Supply chain integration  Various supply chain strategies  Push strategies  Pull strategies  Push-pull systems  Matching products or industries with supply chain strategies  Impact of the Internet on supply chain integration  Effective distribution strategies  Direct shipment  Warehousing  Cross-docking

Push Strategies  Production decisions based on long-term forecasts  Ordering decisions based on inventory & forecasts  What are the problems with push strategies?  Inability to meet changing demand patterns  Obsolescence  The bullwhip effect:  Excessive inventory  Excessive production variability  Poor service levels  Hard to predict production capacity or transportation capacity

Pull Strategies  Production is demand driven  Production and distribution coordinated with true customer demand  Firms respond to specific orders  Pull Strategies result in:  Reduced lead times (better anticipation)  Decreased inventory levels at retailers and manufacturers  Decreased system variability  Better response to changing markets  But:  Harder to leverage economies of scale  Doesn ’ t work in all cases

Pull strategies – a Kanban system Production cards Move cards Inbound bufferOutbound bufferInbound bufferOutbound buffer Retailers Suppliers

Push-Pull Systems Push StrategyPull Strategy Raw Materials End Customer Push- Pull Boundary Supply chain time line

Push-pull systems  A shift from a Push System...  Production decisions are based on forecast  …to a Push-Pull System  Initial portion of the supply chain is replenished based on long-term forecasts  For example, parts inventory may be replenished based on forecasts  Final supply chain stages based on actual customer demand.  For example, assembly may based on actual orders.

Consider Two PC Manufacturers:  Build to Stock  Forecast demand  Buys components  Assembles computers  Observes demand and meets demand if possible.  A traditional push system  Build to order  Forecast demand  Buys components  Observes demand  Assembles computers  Meets demand  A push-pull system

Push-Pull Strategies  The push-pull system takes advantage of the rules of forecasting:  Forecasts are always wrong  The longer the forecast horizon the worse the forecast  Aggregate forecasts are more accurate  Risk Pooling impact  Delayed differentiation is another example  Consider Benetton sweater production

What is the Best Strategy? Pull Push Pull Push Demand uncertainty (C.V.) Delivery cost Unit price L H HLHL Economies of Scale I Computers II Furniture IV Books & CDs III Grocery

Selecting the Best SC Strategy  Higher demand uncertainty suggests pull  Higher importance of economies of scale suggests push  High uncertainty/ EOS not important such as the computer industry implies pull  Low uncertainty/ EOS important such as groceries implies push  Demand is stable  Transportation cost reduction is critical  Pull would not be appropriate here.

Selecting the Best SC Strategy  Low uncertainty but low value of economies of scale (high volume books and CDs)  Either push strategies or push/pull strategies might be most appropriate  High uncertainty and high value of economies of scale  For example, the furniture industry  How can production be pull but delivery push?  Is this a “pull-push” system?

Characteristics and Skills PortionPushPull ObjectiveMinimize costMaximize service level ComplexityHighLow FocusResource allocationResponsiveness Lead timeLongShort ProcessesSupply chain planningOrder fulfillment

Locating the Push-Pull Boundary  The push section:  Uncertainty is relatively low  Economies of scale important  Long lead times  Complex supply chain structures:  Thus  Management based on forecasts is appropriate  Focus is on cost minimization  Achieved by effective resource utilization – supply chain optimization  The pull section:  High uncertainty  Simple supply chain structure  Short lead times  Thus  Reacting to realized demand is important  Focus on service level  Flexible and responsive approaches

Locating the Push-Pull Boundary  The push section requires:  Supply chain planning  Long term strategies  The pull section requires:  Order fulfillment processes  Customer relationship management  Buffer inventory at the boundaries:  The output of the tactical planning process  The input to the order fulfillment process.

Locating the Push-Pull Boundary

Demand-driven strategies  Demand forecast: Using historical data to develop long-term estimates of expected demand  Demand shaping: Determining the impact various marketing plans such as promotions, pricing discounts, rebates, new product introductions and product withdrawal on demand forecasts  Inaccuracy of the forecast has a detrimental impact on supply chain performance: lost sales, obsolete inventory, inefficient resource utilization  Employing supply chain strategies to reduce the impacts of forecast inaccuracy  Select the push-pull boundary so that the demand is aggregated over different dimensions: products, geography, time  Use market analysis and demographic and economic trends to improve forecast  Determine the optimal assortment of products by store to reduce the impact of competing SKUs in the same market  Incorporate collaborative planning and forecasting processes with customers to better understand market demand, impact of promotions, pricing and advertising

Impact of Internet on SCM  What does internet change for a supply chain?  Enables a whole new business model.  Online purchasing, direct shipping, auctioning, secondary markets  Improves or enables integration between different parties of the supply chain  Enables information sharing  Enables collaboration  Reduces lead times  Reduction in order processing times  Improves product availability

Impact of internet  E-business: a collection of business models and processes motivated by Internet technology and focusing on improvement of extended enterprise performance  Business-to-consumer (B2C): “direct to customer”, retail activities over the internet  Business-to-business (B2B): business conducted over the internet between businesses  Impact of internet  Move from push to pull systems  Significant failures as a result  Move from pull systems to push-pull systems  Many click-and-mortar companies established. Many brick-and- mortar companies opened online stores. Some has been successful: Dell, Cisco, Amazon. Many have failed

Grocery Industry  Example: Peapod  Founded early 90s  Implemented a pure pull system. When an order is received, the products are picked from a nearby supermarket. Problems  Service problems: significant stock-outs  Reduced profit margins  Moved to a push-pull system. Set up a number of warehouses  Stock-out rates are lower  As compared to traditional grocers, the demand is aggregated over a larger geographical area  Transportation costs versus response time  Shipments in small batches  Response time within 12 hours  No sufficient density of customers to control transportation costs  Many failures so far: shoplink.com, streamline.com, etc

Book industry  Example: Amazon.com. World’s largest bookseller  Founded in 1994  Implemented a pure pull system where it utilized Ingram Book Group to supply customer demand. Appropriate when Amazon was building its brand name. Issues became clear later when the demand increased  Ingram’s distribution capacity supports many other booksellers. Service issues for Amazon during peak demand  Amazon had to share its profit margins with Ingram  Amazon established several warehouses, where the inventory is procured using a push strategy, orders are shipped using a pull strategy  May still use a pure pull strategy for slower items  World’s largest bookseller with $3.9 Billion sales in But yet to make a profit  Response time is not as critical as grocers. May use parcel services.

Retail industry  Traditional retailers added online shopping component to their offering: Wal-Mart, Kmart, Target and Barnes and Noble. Advantageous over pure Internet companies  They already have the distribution and warehousing infrastructure in place  Established brand name  Easy returns and reverse logistics  Different strategies for different products  High-volume, fast-moving products stocked in stores and available online  Low-volume, slow-moving products are stocked centrally and available only online  Moving from a traditional business to internet based business may require different skills that are not present in many traditional businesses

Impact of internet on fulfillment Traditional fulfillmentE-fulfillment Supply chain strategyPushPush-pull ShipmentBulkParcel Reverse logisticsSmall part of businessImportant and highly complex Delivery destinationSmall number of storesLarge number of geographically dispersed customers Lead timesRelatively longRelatively short

Spearman  Analyzes service level in pull systems  Kanban system versus a base stock policy  Kanban would not place an order for more parts if a demand had arrived when there was no stock in the outbound stock point  Kanban has a constant WIP  Stochastic ordering and Kanban systems. Two Kanban systems A and B, A has service times S A and B has service times S B.  If S A is stochastically larger than S B, then system A has worse service than system B  Two normal distributions with same variance,  A <  B  If S A is stochastically larger than S B, in the sense of increasing convex ordering, then system A has worse service than system B  Two normal distributions with same mean,  A <  B  More Kanban (WIP) leads to higher service in Kanban systems

Spearman - continued  The study argues that the superiority of Kanban systems is not the fact that material is pulled everywhere, but the fact that the WIP is constant  Develops a system called a constant WIP system, or CONWIP  Pulling only at the first station  Pushing on the rest of the chain using CONWIP backlog  Can utilize common setups  Consumption of item A may lead to start of the production for item B  CONWIP system outperforms Kanban system  Better service  May be more appropriate for large setup times, changing product mix.

Rajagopalan  Develops a model to decide whether an item should be MTO or MTS  Items that are MTO. No inventory cost. Only setup time. Orders of a same item within a bucket still shares the setup. Ship to customer within T with probability P O  Items that are MTS. Follows an (r, q) policy. Cycle stock and safety stock. Service level (type I) of P S  Decisions for items are not independent of each other. Trade-offs  Making an item MTO reduces the inventory for that item, but  Leads to more setups, thus higher capacity utilization and larger lead times  Leads to more variability in lead times  Thus larger safety and cycle stocks for MTS items, poorer service for MTO items  Decreasing the lot size for MTS similar effects

Rajagopalan - continued  Insights  MTO systems may be less costly but not always feasible  Medium demand items are attracted to MTS  Higher holding costs -> MTO  Higher setup costs -> MTS  Larger processing times with high demands -> MTO  Impact of priority queues (MTO orders are prioritized)  Leads to more MTO items  However leads to more cycle stocks and safety stocks for MTS items

Stochastic ordering  =30  =5  =20  =5

Increasing convex stochastic ordering  =20  =8  =20  =5