product line to a build-to-order model; cell redesigning for optimal operator/work balance; eliminated kitting and moving 100 percent to supermarkets on the line; and reducing the number of operators from 52 to 34 in the work center. As a result, labor productivity is three times for critical sub-assemblies in the work cell. Additionally, no finished goods inventory in the system implies better cash flow. Lessons learned The most fundamental lesson from designing and executing the Lean Pull Replenishment initiative at the telecommunications company has been that it is possible to achieve customer service excellence and improve financial results at the same time! Although this initiative may sound like an operations initiative, it has been very much a customer-centric and a business focussed strategy. It is necessary to engage with all facets of the business sales, product management, operations, supply chain, logistics, IT, HR, and engineering for a successful outcome. If it involved only the plants, success would have been very limited and the business would have been sub-optimized. Consistency of purpose among senior management was important for success. They stayed true to their commitment in allocating necessary resources for deploying and leading this initiative. Since the company started this initiative from scratch and was completely homegrown, senior management gave the core team members the time, space, and tools to learn new methodologies and skills, deploy them for a successful outcome. Senior management recognized that the Lean Six Sigma tools are not very complicated. It takes strong leadership and a change in culture, which is bottom-up driven to transform a company from decades-old MRP-based Push system to a radically new philosophy of a Pull system.

Customer service excellence

105

Quote 14 days

Receive order

Confirm 14 days

Release production order

Build order

Ship order

Figure 20. Simplified order management process after Lean Pull Replenishment has been implemented

IJPPM 62,1

106

Tight scoping of projects and value streams was important to build confidence and success in the companys changes. Boil the ocean types of projects did not lead the company anywhere. Throughout the deployment, the team followed the principle of start small, show success and replicate. This helped the teams take on meaningful projects, were able to learn the tools as applied to their projects and show quick wins. As was observed, success breeds success and soon, many people working on small projects were able to make a huge difference in delivering positive results to the overall initiative. Successful projects can also make the champions and sponsors to move onto the next project without taking the time to recognize extraordinary efforts and results by the initiative and project teams. Upon completion of every value stream, at the end of every kaizen and achievement of sustainable results, the management team made sure that the teams, from supervisors to operators were recognized consistent with local culture and budget. These small celebrations of victories energized our people to achieve more results. Recognition was a key part of the initiative, events very much looked forward to by the teams, as well as, senior management. Managerial implications Usually Lean and Six Sigma efforts are viewed primarily as operational efficiency projects. While a significant source of productivity improvements, many high-level managers, especially in sales, do not demonstrate support other than perfunctory verbal commitments. What made the Pull Replenishment initiative unique was the fact that the sales team participated in and supported us from Day 1. Pull was positioned and sold as a sales enablement initiative. This organizational alignment and support ultimately created the momentum necessary to improve customer delivery, employee engagement, and shareholder value. In the authors collective experience, there has never been sales involvement in any Lean Six Sigma rollout prior to the telecommunications companys Pull Replenishment. Generally, there is reluctance on the part of management and the organization as a whole to involve Sales. This is understandable. Sales should spend their time closing new business with customers. Lean Six Sigma is viewed as internally focussed, and thus not meriting sales time. Additionally, operations and Lean Sigma leaders traditionally do not have the kind of working relationships necessary to persuade their sales counterparts to participate in Lean Six Sigma. Neither do most operational-types possess the kind of vocabulary to excite and engage sales. Consider a typical improvement: We improved cycle time by 33% and reduced our costs by $1 M. Someone in sales would be unmoved by such efforts and possibly think that such improvements are par for the course. What did this case study do differently? It engaged in a voice-of-the-customer process from Day 1. By listening to sales frustrations with delivery, authors were able to analyze the data and correlate sales own actions to the problems they were experiencing. Expediting, it turned out, consumed 40 percent of sales time, yet produced statistically insignificant improvements. When the research team showed that data to sales leadership and told them, Well give you 40% of your time back to focus on closing the business, they reacted enthusiastically. Sales turned out to be the projects biggest cheerleader. In the pecking order of organizational influence, sales tends to exert disproportionate influence. Without sales engagement and approval, most largescale initiatives wither from organizational inertia. However, with sales backing, the

telecommunications company was able to execute significant process changes that resulted in an overall business transformation. The experience was synergistic. The team improved customer delivery from 50 to 14 days (95 percent CI) and thus demonstrated customer relevance. This made sales excited and positive, which created a positive ripple effect in the rest of the organization. Senior executives would do well to externalize their large-scale transformational initiatives, meaning, Make it meaningful to the customer! To put it bluntly, very few people care enough about pure productivity improvements to create the organizational alignment necessary to generate the required changes. Breaking down organizational silos and mapping enterprise processes end-to-end are prerequisites to success. This project focussed its efforts on selling to sales. These efforts paid off and the company was able to demonstrate significant statistical process capability improvements. Such improvements yielded considerable cost savings and customer delivery improvements. All the while, the entire organization was involved from sales through supply chain, which resulted in a high level of employee engagement that generated a real transformational breakthrough. Recommendations for future work Authors experience implementing Lean Pull Replenishment highlights a few key insights: one, that the vast majority of the positive results stem from replacing hedging behavior with trust. Trust, in our case, manifested itself in a simple 14-day replenishment script that everyone followed. Two, that simplicity is absolutely necessary. Without simplicity, executing a large-scale initiative down to the operator level is impossible. The multiple grand failures we saw in attempting corporate transformation initiatives highlight the execution risk created by complex undertakings. Third, and apropos of future recommendations, we believe our template works across the entire value chain from supplier through customer. The greatest value yet to be captured will come from Value-Chain Pull Replenishment. The company has already begun this journey with several suppliers, value-added resellers, and distributors. The data confirms that across the value-chain the inability to forecast accurately creates massive bullwhip effects. The bullwhip effect (or whiplash effect) is an observed phenomenon in forecast-driven distribution channels. It refers to a trend of larger swings in inventory in response to changes in demand, as one looks at firms further back in the supply chain for a product. The concept first appeared in Jay Forresters Industrial Dynamics (1961) and thus it is also known as the Forrester effect. Since the oscillating demand magnification upstream a supply chain is reminiscent of a cracking whip, it became known as the bullwhip effect. MIT created an elegant simulation called the Beer Game to illustrate the bullwhip behaviors (Kumar et al., 2007). Indeed, even with our internal operations on Pull Replenishment, we continue to experience tremendous operational and supply-chain disruptions from demand spikes created from our channel partners. The cost to the entire value chain is enormous. In the companys Juarez facility, the company hires and fires hundreds of workers in an attempt to keep up with the bullwhip. Companys purchasers order multiples of what is actually required from suppliers. Suppliers for the company choke on the massive orders resulting in costly delays. When the company superimposes end-customer demand on top of channel PO, it is observed that most of the expediting, overtime, and over-purchasing behaviors are largely unneeded. Every player in the value chain is reacting to each other thus exacerbating the problems.

Customer service excellence

107

IJPPM 62,1

108

Authors believe the solution is Lean Pull Replenishment based on customer usage. By replenishing companys channel partners in 14 days with weekly shipments, authors have modeled that 50 percent of excess inventory can be removed in the supply chain and reduce operating costs by 10-20 percent. The company is aggressively pursuing kaizen events with channel partners, who have responded with openness and enthusiasm. Nevertheless, creating trust when it comes to implementing across organizational boundaries will continue to be the key factor in execution. Trust is the elusive x-factor which enables timely and successful execution. Indeed, authors could write an entire future paper on this topic. Suffice it to say, companys longstanding relationships with its external partners have facilitated companys current foray into Value-Chain Pull Replenishment. Authors look forward to sharing the results in the near future.
References Allen, T.T. (2010), Introduction to Engineering Statistics and Lean Sigma: Statistical Quality Control and Design of Experiments and Systems, 2nd ed., Springer Verlag, London. Alukal, G. (2003), Create a Lean, mean machine, Quality Progress, Vol. 36 No. 4, pp. 29-35. Anthony, J. and Banuelas, R. (2004), Six Sigma in the software industry: results from a pilot study, Managerial Auditing Journal, Vol. 19 No. 9, pp. 1025-32. Choe, D. (2011), Lean pull replenishment to improve customer service, Management Innovative Exchange (MIX) Conference, February 21, Chicago, IL. de Koning, H. and de Mast, J. (2006), A rational reconstruction of Six Sigmas breakthrough cookbook, International Journal of Quality and Reliability Management, Vol. 23 No. 7, pp. 766-87. de Koning, H., Verver, J.P.S., van den Heuvel, J., Bisgaard, S. and Ronald, J.M.M. (2006), Lean Six Sigma in healthcare, Journal for Healthcare Quality, Vol. 28 No. 2, pp. 4-11. De Mast, J. (2007), Integrating the many facets of six sigma, Quality Engineering, Vol. 19 No. 4, pp. 353-61. Fliedner, G. and Vokurka, R.J. (1997), Agility: competitive weapon of the 1990s and beyond, Production and Inventory Management Journal, Vol. 38 No. 3, pp. 19-24. Furterer, S. and Elshennawy, A. (2005), Implementation of TQM and lean six sigma tools in local government: a framework and a case study, Total Quality Management, Vol. 16 No. 10, pp. 1179-91. Hahn, G., Hill, W., Hoerl, R. and Zinkgraf, S. (1999), The impact of six sigma improvement a glimpse into the future of statistics, The American Statistician, Vol. 53 No. 3, pp. 208-15. Harry, M. (1998), Six Sigma: a breakthrough strategy for profitability, Quality Progress, Vol. 31 No. 5, pp. 60-4. Johnson, J.A., Widener, S., Gitlow, H. and Popvich, E. (2006), A Six Sigma Black Belt case study: G.E.P. boxs paper helicopter experiment part b, Quality Engineering, Vol. 18 No. 4, pp. 431-42. Kumar, S., Chandra, C. and Seppanen, M.S. (2007), Demonstrating supply chain parameter optimization through beer game simulation, Information Knowledge Systems Management, Vol. 6 No. 4, pp. 291-322. Lakham, N. (2008), When push comes to pull: its kanban, Ultriva Inc, available at: www.ultriva.com/Articles/WhenPushComestoPull.pdf (accessed March 10, 2012).

LaLonde, B.J. (1997), Supply chain management: myth and reality?, Supply Chain Management Review, Vol. 1, Spring, pp. 6-7. Liker, J.K. (2004), The Toyota Way, McGraw-Hill, New York, NY. Linderman, K., Schroeder, R.G., Zaheer, S. and Choo, A.S. (2003), Six Sigma: a goal-theoretic perspective, Journal of Operations Management, Vol. 21 No. 2, pp. 193-203. Maguad, B. (2006), The modern quality movement: origins, development and trends, Total Quality Management, Vol. 17 No. 2, pp. 179-203. Martin, J.W. (2007), Lean Six Sigma for Supply Chain Management: The 10-Step Solution Process, McGraw-Hill, New York, NY. Ohno, T. (1988), Toyota Production System, Productivity Press, New York, NY. Rasis, D., Gitlow, H.S. and Popovich, E. (2002a), Paper organizers international: a fictitious Six Sigma Green Belt case study, Quality Engineering, Vol. 15 No. 1, pp. 127-45. Rasis, D., Gitlow, H.S. and Popovich, E. (2002b), Paper organizers international: a fictitious Six Sigma Green Belt case study, Quality Engineering, Vol. 15 No. 2, pp. 259-74. Shenchuk, J.P. and Colin, L.M. (2000), Flexibility and manufacturing system design: an experimental investigation, International Journal of Production Research, Vol. 38 No. 8, pp. 1801-22. Wang, G., Huang, S.H. and Dismukes, J.P. (2005), Manufacturing supply chain design and evaluation, International Journal of Advanced Manufacturing Technology, Vol. 25 Nos 1-2, pp. 93-100. Womack, J.P., Jones, D.T. and Roos, D. (1991), The Machine That Changed The World: The Story of Lean Production, HarperCollins Publishers, New York, NY. Yang, H.M., Choi, B.S., Park, H.J., Suh, M.S. and Chae, B. (2007), Supply chain management Six Sigma: a management innovation methodology at the Samsung group, Supply Chain Management: An International Journal, Vol. 12 No. 2, pp. 88-95. Further reading Allen, T.T., Tseng, S.-H., Swanson, K. and McClay, M.A. (2010), Improving the hospital discharge process with Six Sigma methods, Quality Engineering, Vol. 22 No. 1, pp. 13-20.

Customer service excellence

109

Corresponding author Sameer Kumar can be contacted at: skumar@stthomas.edu

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints