Category Archives: Problem Solving

Problem Solving Series II: Formulate the Problem

This is the second post in a series of posts designed to introduce you to problem solving strategies that you can use to attack your sourcing and supply chain problems. Last Sunday we discussed eleven strategies you could use to help you understand the problem. Today we are going to discuss a generic methodology for attacking problem formulation.

The easiest way to formulate a problem is to simplify the task. There are three basic strategies you can use to accomplish this.

( 1 ) Simplify the Problem

If the problem is too challenging, simplify the problem. In spend reduction, don’t look at the category but look at each commodity. Don’t look at the commodity as a whole, but as a collection of cost components. Look for the key elements, visualize them, and redefine the problem as a simpler problem on those key elements.

( 2 ) Solve One Part at a Time

If you need to rationalize your supply base, instead of trying to rationalize your supply chain as a whole, just focus on rationalizing it for a key category. Then repeat the process. Instead of trying to define the optimum transportation route from Mumbai, India to Atlanta, Georgia, redefine the problem as trying to find the optimal land and sea route and the optimal air route, and take the best option. Then, instead of trying to find the optimal land and sea route, find the best routes from Mumbai to the nearest ports, the available sea routes from those ports, and the land routes from the available receiving ports to Atlanta. The best route will then be a combination of one route in each set. If one option in each set is clearly the best, and they match up, you have your route. If not, simply load this small amount of data into a network optimizer and you will quickly have your answer.

( 3 ) Redefine the Problem

Instead of trying to determine the minimum number of suppliers you can use to supply a given product to your different receiving centers with seven days notice, just focus on finding a fixed size sub set of suppliers from your current supply base that cover most of the receiving points. Chances are you can eyeball this. Then, for each receiving center not covered, choose the supplier that covers the most remaining receiving points. Continue until all your receiving centers are covered. If your starting set was good and most of your suppliers can cover a significant number of your receiving points, this technique could quickly replace dozens or hundreds of suppliers with a handful. You’ve accomplished significant consolidation with very little effort. If you still have too many suppliers, work with the best remaining suppliers to cover more centers more flexibly. Instead of trying to optimize a network with potentially hundreds or thousands of nodes, you are optimizing core supplier capabilities, which is not only an innovative solution, but a better solution in the long run.

Next Sunday we will discuss some strategies to aid with the third step of the Operations Research Modeling Process: Model Construction.

Problem Solving Series I: Understand the Problem

Last Sunday we discussed the Operations Research Modeling Process because it described a basic problem solving process that you could apply to sourcing and supply chain problems.  Today we are going to start a series of posts that are going to discuss some problem solving strategies that will help you in the application of this process.  This series will run on Sundays for at least the next five weeks, so keep checking back for more tips, tricks, and insights that you can use.

The first step of the Operations Research Modeling Process is “Recognize the Problem”.  Today we are going to discuss eight problem solving strategies that you can use to aid in this process and clarify them with examples from the sourcing space.

( 1 ) Clarify the Problem

It is often easier to solve specific problems then vague generalizations.  If you need to reduce spend on a high-spend category, then determine precisely what commodities or materials make up that category and reformulate the problem as reducing spend on each of the commodities or materials individually.  If you need to help engineering identify the appropriate widget to source for a new prototype, determine what properties are required vs. what properties are desired.

( 2 ) Identify Key Elements

Some aspects of a problem are more relevant than others.  Take our spend reduction problem.  Reducing spend on a commodity can be re-stated as reducing spend on the individual spend components.  Identifying which components are the highest spend, especially with respect to a should-cost model or market index, will help you greatly. 

( 3 ) Visualize the Relevant Process

Sometimes you know all the factors that contribute to a problem, and sometimes you do not.  Our spend problem is also relevant here.  Maybe global transportation costs from your suppliers in one of your low cost countries are higher then you think they should be.  However, just looking at costs per unit or quoted freight rates are not going to give you the whole picture.  However, if you visualize the process, which starts with a crate being packed and put on a palette, continues with the palette being loaded into a truck, transported to a warehouse, sitting there for three days, being loaded onto a boat, crossing the ocean, being unloaded to a warehouse where it sits for four days, being loaded onto another truck, and finally delivered to your factory, you will see that there are significant excess inventory costs on both sides of the ocean.  If you could synchronize shipments on both sides of the water, you could eliminate significant built in inventory holding costs even if the land and sea freight rates are not negotiable.

( 4 ) Draw a Diagram

Visualizing a problem can increase your understanding, but if the problem is complex, it might be a lot easier to draw a diagram to help you keep track of the different aspects of the problem you have visualized. 

Lets say that you need to consolidate your supply base because you are using over a dozen suppliers with over three dozen collective shipping locations for a commodity and you have determined you should be able to mitigate single-supply risks with only three or four suppliers and you need to choose who those suppliers will be.  Lets also say that you want to retain a flexible demand-driven supply chain and make sure that every receiving location can be supplied by at least one supplier in seven days or less.  Finally, you need to reduce costs by at least 10%.  We now have a problem that is beyond simply considering Total Cost of Ownership (TCO), which is the current state of the art supported by most of the advanced decision optimization systems on the market today, and well into Total Value Management (TVM), which is, generally speaking, not (directly) supported when your problem includes more then one of the standard well-defined optimization problem space boundaries, and this one includes traditional sourcing (TCO/TVM-based determination), inventory and logistics, and network design.  In order to use even your state of the art TVM-based decision optimization system, you will first have to determine which groups of suppliers can satisfy your time-based shipment requirements as well as which business rules dictate which suppliers can and cannot be used in conjunction.  Drawing a diagram of all the potential shipping and receiving destinations will help you determine which supplier shipping centers can supply which of your receiving centers, and that information will be crucial to building a valid model.

( 5 ) Consider a Specific Example

Sometimes problems are just too abstract for us to solve easily.  For example, let’s say we want to consolidate our supply base but want to know the minimum number of suppliers it will take to meet our supply chain flexibility requirements.  If we have access to the right solvers, a solver can determine this for us given a mathematical model.  However, we need to be able to appropriately specify the model and we need a check for its validity.  A specific example will help us here.  Choose four suppliers and determine whether or not each receiving center can be supplied by those four suppliers.  If it can, we know that the optimal answer is at most four, so if the solver comes back with a higher number, we did not properly formulate the model, and probably need to clarify our understanding.  If it cannot, we know that those four suppliers cannot constitute a solution and any solution with four or less suppliers must include a different set of suppliers.  Furthermore, by working out all the relationships, we will have specified the majority of constraints for that problem instance and if we feed just that sub model to the optimizer, then we have a good check on the accuracy of that model.

( 6 ) Consider Extreme Cases

Although we again consider specific examples here, we are testing what happens at the boundaries of our problem space.  For example, let’s say we are trying to determine whether or not the fuel surcharges that were charged by our current carrier over the last year are reasonable and fair.  To do this, we would build a should cost model for each route based upon an the lowest fuel rate and the highest fuel rate for the year (the extreme cases).  If most of the average charges for each route fell into our should cost ranges, then the rates were probably fair.  If most fell outside of the range, the rates were probably not fair and we should either change carriers or build in maximum fuel surcharges based upon a standard market index.

( 7 ) Consider Levels

Sometimes solving a problem requires preventing the problem.  This requires determining what led to the problem and fixing that process.  For example, let’s assume a preferred supplier is adamant that rates cannot be reduced because the profit margin is too slim even though direct competitors are quoting 10% less.  If the direct competitors are located in the same geographic market, chances are that either the supplier’s overheads are too high, the amounts they are paying for the raw materials are too high, or their processes are inefficient.  If you examine each in turn, you might find an obvious weak point that you can help them fix and allow you to continue your relationship at a significantly lower cost.  For example, let’s say their material costs are too high.  If this is a product you buy from multiple suppliers to mitigate risk, then you could save each of your suppliers money, and lower your costs, by sourcing the raw materials for all of them at a higher volume and associated spend leverage.

( 8 ) Change Perspective

Let’s say you are in a tough negotiation that isn’t going way.  Consider the viewpoint of the other party.  Why won’t she go lower?  Why can’t she?  If it is because their production costs are simply to high to support lower rates, then maybe working with them to find ways to produce the product at a lower cost is the answer.

Now, I know all of these are obvious, but in the midst of a high-pressure dilemma, especially one that must be solved quickly, it is easy to forget a couple of these strategies, and you never know which approach is going to work in advance.  Reviewing them regularly and attempting to apply them whenever a problem comes up will not only keep them fresh in your mind, but increase your problem solving aptitude – and that’s what strategic sourcing is fundamentally about.  When we are identifying, attacking, and removing any unnecessary costs and inefficiencies in the supply chain, we are identifying, attacking, and solving problems. 

Next week we will discuss some strategies to aid with the second step of the Operations Research Modeling Process: Problem Formulation.

The Operations Research Process

Operations Research is “the use of mathematical models, statistics and algorithms to aid in decision-making. It is most often used to analyze complex real-world systems, typically with the goal of improving or optimizing performance. It is one form of applied mathematics.” (Wikipedia)

The quest for improvement has been a continual one and operations research has been one of the areas that has been traditionally focused on improving operations across the company, particularly in production, operation, scheduling and physical systems.  As such, there is a wide body of knowledge upon which can draw to improve procurement and sourcing operations when properly applied and modified.  Even Six Sigma’s (Strategic Sourcing) toolbox makes extensive use of techniques and processes that have their foundations in operations research.

But today we are going to talk about the basic process.  There are many good overviews of operations research on the internet, but one of the best sites I found was the “Operations Research Models and Methods” site (formerly at http://www.me.utexas.edu/~jensen/ORMM/) that was maintained by Dr. Paul Jensen at the University of Texas.  On the site, he overviewed the basic Operations Research Process, which, simply put, is:

(1) Recognize the Problem
(2) Formulate the Problem
(3) Construct a Model
(4) Find a Solution
(5) Define the Process
(6) Implement the Solution
(7) Repeat and Refine

Essentially, the operations research process is your basic problem solving process, like the introductory problem solving process you might encounter if you were studying (cognitive) psychology, the art of mathematics, or (classical) engineering.  Furthermore, it neatly captures the key steps you will have to work through as you attempt to improve and evolutionize your sourcing process.

(1) You first have to define what your primary problem is and what your key goals are.  Are you spending too much money?  If so, where.  Are you spending too much time on the process?  If so, why?  Etc.

(2) Then you have to formulate and frame the key problem.  For example, you believe you’re spending too much on your high volume direct materials or you are spending too much time in your data collection process.

(3) Once you have precisely formulated the problem to solve, you need to model what you believe the solution should look like.  Many individuals and organizations skip this step and go straight to the solution identification step.  However, if you don’t know what the solution should look like, you risk selecting the wrong solution.  For example, if you believe you are spending too much, you might select a (reverse) e-Auction platform.  However, if current raw commodity prices are high, this might not save you any money.  Conversely, if you instead sought out a strategic supplier who would work with you to improve processes and component reusability, you might save a bundle.  Always understand what the solution should look like and what it should accomplish before selecting a solution.

(4) Often this step will be accomplished in practice by selecting a readily available technology, methodology, process, or model from the public domain or commercial marketplace.  Don’t try to reinvent the wheel, chances are your problem is not unique and someone else has already solved it for you.  For example, the inventor and followers of TRIZ (an innovative problem solving methodology that we will discuss at a later time) have collectively reviewed over 2 million patents and discovered that less then 4% contained a new concept and only 1% contained a revolutionary discovery.  The rest were merely improvements on existing solutions and processes.  In other words, there is at least a 95% chance that a solution to your problem already exists, and at least a 99% chance that a solution to a similar problem exists that can be adapted to your problem.

(5) Once you have a selected a solution — be it a technology or a new methodology, you need to define how it is going to be integrated into your current operational processes.  This step is easy to overlook, but if the introduction of a new process or technology disrupts your daily operations, you will not realize the full benefits.

(6) Once you have identified the right solution and determined how to successfully integrate it into your operational processes, you need to implement it and reap the rewards.

(7) Finally, you need to monitor the process, measure the improvements, and look for ways to continually improve it.  Innovation is a continual activity.  There’s always room for improvement, but if you do not look for it, I guarantee you will not find it.

We’ll discuss other problem solving methodologies in the future, including some from psychology, innovation, and the Six Sigma toolbox, but first we are going to review the core executable parts of the sourcing cycle where technology solutions can have the greatest impacts to set the stage for what is to come.

Comments? Criticisms? Creative Insights?  Email us at the contact information in the FAQ.