Throughout my career, I’ve seen a number of companies apply ERPs to manage their service parts supply chains, and despite their best efforts, they encounter the same problems: disappointing fill rates, improper mix of parts, and unnecessarily high inventory costs. Why did they go down that path?
The challenge is that at the highest levels of abstraction, every supply chain shares the same terminology. They all forecast, set target inventories, source and deploy. But how supply chain professionals execute these processes depends on whether they're managing manufacturing or service supply chains. Failure to rigorously investigate whatever solution you’re planning to deploy can lead to staggeringly disappointing results.
Why? The big-name ERP systems are built to support manufacturing supply chains, which operate under conditions entirely different than those of service supply chains. There are at least six major reasons why manufacturing supply chain management techniques are inappropriate for service parts operations:
Manufacturing supply chains focus on the finished good. The forecast is in units of product. The only time you apply logic to the components is after an MRP explosion to determine their dependent demand for sourcing.
In addition, the products are assembled in a handful of plants. This scale is minuscule compared to that of a service operation, which must account for the individual demand of tens, if not hundreds of thousands of parts for multiple generations of products that are still in use.
These parts can be represented in thousands of demand-triggering and supply-carrying locations. The scale of this problem is exponentially larger than it is for a manufacturing operation. That’s why it’s so difficult to adapt ERP manufacturing modules for service. You cannot put feathers on a dog and proclaim it a duck.
The foundation of any supply chain solution is demand planning. Manufacturers and retailers like demand, and attempt to stimulate its growth. This is philosophically different than a service organization: Demand indicates that a product has failed.
One does not attempt to stimulate product failure. This means that the data used in product forecasting, cost / margin, advertising budget, market share, etc. have no place in predicting service parts demand. Parts demand is probabilistic and based on failure rates, product activity profiles, and maintenance schedules, requiring a totally different tool set.
The probabilistic nature of parts demand is incentivizing many parts planners to use Internet-of-Things data to help them develop more robust forecasts. My colleague, Vinod Arekar, discusses the implications of this trend in further detail below:
Manufacturing has the luxury of only needing to accommodate current products’ bills of materials, and can only source all of the component parts from new-buy sources.
Service, on the other hand, must plan parts availability for every product configuration supported, many of which may be decades old and no longer have new-buy sources. This increases the scale of parts forecasted and planned, again, by exponential factors.
As an additional consequence, an alternate, often primary, source of supply has evolved: reverse logistics and repairable parts. For many service verticals (e.g. aviation, industrial equipment) the high cost and long lead time of highly engineered parts requires that they be returned, repaired and put back into service as an alternative to buy or build. All manufacturing and retail supply chain solutions have to handle is the trickle of customer returns.
Optimization, to a manufacturer, is maximizing the productive capacity of facilities with minimal supporting component inventories. Since there are typically one or two plants, the optimization logic has evolved around capacity planning and load leveling.
For service parts, it is all about meeting part or product availability targets at minimum inventory investment. Service networks have multi-echelon networks with central, regional, branch / dealer and even mobile engineer van stocks. When this is compounded with differing availability targets by part group, location type, and contract service level agreements covering tens of thousands of parts, inventory planners need a very highly specialized and complicated set of algorithms. One such algorithm is multi-echelon inventory optimization, a capability I discuss in the post below:
Manufacturers and retailers discontinue products that do not meet sales goals. Service has no such luxury. In fact, slow-moving parts often represent critical, high availability, high-cost and long-lead-time items – the hardest for which to forecast and plan inventory deployment. This further requires additional specialized functionality to deal with forecasts of fractional units and pooled inventory. Without these tools, the only alternative is to deploy scads of expensive and underutilized inventory.
Manufacturers do not have to decide where to deploy part inventories. The parts come into the plant and go out as products into another organization's supply chain.
For service, you can either put a part everywhere or deploy logic that creates a fluid distribution network, where typically high-cost, low-demand parts can be redistributed as demand occurs. When combined with service-centric inventory optimization, a very powerful platform supports higher levels of availability at lower costs of investment – the supply chain Holy Grail.
As with most things, you need to use the right tool for the right job. Manufacturing and service supply chains are fundamentally, philosophically, and architecturally different. Manufacturing supply chain tools won’t work for service (and vice versa). Changing marketing collateral and PowerPoint presentations does not transform a manufacturing solution into a service solution.