Part I of this blog series introduced the notion of long tails in modern supply chains. That blog post also introduced the vendor ToolsGroup and its solution for planning and optimizing finished goods in distribution environments.
So, How Does ToolsGroup Solve the Distribution Puzzle?
Most of the benefits are driven here by the distinctive Stock Mix Optimization capability that has delivered higher service levels with much less inventory for ToolsGroup customers. As its name suggests, the feature is used to define and manage the right mix at each location in the supply network to deliver the targeted customer service level.
The inventory optimization system automatically adjusts service level targets for individual stock-keeping units (SKUs) to obtain the most optimal “mix” of inventory, and hence to attain the target global service objective while deploying less inventory.
This is done by identifying the ideal mix of inventory, considering that each SKU-Location combination has its own efficiency curve, and taking into account variables like demand and supply variability, optimal lot size, upstream mix optimization, lead time, replenishment frequency, etc.
A closely related, and also proprietary, ToolsGroup capability is STS curves (”Stock to Service”), used to show the relationship between the average inventory level and the service level (e.g., $11.2 million of inventory yields the 98 percent service level).
As one varies the inventory investment, the level of service naturally changes. Plotting these individual points creates a curve where the inventory rises dramatically as we approach the ideal (albeit impractical) 100 percent service level. Hence, as a company’s service level targets go up, the amount of inventory required increases significantly, and the need for inventory optimization, which offers improved inventory efficiency, becomes increasingly more acute.
For more details on how ToolsGroup’s STS curves fare, when it comes to so-called “efficient frontiers” against other inventory optimization curves (that might use the poisson, normal or exponential distribution models), see this brochure. It is apparent from there that STS curves reach higher service levels (measured in percentages) at significantly lower average stock levels (measured in periods).
In addition to the above-mentioned capabilities and the largest installed base, domain experience and recently acquired brand new customers in distribution-intensive verticals such as food, consumer packaged goods (CPG), durables, retail/wholesale, aftermarket parts, fashion and so on, I believe that ToolsGroup remains “best in class” by differentiating with its traditionally strong demand modeling and inventory modeling capabilities.
This includes the ability to seamlessly model a wide range of product behaviors (for both fast- and slow-moving SKUs). ToolsGroup has had the demand modeling feature for more than a decade — it is basically its DNA. ToolsGroup has the ability to accurately model demand and inventory behavior, and like the mix optimization and STS curves features, it has long been an intrinsic part of ToolsGroup’s core DPM (Distribution Planning Model) product suite.
The suite contains several models with more than 20 parameters for modeling demand across a broad variety of conditions such as slow-, medium-, and fast-movers, different lot sizes, different lead times, highly variable lead times, during periods of new product introduction (NPI), and so on.
For those math and stats jocks that are interested in more detail, I suggest indulging in one of ToolsGroup’s original white papers entitled “New Concepts in Inventory Optimization”. The paper describes all of these rocket-science-like number-crunching models with gusto (several espresso shots are also recommended to keep the mind sharp & focused).
Self-Adapting via Demand-Sensing
Another DPM system differentiation is that it is an automatic and self-adaptive solution that yields a sustainable process for day-to-day operations. Namely, forecasting in traditional weekly and monthly buckets might seem deceptively adequate even for slow-moving items.
For instance, some item might show about a 90-unit sale trend every month, but does it all happen at the uniform rate of three units per each day, 30 units on three random days, or all 90 in a particular single day?
That’s to say, the same SKU may look like a fast-mover (with no zero-demand periods and relatively stable demand) if its demand behavior is observed in monthly buckets, but looks more like a slow-mover if observed in weekly buckets, and typically appears “lumpy” at the daily level.
Store demand signals thus have to propagate upstream through the retailer’s supply chain, synchronizing constrained replenishments across all supply chain tiers. The retail supply chain optimization system has to then dynamically monitor, sound the alarm, and compensate for errors caused by external factors.
In other words, the system has to understand the statistical nature of the demand signal and automatically self-adapt for changes in demand and supply characteristics. DPM accordingly features a special logic for demand signal propagation that combines the capabilities of both demand modeling and inventory modeling. The underlying idea here is to model the impact of replenishment parameters and constraints at each echelon of the supply chain.
To that end, a couple of years ago ToolsGroup espoused a strategy of an optimized replenishment process via the following three business practices:
The idea of the above three-legged-stool process is to rationalize the trade-off of inventory vs. service, given the constant demand variability and re-supply parameters (policies). With S&OP at its core, the practice helps with generating an inventory and service level plan by item and location that optimizes inventory mix and investment to achieve a defined global business objective.
In a nutshell, the goal is to create a reliable mix of inventory, positioned to achieve an enterprise-wide service policy, while leveraging and monitoring the demand signal and minimizing the supply chain noise.
By doing so, ToolsGroup DPM has long supported a much needed Service Segmentation and Inventory Optimization collaboration process between vendors and retailers. This way, inventory could be optimized from suppliers all the way to the retail shelf to deliver store-level service targets, thus better compensating for inevitable demand volatility and supply uncertainty.
Striving for Service Level Excellence
This idea of addressing service level degradation in the retail supply chain (whereby the 98 percent service level at supplier’s distribution center [DC] easily drops to 95 percent at regional DCs or vendor managed inventory [VMI] locations, and even down to 93 percent at retail stores outlets) was somewhat a progenitor of the current long tails theme at ToolsGroup.
Namely, all too often, the high “item/store” forecast error causes, besides stock-outs in the store, costly overstocks at regional DCs, which then propagates to an avalanche-like bullwhip effect throughout the supply chain.
Yet some industry-leading companies are breaking through customer service-level barriers and achieving fill rates above 99 percent, increasing top-line revenues and meeting heightened customer demands. To support this drive toward a higher standard, ToolsGroup in late 2007 introduced the Service Optimizer 99+ (SO99+) suite. The newly launched offering supplants DPM.
Prior to the launch, ToolsGroup’s cross-industry survey of more than 200 supply chain professionals and operations managers found that more than half of the companies surveyed were focused on improving customer service rates, with most important reasons being: achieving service level agreement (SLA) targets, increasing margin to the top line, and improving market share.
What has held companies back in the past from achieving the “99+” percent customer service level has been a combination of poor forecast accuracy and incorrect safety stock calculations. Persistent forecast bias and the inability to correctly allocate the forecast to a daily SKU and account-location level, have caused high daily forecast error. This situation, combined with widespread use of obsolete and inadequate inventory modeling, has lead to frequent stock-outs and excess inventory.
Basically, companies have two ways to achieve service level excellence. One is the traditional painful way of learning to live with systemic instability and to manage the uncertainty through an army of expeditors constantly fighting fires. This is certainly not sustainable in the long run due to excessive extra inventory, supply chain noise and volatility.
A more sustainable way is to model inventory reliably, localize the forecast and ensure adherence between plan and delivery. ToolsGroup’s Service Optimizer 99 suite addresses this problem head-on through Forecast MicroAllocation, a new module on top of the traditional demand modeling and inventory optimization modules.
Forecast MicroAllocation transforms demand plans into highly detailed forecasts, allocated by day, SKU and account-location. Combined with the existing statistical inventory modeling and mix optimization capabilities, it helps companies drive service levels up to coveted 99 percent and above, while improving supply chain stability and reducing overall inventories.
Service Level Inhibitors and How Forecast MicroAllocation Solves These
Why are forecasts, especially the near-term and more granular ones, notoriously inaccurate?
For one, forecasts have long been plagued by bias due to wishful financial thinking, whereby collaboration becomes negotiation and sales reward systems create expectations and pressure to perform. To that end, Forecast MicroAllocation constantly monitors for bias with a statistical and rules-based approach for determining alarm conditions and via continuous benchmarking of responsibility areas.
Secondly, demand profiling has been a daily battle by SKU/location, whereby oversimplified disaggregating of forecasts and actual demands yields high daily forecast error. In that regard, the new ToolsGroup solution disaggregates forecasts to daily, SKU and account (location) levels and integrates account-focused data into forecasting process.
Last but not least, real-time forecasting has hardly ever been accomplished via rudimentary consumption logic. For instance, companies are not able to say whether a forecast of 100 sold units in one week was consumed by a number of smaller orders from a lesser known retailers (mom-and-pop shops) or by a large order from retail giants (Kroger or Wal-Mart that will not take “no” for an answer). Accordingly, Forecast MicroAllocation performs detailed demand pattern sensing and features a forecast consumption logic with reactive adjustments.
The final part of this blog series will describe how the above-mentioned ToolsGroup’s capabilities have been parlayed into solutions for long-tail environments. In the meantime, your comments, opinions, experiences, etc. are highly encouraged and appreciated.
[…] Part II of this blog series explained ToolsGroup’s value proposition for achieving service level excellence in distribution environments. The point of the Service Optimizer 99+ (SO99+)? suite’s name? is that a “99+ percentage” represents the gold standard in customer service levels, and it takes a product purposely built to achieve service level excellence and to support such a high standard. […]
[…] one hand, the ever? more pressured manufacturing environment demands acceleration of the stock-keeping units (SKU) mix and shorter lead times, all due to ever more demanding and fickle customers; […]