How to Plan and Manage in Times of Uncertainty and Volatility? Based on Reality and Facts, Duh! (Part 1)




My recent attendance at Progress Revolution 2011, Kinexions 2011, and several Boston APICS Chapter professional development meetings, where a plethora of companies talked about their operational experiences of late, made me realize that “business as usual” practices no longer work.

For one thing, while long-term planning remains an important exercise for senior executives’ strategic and visionary purposes (evaluating what-if scenario options and making long-term decisions), many recent events have caused serious paradigm shifts.

Trying to make rocket science-based optimized long-term plans has nearly become a fool’s errand. For example, the recent Japanese earthquake and the still ongoing floods in Thailand had quite the impact on high-tech brand owners worldwide, given that some finished goods (gadgets) manufacturers source 30 percent or even more of their critical electronic components from these regions.

At Kinexions 2011 we all heard the following sad supply chain stats: 48 percent of weekly demand plans have errors, with only 5 to 10 percent average net promoter scores (NPS), as the measure of customer loyalty, and measly 0.06 percent compound annual growth rate (CAGR) on return on capital (ROC) as results.



The old adage is that “the best laid schemes of mice and men go oft awry.” For perhaps the best example of this adage, even outside the realm of supply chain management (SCM), one can think of the Boston Red Sox’ epic September 2011 collapse that is still fresh in our minds.

Planning vs. Execution – A Cautionary Baseball Tale

Indeed, it is difficult to argue with the Red Sox front office’s 2011 season planning during the past off-season. The proprietary “Carmine” program that crunches the baseball stats and suggests the roster and individual players’ pay based on their (historical) performance did very good work according to all sports experts and analysts. After all, this “Moneyball” tool has been largely credited with Red Sox’ two Major League Baseball (MLB) World Series in the past decade.

The 2011 Red Sox were touted by many experts as the best team ever and a sure 2011 World Series finalist by some really renowned baseball connoisseurs at ESPN and elsewhere. In SCM lingo, there was a “forecast consensus” of a sort prior to April 2011 that worked well until the end of August 2011.

So, what on earth happened in September 2011, and how did the team blow a 9.5 game lead in such a record short time? During his poignant and captivating 70-minute interview on the local Boston sports radio show, Red Sox owner John Henry was asked whether there was anything different that he could have done during the collapse, in hindsight. He was only able to utter a facetious answer that perhaps he could have thrown a few innings in September, given how atrociously bad were his overpaid and under-disciplined star pitchers at the time (the latter part he did not say, but we got the implicit message).

Well, perhaps his answer should have been to have some “plan-monitor-respond” (“detect-and-react” or “sense-and-respond”) mechanism (a feedback loop) in place, so as to be able to react on his feet. Jack Welch’s following quote from his “Straight From the Gut” book might come in handy here:
“Business success is less a function of grandiose predictions than it is a result of being able to respond rapidly to real changes as they occur.”

The Red Sox’ long-term 2011 season planning was arguably fine, even outdoing their NY Yankees foes during the off-season.  No one in the MLB punditry was disputing the new players’ acquisitions per se, other than whether they were overpaid (it is Henry’s money, after all).

But what about the team’s short-range execution and responsiveness to unpredictable events? Shouldn’t someone in the Red Sox brass have noticed weird clubhouse and dugout practices and reacted to star players ganging up and chucking down fattening chicken and guzzling beer cans during games? If there was some quick decision-making system in place, perhaps someone could have reacted in time with some urgent trades after noticing their star players gaining 15 pounds during the season while their earned run allowed (ERA) figure “gained weight” (from 3 to well over 6) as well?!

Whacky Demand Trends Playing Havoc with Planning

Coming back to non-sports businesses and supply chains, one can think of unprecedented changes in the level of demand that turned all traditional supply chain planning (SCP) practices upside down. Indeed, many might remember what happened in the 2009 holiday season with precipitous down demand trends that butchered all the traditional seasonality trends.

Moreover, due to the ongoing decreased buying power of consumers, there have been dramatic changes in product mixes. For example, new car sales have lately been ever-smaller percentages of total car sales (remember the “cash for clunkers” initiative a while back?).

As a result, global suppliers are reducing their inventories, except for petroleum, coal, beverage, and tobacco products, which then creates supply shortages and the need to properly allocate parts for particular orders. For their part, as a result of depressed aggregate demand, customers are reducing inventories as well: merchant wholesale inventory figure was at its maximum in July 2008 and much lower ever since.

What is the Impact of these Supply-Demand Trends?

In such an upside down reality, the efficacy of traditional intrinsic time series forecasting tools (that predict future based on history) is degraded. It has never been fun to drive a car by looking in a rear view mirror, except for stunt purposes. Thus the cost per forecast unit goes up, while forecast accuracy is going down (often below 50 percent).

Many companies have painfully realized that their forecasting ability is not going to get any better and at the customer and item level 30 days out the mean absolute percentage error (MAPE) seldom gets above 75 percent. Anecdotal evidence is that these numbers are only true for consumer product goods (CPG) with longer product life cycles and relatively slow technology changes.

In the high-tech sector, including consumer electronics, the MAPE is seldom above 50 percent. From this we can deduce that the supply plan is at best correct half the time, though of course assembly postponement strategies tend to increase the accuracy as one moves back up the supply chain. However, the further upstream the supply chain one goes, the greater the consequences of the bullwhip effect are.

In summary, as companies have greater exposure to the aforementioned “demand shocks”, suppliers have less buffer inventory (safety stocks), whereby lead times (latencies) are increasing. To this end, there have been the following two classical responses of SCM executives at enterprises:

  1. Increase safety stocks -- and good luck with the chief financial officer (CFO) who will “love” this tying up of his/her working capital and cash.

  2. Reduce customer service levels -- and increase the number of “stock out” situations with customer disappointments as a result.


Needless to say, these are not great choices at all: at least you only get yelled at for having too much stock (and risk of obsolescence), but you can get fired for not having enough and losing business. Pick your poison, a root canal or a colonoscopy?

What could be a Better Way?

Make no mistake, the point here is not to dismiss planning outright: after all, good planning helped the Red Sox win two MLB championships. However, when talking about planning you have to be careful to differentiate between long-range and short-range planning.

Long-range planning is what SCP and enterprise resource planning (ERP) applications have handled well for a long time. But the harsh reality of today is that there is no more a “let's lay a huge rigid 'push manufacturing' facility down” (a la Ford’s River Rouge complex) attitude in this demand and supply volatility within complex global supply chain networks.

In addition, Ford’s ancient practices of vertically integrated supply chains (where an enterprise is in total control of all of its suppliers) and telling customers that “it can be any color as long as it is black” can now only be found in schoolbooks and not really in practice. The sad truth, however, is that ERP systems from a few decades back were designed to serve rigid enterprises whose business models would not change often, if at all.

At the same time, supply chains have become highly non-linear systems that must balance a whole host of competing objectives based upon incomplete information. And yet the likes of former i2 Technologies and Manugistics (now at JDA Software) and SAP threw long-term planning optimization at the problem in the 1990s, assuming not only that the problem could be linearized, but that the input data was complete and precise. If we add to that the massive outsourcing from multiple places, product mixes and mass customization that have taken place in the last 20 years, which all leads to longer lead times and more volatile demand, the problem has shifted from “boiling the ocean” with an optimized plan to knowing what is happening and making a “best possible” or “good enough” decision quickly.

Jabil Circuit, which designs and manufactures electronic circuit boards for major brand owners (or OEM’s) in a diverse group of industries including automotive, computing and storage, consumer products, medical, networking, peripherals, and telecommunications, presented the following stats at Kinexions 2011: a US$12.5 billion annual spend, 55 manufacturing facilities worldwide, 150 OEM customers, and 12,500 suppliers delivering thousands of components (some of which are special components and subassemblies while others are common parts). Fast decision-making at the time of truth (fulfilling orders) is critical in such complex environments.

Knowledge is power in planning and can help in multiple ways such as by improving lead time reliability, information visibility, demand management (shaping), etc. But SCP and ERP systems have long been disconnected from supply chain execution (SCE) and manufacturing execution systems (MES) because they deal with different issues and look at different data sets and accompanying analytics.

SCP vendors were focused on creating an ideal plan, but, because of demand volatility and product proliferation, it is extremely difficult to create an accurate forecast, and therefore the supply plan one creates is by definition inaccurate. Using long-term optimized planning to create an optimized supply plan isn’t going to improve the forecast, and when the forecast is incorrect, the quality of the optimization is then suspect.

There were also the following two architectural flaws that prevented i2 and all other SCP solutions from providing a capability to merge planning and execution:

  1. They provide separate modules for each planning process that use different data models and different analytics, and therefore require movement of data between these modules to evaluate the end-to-end effect of a change. This need in turn prevents them from providing a single view. While they may sell a separate module that provides a single view of the supply chain, by definition this view can only be inspected and not exercised because the analytics reside in other modules.

  2. Because of the data movement between separate modules and because many of the individual modules rely on optimizations engines that take a long time to run, the latency in decision making is often longer than the time available to make a decision in the execution window.


On the other hand, the problem for pure-play SCE and MES vendors is that they do not connect the dots. At best they can only tell you that something happened that you did not expect (or a non-event in the case of something desired not happening), not what the consequences are and therefore the severity of the event. Without that capability how can you determine the most critical issues?

If we know that our plans are wrong to start with – because we can’t forecast and predict accurately – why do we still insist on executing to a plan that is wrong? On the other hand, if you need to make a change, shouldn’t you be able to evaluate the holistic consequences of your decision?

Creating a sound plan is the first step, and then volatility, uncertainty, complexity, and ambiguity (VUCA) occur, so what now? What if the corrective action impacts other demand of high value? What if you expedite supply to satisfy an important customer, but that action completely erodes your margin?

Where is the next breakthrough going to come from: planning better or learning to respond profitably to real demand? While we should always do what we can to improve our forward visibility to demand (including predictive and/or near real-time analytics, demand sensing, etc.), if we start with the premise that the forecast will always have significant errors, the capabilities to develop are the ‘time to detect’ and the ‘time to correct’. These are the essential elements of response management and responsive supply chains: knowing sooner and responding with confidence.

Part 2 of this series will continue the responsive supply chains discussion, present some notable solutions, and conclude with some best practices and processes. In the meantime, please send us your comments, opinions, etc. We would certainly be interested in your experiences with these software tools and your best practices in this economic volatility and uncertainty.

Related Reading
 
comments powered by Disqus