Sourcing Decision Optimization
Strategic Sourcing Decision Optimization
The Inefficiency Eliminator
In today's highly competitive marketplace, rampant inflation, diminished returns from reverse auctions, and constrained capacity could combine to significantly increase an organization’s Cost of Goods Sold unless advanced sourcing technologies and decision analytics are employed. This wiki defines decision optimization, describes the basic requirements of a decision-optimization based solution for sourcing, indicates how it supports total value management, and outlines its place in the sourcing process.
Decision optimization is important as it can significantly reduce cycle times, enable significant realizable cost savings, and guide a buyer to the best overall total value management sourcing decision. Sophisticated analysis that takes weeks or months in Excel can often be done in a few hours in a good decision optimization tool and Aberdeen has found that the application of optimization tools to analyze total costs, and of flexible bidding functionality to uncover creative supplier solutions has enabled early adopters to identify an average incremental savings of 12% above those that basic, price-focused auctions alone have generated.
It is important to realize that not all decision support or optimization tools meet the strict requirements provided herein for true strategic sourcing decision optimization, since many heuristic, simulation, and evolutionary methodologies are not guaranteed to produce the optimal answer on every problem, even given infinite run time and processing power by the nature of their design. (However, a decision support tool based on mixed integer linear programming or constraint programming techniques will usually qualify since these techniques, if properly applied, are usually guaranteed to produce the optimal answer if given sufficient time and processing resources.)
At a high level, decision optimization requires solid mathematical foundations, true cost modeling, sophisticated constraint analysis, and What If? scenario analysis capabilities. In other words, the solution must be based on an optimization technique that is sound and complete, support the definition of all associated costs and constraints, and allow for the construction, solution, and comparison of multiple What If? scenarios.
Decision optimization, defined as the application of rigorous mathematical techniques that are provably correct and repeatable given the same scenario definition, is a time-tested and proven methodology of arriving at the best answer for a problem - it is not a new craze and dates back to 1947 when G. Dantzig developed the simplex algorithm.
Finally, best results are obtained when best practices are used. This paper concludes by defining ten strategies to guide a buyer down the right path.
- Combine the art with the science
- Establish clear goals
- Look at the big picture
- Encourage supplier innovation
- Reward suppliers for quality and innovation
- Evaluate the full range of possibilities
- Engage all relevant stakeholders
- Apply analytics to the front end of the process
- Identify process improvements and
- Target 100% of spend - decision optimization is applicable to any sourcing problem, not just the large ones!
Why should one be concerned with strategic sourcing decision optimization?
As highlighted in a recent report from Aberdeen, four significant market forces are converging that, when combined, could significantly increase an organization’s Cost of Goods Sold unless advanced sourcing technologies and decision analytics are brought to bear on the problem. Hands down, there is no advanced sourcing technology that is more widely applicable than decision optimization.
Many commodity categories, especially energy related or influenced categories, are increasing steadily with inflation expected to continue for the near future. The days of constant deflation appear to be over. Thus, it is absolutely critical that the best buy be made every time.
Diminished Returns from Reverse Auctions
Although auctions have historically yielded double-digit returns in their initial application, successive applications quickly squeezed out all of the fat in supplier margins leaving little or no room for additional returns in repeated application. The low hanging fruit has been picked.
Furthermore, some poorly planned and executed auctions have resulted in increased prices in the long term as some suppliers, desperate for new business, bid at or below their total cost of goods sold hoping to make the loss up in the long run. They then ended up forfeiting on the contract or going out of business when they failed to make up the loss and ran out of money. This resulted in the buyer having to secure new sources at the last minute at significantly higher unit costs and transportation costs. Had the buyer used a decision optimization application which can enforce a risk mitigation strategy by way of appropriate constraints, the buyer would have been able to quickly switch to one of the alternate supply sources at minimal impact and minimal additional cost.
In many markets, demand is meeting or exceeding supply. This necessitates a heightened focus on supply assurance in the short and long term. Examples include raw materials used in consumer goods production, such as iron ore, and in numerous agricultural commodities, due to the effects of increased natural disasters such as hurricanes, typhoons, and earthquakes, in recent times. A decision optimization tool can take into account each supplier’s capacity restrictions and insure an award is not made that a supplier cannot support.
Renewed Focus on Top Line Value
With a renewed focus on top-line revenue performance now that the recent information technology boom and bust cycle has taught us that mindshare and market share alone do not a successful company make, executives are looking for ways to combat price deflation and maintain product and service innovation. A good decision optimization tool allows the absolute best decision to be made, avoiding unnecessary cost. Those dollars can then be diverted to new innovation efforts, multiplying the benefit.
A previous wiki defined strategic sourcing as the process of identifying, evaluating, negotiating, and configuring the optimal mix of products, suppliers, processes, and services to support business objectives from a total value management (TVM) perspective. In short – it is the process of selecting the right product, at the right time, from the right supplier, at the right price. However, this is much easier said then done.
That’s why one turns to sourcing optimization, the cornerstone of the strategic sourcing process, defined as the use of advanced analytical tools to simultaneously negotiate and evaluate complex sourcing scenarios and bid structures against a wide range of interdependent sourcing objectives, variables, and constraints.
Strategic Sourcing is a process-based methodology that makes use of advanced technologies to help an organization make better buying decisions. For more information, see the companion wiki Strategic e-Sourcing Best Practices.
An Introduction to Decision Optimization
Before strategic sourcing decision optimization can be defined, the capabilities of a strategic (e-)sourcing decision optimization tool described, the benefits that it provides explained, or some basic strategies for success outlined, one needs to understand what decision optimization is.
A Basic Definition
As with many emerging technologies, decision optimization means different things to different people and every individual asked could quite possibly provide a different answer. However, almost all of the good definitions have two things in common. They all focus on the derivation of the absolute best decision and they all refer to the use of a rigorous algorithmic technique to arrive at that decision. Therefore, decision optimization can be defined as follows:
Note that our definition has four key components.
(1) Rigorous Analytical Techniques
The techniques must be mathematically sound and complete. Therefore, the application of Mixed Integer Linear Programming techniques would be valid, since hybrid simplex approaches will provably converge on an optimal answer given sufficient time, but the application of many heuristic, simulation, or evolutionary approaches may not be, since many of these techniques do not guarantee full exploration of the potential solution space.
In laymen terms, the techniques applied to the problem must be capable of analyzing every possible solution (complete) and do so in a correct fashion (sound).
(2) Well Defined Scenario
The scenario must completely and accurately represent the problem at hand. For example, if a buyer is sourcing a direct material, it must include all indirect and incurred costs, such as freight, tariff, storage, processing, and marketing differential costs, all regulatory constraints on volumes and compliance, and all business constraints dictated by organizational policies and supply chain strategies.
(3) Best Decision
The process must be capable of producing the absolute best decision given a sufficiently large finite amount of time. It is not sufficient that the process simply yield a better solution than could be obtained with an e-auction, revised sourcing strategy, or strategic endeavor. The process must be capable of producing the best answer.
(The key word here is capable because some problems are so large and so complex that it might take the best software tools available on the fastest computers hours, days, or even weeks to determine the absolute optimal answer and prove it. Moreover, in reality, a solution that is within 0.001% of optimal with a provably high probability will usually be sufficient, especially when one is dealing with a spend under 100M. If such a solution was significantly better than the best solution producible without the tool, and such a solution could be computed in minutes, when it would take days or weeks of computation to prove it optimal and / or find the optimal solution, in practice one often stops the analysis at this point since 1K is insignificant in a 100M spend.)
(4) Repeatable and Provable
A re-application of the process must produce the same solution or another solution that is equivalent with respect to the objective of maximizing the total value of the award. Furthermore, the underlying mathematical algorithms must be provably correct.
What It is -and- What It Is Not
As per our definition, decision optimization is the application of rigorous mathematical techniques that are provably correct and repeatable given the same scenario definition. It is not the application of just any advanced analytical technique to the problem. The analytical technique must be capable of analyzing every possible solution to the scenario and be capable of performing the analysis accurately.
This means that sourcing award allocation tools based on mixed integer linear programming are valid decision optimization tools since they are based on simplex algorithms and branch and bound techniques known to be sound and complete. [The reason for this is that, with every step, the simplex algorithm mathematically proves that a certain potential set of awards must be worse than the award currently being evaluated, limiting the number of awards that remain to be analyzed in the next step. Thus, the algorithm is guaranteed to terminate after a finite number of steps at the best solution.]
This also means that sourcing award allocation tools based on Monte Carlo simulations are not true decision optimization tools. Although Monte Carlo simulations analyze a large number of possible assignments according to statistical distributions designed to cover the maximum number of the most likely possibilities, the selection of the possible assignments analyzed is not guaranteed to contain the optimal solution. Thus, although such tools may be capable of providing an answer that is probably optimal after a sufficiently large number of runs, it is not provably optimal, and these tools fall into the category of decision support, and not decision optimization.
Decision optimization is an advanced analytical tool that can be applied to a buyer’s toughest scenarios to achieve the correct answers under correct conditions. It is not intelligent.
This may seem a silly thing to state, but even in today’s age many people think that IT is a replacement for manpower. Although IT can automate many tasks for us and reduce the amount of human effort required, sometimes substantially, it does not take the human out of the picture. This is especially true in sourcing.
The buyer has to decide what to source, when to source, what strategy to apply, and why before decision optimization can even be thought about. Then the buyer has to define the scenario, insure all the data is present and correct, make sure all the constraints are present, and analyze the alternatives against the hard data and soft targets.
Now, one might say that an inventory tracking system can tell determine when an organization is low on stock, a forecasting system can predict when an organization will likely run out of inventory and how many additional units will be needed in the following year, an expert system can determine what strategy is statistically likely to give the best results, an RSS feed can push updated pricing data into the sourcing suite, a rules engine can extract the appropriate rules, and a decision support system can define all the scenarios that should be optimized and compared and then select the best one according to a learned heuristic. And one would be right. But this would not take into account that a drought just wiped out 20% of the world’s supply of the produce item that was about to be sourced, in a market where demand typically exceeds supply by 5%, that a supplier might be willing to offer a discount if the organization increases the purchase volume of another item, that a viral marketing campaign could cause a 50% spike in demand in three months time, or that there is the expectation that new governmental regulations will come into effect mid way through the contract term and the organization should be insuring that an award that would comply is selected. In other words, the amount of manpower required can be significantly reduced, which decision optimization does, but it does not take the human out of the picture. However, with a decision optimization tool, the amount of spend a human can analyze essentially increases exponentially.
Given a complete scenario, a decision optimization tool will find the optimal answer. It can not build the proper scenario. It significantly increases the productivity of a skilled sourcing professional – it does not replace the need for such an individual at the helm.
Decision optimization is a time-tested and proven methodology of arriving at the best answer for a problem. It is not a new craze.
The formal beginnings of optimization trace back to 1947 when G. Dantzig developed the simplex algorithm: the first general methodology for finding the optimal solution of a general linear program. Optimization turns 60 in 2007 (when this wiki was first started). It works. The development of appropriate sourcing models may be a recent event in optimization’s history, but the mathematical underpinnings are solid.
Decision optimization is applicable to any sourcing problem!
One myth that is found over and over again in the literature is that decision optimization is only for “large problems” or “complex categories” or “strategic spend”. That the effort involved is not worth it except in these cases.
Well, to put it bluntly, this is the biggest piece of BS the (e-)sourcing solution providers have ever released upon the unsuspecting sourcing community.
First of all, if the e-sourcing product suite being used complements a proper sourcing process, then a buyer will have identified the sourcing strategy, identified and qualified the suppliers, documented an accurately forecasted demand, collected all of the constraints, attributes, and weightings, and collected all of the bids – in other words, everything required to define a proper sourcing scenario. It should simply be a matter of identifying all of this relevant data to the decision optimizer – which is a minimal effort compared to all of the effort that went into collecting the information, verifying its accuracy, and making the decisions that took the buyer to this point.
Moreover, if the tool is properly constructed and easy to use, the constraints would have been categorized as regulatory requirement, scenario requirement, corporate policy, and sourcing strategy, and the tool will automatically be able to construct and solve an unconstrained, a requirement constrained, a set of individually constrained scenarios for each constraint, and a fully constrained scenario that will not only present an optimal award strategy but allow the buyer to understand approximately how much each strategy constraint and policy constraint is costing the organization.
Therefore, not only should decision optimization be rather effortless compared to the other steps of the process, but it should be capable of providing more automation and user support than the other parts of the sourcing process.
Secondly, a problem does not have to be very large to be complex. Believe it or not, one item, three suppliers, six shipping locations, three discounts, and three potential carriers per lane is complex. This alone amounts to at least 36 different costs for the same item. Even with a spreadsheet, a buyer is going to spend at least an hour creating and analyzing all the possibilities in an error prone process. An optimizer will take less than a second.
Thirdly, there is no drawback. What’s the absolute worst result obtainable from the application of decision optimization? A proof that the solution arrived at without using the tool is optimal and that it is not possible to save a single cent. How bad is this? Not bad at all. Actual proof that out of the myriad of possibilities, often in the thousands or millions, the best one has been chosen. The odds have been beaten – with certainty!
However, since this practically never happens (neither Iasta nor the primary author of this wiki has ever encountered or even heard of a single real world sourcing project where the current solution was the best solution or where an analyst found the best solution without a decision optimization tool), the buyer is guaranteed results. If there is no possibility for loss, why not use it?
So, unless you’re a small organization where your annual spend is less than a few million, since the expected savings would not exceed the licensing costs of the tool significantly enough to make licensing the tool a good business decision, there’s really no excuse not to be using strategic sourcing decision optimization.
Decision Optimization and Total Value Management
Total Value Management is a comparative cost metric that quantifies a sourcing plan according to the overall cost of each acquired unit of product relative to the overall value of the buy subject to the sourcing strategy and supply chain goals of the organization.
It is more than just total cost of ownership. Although total cost of ownership sounds like the right metric to use for strategic sourcing as it captures all direct costs, indirect costs, and in a proper implementation, all quantifiable market costs, it often misses the impact costs from deviating from overall sourcing and supply chain strategies – wherein the true value of strategic sourcing lies. The harsh reality is that a 5% savings on TCO for a specific sourcing project could actually yield a 10% loss if it sacrifices a future savings of up to 5% per year for the next three years which would have resulted from selecting a different supplier willing to undergo production process renovations to improve their efficiency and lower their costs in the long run if given a multi-year commitment.
Total Value Management is the natural extension of Total Cost of Ownership augmented to align with the corporate sourcing and supply chain strategies and to take into account the current and future costs associated with deviating from the overall supply chain strategy.
Decision Optimization in Strategic Sourcing is the application of rigorous mathematical techniques to produce optimal award allocations from a total value management perspective. The scenarios it works on are Total Value Management Scenarios.
Optimization-Based Decision Support
As was stated in the last section, the primary difference between a decision support system and a decision optimization system is the presence of rigorous analytical techniques that are repeatable and provably accurate. However, the existence of solid mathematical underpinnings alone is not enough – the system must be able to model and solve an accurate representation of the real-world sourcing problem. Even the best optimization algorithm in the world is useless if it does not solve the problem.
The system must contain the capabilities required to accurately represent the real world sourcing scenario and must be able to translate that representation into an accurate low level model representation compatible with the underlying algorithms. However, this requires more then just capturing basic cost data, forecasted demands, and capacity constraints. Although this may be enough information to form a minimally valid model, it is an oversimplification of the real world scenario that does not capture policy or strategy constraints or allow for the formation of a more accurate supply chain picture.
Before basic requirements for an optimization-based decision support system for strategic sourcing are specified in detail, it should be pointed out why there is a need to go to such great lengths to do so. In recent times, a growing number of solution providers have claimed to possess ‘optimized’ applications that achieve the goal of producing the absolute best solution for a scenario – every time, when in fact this is not the case. One only has to look to the media to discover a number of situations where many solution providers, including internationally renowned software companies, have failed to deliver on this promise.
Even today, after an extensive literature survey, one will quickly realize that the majority of providers claiming to offer decision optimization do not define precisely what they mean by optimization or describe exactly how their products enable an enhanced degree of value, savings, or productivity. (There are a few notable exceptions, but they are few and far between and generally countable on a single buyer’s fingers.) This leads us to the critical question: Do the majority e-sourcing solution providers really deliver true decision optimization or are they just trying to jump on the bandwagon?
Therefore, it is vital to explicitly define what a decision optimization system is. It is a system that meets four basic requirements: solid mathematical foundations, true cost modeling, sophisticated constraint support, and what if? capability.
Solid Mathematical Foundations
The tool must be based on a sound (correct) and complete algorithm (capable of analyzing every possibility, not just statistically relevant or likely possibilities) that analyzes an accurate representation of the problem and not a (heuristic) simplification thereof. Generally speaking, a decision support tool built on mixed integer linear programming or constraint programming techniques will meet these rigid requirements and provide true decision optimization while tools based on heuristic techniques, evolutionary methodologies, or simulation models will generally not meet these requirements. This is because many mixed integer and constraint based techniques provably evaluate, directly or indirectly, every possibility while evolutionary and simulation techniques only evaluate a finite number of (what they believe to be the most statistically relevant) possibilities.
Tools based on non-linear techniques, gradient search, and tabu search may or may not be capable of producing the optimal solution (every time) and fall into a grey area between decision optimization and decision support. It depends on the specific implementation of the algorithms used.
Non-linear problems are much harder then (piecewise) linear problems and it is generally not possible to prove a solution optimal beyond a limited degree of accuracy or confidence. If the algorithm used would cover a sufficient amount of the search space to reach a specified degree of accuracy and confidence given a sufficiently large finite amount of time, then it is a true decision optimization algorithm. If it is based on statistics, probabilities, or potentially incomplete meta heuristics and can not be guaranteed to reach any fixed level of confidence in a finite timeframe, then it is just a highly sophisticated decision support system.
This paper is not dismissing the utility of systems based on mathematical techniques that are being classified as decision support as many of these techniques, when fine-tuned, can produce near-optimal solutions with sufficiently high confidence in relatively short time frames for the majority of real world scenarios the system processes. If one has a sufficiently complex scenario that would take the best decision optimization system weeks or months of computation to arrive at and prove an optimal solution and one of these decision support systems could arrive at a solution that was expected to be within 0.1% of that solution in days or hours, and the spend was under 10M, the solution would produced by the decision support system would be more then adequate for general use and much more timely. Thus, when properly used, such systems do have their place.
However, even though this wiki-paper is not attempting to dismiss the utility of decision support systems based on potentially incomplete mathematical techniques in strategic sourcing, this paper is cautioning users on the use of these decision support systems. The reality is that even though such a system might produce a workable ‘best’ answer 99 times, there is no guarantee that the answer produced the 100th time will be a workable ‘best’ answer. These systems have their weak points, and every now and again, by the very laws of statistics they are often based on, they will produce a ‘bad’ answer.
Therefore, if these decision support systems are used (and there may often be good reasons to use these systems from a performance or system cost of ownership perspective), one should always validate the answer(s) produced through a true decision optimization system. If the decision optimization system is not capable of producing a better answer in an equivalent timeframe, one can usually assume with a sufficiently high level of confidence that the solution produce is optimal to within a small degree of accuracy. However, if the decision optimization system produces a better answer in the same time frame, then the buyer knows that she has hit one of the statistically mandated ‘bad’ answers and that the buyer should allow the decision optimization system to run to conclusion and accept its answer.
[In fact, an experienced optimization expert, in all honesty, will know that if a user is serious about high performance optimization, then the user should consider simultaneously running multiple implementations of the same basic algorithms, and that the user might even consider purchasing multiple similar products and running them in parallel. Even though all major commercial implementations of mixed integer linear program solvers are complete, they all use heuristics to order the evaluations of the decision points to try and find a faster path through the search space or a shorter proof that the current solution is optimal. A bad choice early on in the selection of a decision point can exponentially increase the solve time and a good choice early on can exponentially decrease it. This is why some solvers start by running a number of algorithmic variations for a finite number of steps internally before settling on one particular algorithm.]
True Cost Modeling
Despite a plethora of systems that purport to be examples to the contrary, simply capturing a bid on one unit of an item is not sufficient to qualify a system as a strategic sourcing decision optimization system. In reality, the cost of goods sold calculation is much more complex – and so are the bids provided by many suppliers, although not necessarily for the same reasons.
The reality is that even with the simplest of products, there are fixed costs and variable costs and that these change at fixed production levels. For example, there are fixed costs to set up and start a production line and then variable costs for each production level depending on raw resources, energy, and manpower required to produce the product. A true decision optimization system for strategic sourcing should be able to capture and analyze the fixed and variable costs at each tier under consideration.
Furthermore, in addition to the direct purchase cost, there are indirect transportation costs, storage costs, duties, tariffs, and utilization costs as well as impact costs to associate with an award from marketing, customer satisfaction, and other total value management viewpoints. The system must also be able to capture all of these associated costs, when relevant.
Some suppliers will offer discounts on bulk purchases, especially when multiple items are combined in a single purchase. In addition, there may be pre-existing multi-period agreements in place with some suppliers or business requirements that would impose a penalty if a certain supplier and / or product is not selected as part of the total award. Therefore, the system must also support discounts and penalties on a supplier and / or item basis.
Sophisticated Constraint Analysis
Even though sophisticated e-auction systems with complex transformation bidding support may lead one to think otherwise, the lowest bid is not the optimal solution. In reality, there a plethora of regulatory, business, and strategic constraints of a quantitative and qualitative variety that a buyer needs to adhere to in order to arrive at a true optimal solution that does not have unintended consequences.
Regulatory constraints will often limit where an organization can source from, what requirements the suppliers have to meet, and what limits have to be adhered to. Furthermore, they may also define incentives that an organization wants to take advantage of. For example, there may be a trade embargo on Palestine, the suppliers from Pakistan might need to adhere to increased security regulations, and the organization might be forced to source a minimum amount from homeland suppliers. In addition, there might be government incentives or subsidies for the organization to do business with certain minority suppliers.
Business constraints will often define pre-existing agreements that need to be adhered to, restrictions on what products and what quantities thereof can be shipped to, processed, or stored at buyer facilities, preferred carriers, and qualitative requirements that must be met on the products being sourced from an engineering perspective.
Strategic constraints will correspond to a sourcing strategy, as defined by the category requirements and the organization’s overall supply chain strategy, and will define supply assurance and risk mitigation requirements, qualitative objectives based on marketing and customer satisfaction and service goals, and quantitative restrictions on certain suppliers.
This tells us that, at a minimum, a strategic sourcing decision optimization tool should support at least the following four constraint types: capacity / limit, basic allocation, risk mitigation (meta) allocation, and qualitative.
Capacity Constraints allow a buyer to specify real world limits on the amount a supplier can supply and on the amount a warehouse can receive. Limit constraints allow a buyer to further restrict supply based on business rules or strategic decisions. General limit constraints allow a buyer to implement the regulatory constraints that restrict who the organization can source from.
Basic Allocation constraints allow the buyer to select a supplier, or set of suppliers, and specify that a certain award percentage of one or more items must be allocated to this supplier. Basic allocation constraints allow a buyer to capture pre-existing agreements and implement business rules with respect to preferred suppliers or carriers as well as any preferred allocations defined by the chosen sourcing strategy.
Risk Mitigation Allocation constraints allow for the selection of a group of suppliers and the indication that one or more members of this group must receive a minimum or maximum award allocation. Risk mitigation constraints allow a buyer to implement supply assurance strategic constraints and, when combined with discounts, regulatory incentives.
Qualitative constraints allow for the imposition of an absolute or average minimum or maximum qualitative score on each product or product bundle sourced. Qualitative constraints allow a buyer to define engineering requirements (minimum durability rating of 7) as well as marketing and customer satisfaction goals (expected defect rate of 5% or less).
What If? Capability
True Cost Modeling and Sophisticated Constraint Analysis capabilities are indeed extremely powerful tools, especially when compared to any system that does not have these, but the true power of a strategic sourcing decision optimization tool lies in the ability to generate, analyze, and compare “What If?” scenarios.
What if actual demand is 10% greater than forecast? What if actual demand is 10% less than forecast? What if 40% could be sourced from LCC (Low Cost Country) suppliers instead of 20%? What if the incumbent supplier could reduce costs by 5%? What if the organization’s preferred supplier can not meet demand?
Being able to determine the absolute best decision for a sourcing scenario from a total value management perspective is one thing, being able to truly understand the reasons for, and the costs associated with each constraint and decision, is something else entirely.
Therefore, a true strategic sourcing decision optimization tool should support multiple what if? variant scenarios on a base scenario and even facilitate the (semi-automatic) creation of those scenarios. It should also support the quick generation of clear comparison reports that allow a buyer to compare and contrast the costs of unconstrained, minimally constrained (from a regulatory or capacity perspective), partially constrained (from a strategy perspective), fully constrained, and what-if constrained scenarios to understand what each constraint is costing the organization and what each assumption (demand forecast, supply forecast, etc.) could cost the organization if it turned out to be wrong.
Sophisticated technology on its own does not guarantee success. It was said before and it will be said again. In order to be guaranteed of success in one’s endeavors, not only does one need a decision optimization system that meets the basic requirements described in the last section, but there is a need to use it properly. This involves not only following the best practices applicable to the spend category, but also ensuring that the forecasts are accurate, best practices category templates are at the foundation of the scenario, and appropriate what if? variant scenarios for accurate constraint costing are used.
An old proverb attributed to Confucius says “To go beyond is as wrong as to fall short.” That’s as true today as the time it was first uttered. In fact, programmers have an even more general saying that is essentially equivalent from a results viewpoint “Garbage In Garbage Out”. An optimal answer is only optimal if it is solving the right problem.
Even small perturbations in volume could lead to noticeably large differences in an optimal allocation strategy when price tiers, discounts, and bundles from multiple suppliers are at play. Therefore, it is important that a buyer uses the most accurate data available. Selecting a supplier on the premise that the organization will order enough units to qualify for a significant end-of-year rebate is only optimal if the organization hits the target volume – otherwise, the next cheapest supplier without a rebate could be the best choice.
Best Practices Category Templates
A basic requirement of total value management is the capture of the true cost of ownership of each item being sourced. For many items, this will go well beyond unit costs and involve variable transportation costs, duties, tariffs, utilization, storage, and impact costs which will vary by supplier – especially when dealing with agricultural commodities and derivatives. Consider tomato paste – an essential ingredient in the construction of one of America’s favorite food – pizza. Is it canned or bottled? Can it be transported at room temperature or does it require refrigeration?
In order to insure that a buyer does not miss any indirect, incurred, or impact costs and that each cost is defined appropriately, a buyer should start with a category template that includes all of the costs that must be captured, all of the factors that need to be considered in cost identification, and all of the regulatory and supply chain constraints that need to be considered. This will help guarantee that the model constructed will be complete and accurate.
Proper Place in the Strategic Sourcing Process
Section 2.2 alluded to the proper place for decision optimization in the strategic sourcing process when it was noted that if the e-sourcing product suite that is being used complements a proper sourcing process, then the buyer will have identified the sourcing strategy, identified and qualified the suppliers, documented an accurately forecasted demand, collected all of the constraints, attributes, and weightings, and collected all of the bids before making a sourcing decision.
Decision Optimization is the fifth step in the generic sourcing process that was defined in “Strategic e-Sourcing Best Practices” and the final step undertaken before making an award decision. This is because it is a tool designed to help us select the best possibility out of a plethora of options, not a tool designed to identify what the plethora of options available are. That’s where spend analytics, score cards, best practices, supplier qualification tools, demand forecasting, e-RFx, e-Auction tools, and trained sourcing professionals with in-depth category and market knowledge come into play.
Although it is arguably the most powerful tool in a sourcing professional’s toolkit, it is still only a tool nonetheless and not a substitute for proper process or an experienced sourcing professional.
Furthermore, as will be elaborated on in section five when the top ten strategies for success are discussed, it is most effective when the sourcing process is carried out during the initial product design phase before up to 80% (or more) of the product’s costs are locked in.
One area where the marketing materials and available literature does not fall short is when it extols the many benefits and virtues of decision optimization which are achievable when the technology is sound, complete, and properly applied.
Significantly Reduced Cycle Times
There are dozens, if not hundreds of case studies, available from all of the major and minor vendors and analyst firms extolling the huge impact that e-sourcing suites have had on cycle times, typically reducing the average cycle time from 3-4+ months to 3-4 weeks. However, few of these reports focus on the cycle time savings achieved by decision optimization alone. In a complex category, decision optimization will trim weeks, and sometimes months, of analysis down to a few days.
Before decision optimization technology, a buying group had to manually construct a number of award scenarios that they believed would satisfy all of the constraints and then analyze each one from a compliance and cost perspective, and repeat this process over and over until they arrived at the best solution they could find in the time available. With (true) decision optimization, once a model has been populated and fully defined, a buyer just needs to push one button … ‘solve’ … and the optimal answer is computed, often within minutes. The buyer can then tell the tool to solve “what if” scenarios without each of the strategic constraints to determine if the constraints or business rules are worth the cost and generally within a few hours a buyer can have confidence that the selected award decision is right for the organization. If confidence is lacking in the demand forecast, a few scenarios at increased or decreased demands can be run to select an award allocation that is expected to be near-optimal across multiple allocation volumes within a few more hours.
Even when dealing with the most complicated category, a buyer can often settle on a near-optimal award with confidence in a day or two whereas in the past a buyer might not have identified a reasonable solution even with weeks of careful analysis.
Significant Realizable Cost Savings
Although the goal of this wiki, unlike some white papers, is not to overburden the reader with statistics, there is one statistic in particular from Aberdeen’s 2005 Report on “Success Strategies in Advanced Sourcing and Negotiations: Optimizing Total Costs and Total Value for the Next Wave of E-Sourcing Savings” that needs to be restated:
“Early adopters of advanced sourcing and negotiations reported incremental savings of 12% on average, beyond what was obtained with basic e-RFx and auction tools alone.”
And one statistic in particular from Aberdeen’s 2007 follow up report on “The Advanced Sourcing and Negotiation Benchmark Report: The Art and Science of the Deal” that needs to be restated:
“Enterprises that are employing advanced sourcing techniques are still identifying an average savings of 11.9% per sourcing event. Furthermore, best-in-class enterprises are identifying an average savings of 13.7% per event.”
Double digit savings on average, 100% realizable because the award allocation was generated based on complete cost information that also took into account all of the relevant real world constraints. Furthermore, the 2007 Aberdeen report found that organizations not employing advanced sourcing techniques, such as optimization, are no longer obtaining double digit savings on average.
Best Overall TVM Sourcing Decisions
The right buy at the right price every time. No other technology can guarantee that!
Unlike e-auctions, which are only of benefit until all of the fat has been squeezed out of a supplier’s margin, and which generally only net results on direct cost components for a limited number of applications, decision optimization will net results every time as it takes into account all of the continually changing costs, demands, and constraints.
Ten Strategies for Success
In their 2005 Report on “Success Strategies in Advanced Sourcing and Negotiations: Optimizing Total Costs and Total Value for the Next Wave of E-Sourcing Savings”, Aberdeen listed eight strategies which they believed represented the best practices of leading edge companies that used advanced sourcing toolkits. This wiki goes one step further and identifies ten strategies that deliver results when deployed with a sourcing process that integrates decision optimization.
Combine the Art with the Science
Make sure the sourcing solution selected includes a true decision optimization tool that supports cost models, supply market knowledge, and all the basic constraint categories (the art) in addition to using advanced analytical technology built on solid mathematical foundations (the science).
Two of the biggest barriers to success in any sourcing project, as described in the companion wiki “Strategic e-Sourcing Best Practices” are lack of focus and lack of clarity. Just like an ad-hoc on-the-side project will not be successful, neither will a project that just broadly aims to increase value or reduce spend with decision optimization technology. There is a need to fully answer the what, why, when, where, who, and how questions before a valid scenario that will yield an optimal answer can be built. For example, What needs to be strategically sourced? What is the strategy? Why? When is it needed? Where can it be found? Who supplies it? How can it be transported?
Look at the Big Picture
A sourcing project can save 10% but still be a dismal failure due to unintended consequences. For example, if the suppliers were not properly qualified and a new low cost supplier was selected that was not capable of palette transport while all of the other suppliers were, and this supplier supplied the item in question to each of the organization’s major warehouses, this would likely incur additional personnel costs at each of the warehouses and this would more than cancel out the perceived savings of the lower cost supplier.
A strategic sourcing project should always be undertaken with respect to an overall supply chain plan which looks at all of the relevant issues. Such project should take into account any volatility in the market due to capacity or price and revise forecasts and cost estimates as late in the decision process as possible.
Encourage Supplier Innovation Through Flexible Bidding
A single price per unit bid does not paint a very accurate picture of a supplier’s cost of goods sold or identify any potential opportunities for cost savings. On the other hand, a tiered bid structure that incorporates fixed costs, variable costs, and discount opportunities that accurately represents the supplier’s true costs allows for the determination of a plethora of opportunities for immediate and future savings.
With a single unit bid, a supplier would be likely to play it safe and provide his best price at his worst production level. With complete cost information, it is possible to determine when a production line is being fully utilized, when the best price can be obtained, and what opportunities there are for joint ventures to reduce significant cost components.
Reward Suppliers for Quality and Innovation
Every product purchased has impact costs. Poor quality components lead to decreased customer satisfaction and lower sales figures. Components with a high defect rate lead to large numbers of returns which have associated processing costs. On the other hand, high quality components decrease customer service costs and contribute to increased sales. Furthermore, innovative companies have great marketing value.
It is important to take into account these qualitative factors when analyzing bids. Sometimes the most expensive product is the cheapest overall. In a bid-oriented model, be sure to reward high quality and innovative suppliers with negative impact costs. Not only is it the right thing to do, but it will lead to the right allocation in the long run.
Evaluate the Full Range of Possibilities: Unconstrained and Constrained
Do not settle for the best answer given the starting constraints. Find out how much each constraint is costing the organization and if it’s worth it with respect to the sourcing and supply chain strategies that are currently in force. If it’s not, either change the strategy or get permission to make an exception.
Engage all Relevant Stakeholders
Organizations that excel in strategic sourcing have adopted, developed, and enforced best-in-class strategic sourcing procedures across every department in the organization consistent with an overarching “destination” supply chain design.
Not only can engineering and product development help to adequately define the requirements for the product that is being sourced, but they can help with the assessment of supplier offerings. Marketing can advise with respect to whether or not any suppliers are beneficial from an image or partnership standpoint. Customer service can advise of any expected support requirements or costs. Engaging all of the right parties insures that the right information is available upon which to base a decision.
Apply Analytics to the Front End of the Process
Considering that as much as 80% of the final product costs are locked in during initial product design, it is vital that the sourcing process start here and include decision optimization.
Identify Process Improvements
With the increased visibility into a supplier’s cost of goods sold and complete total cost of ownership cost models, one can use decision optimization to identify areas where process improvement could make a significant impact. For example, select a significant cost category and run decision optimization to determine what the net effect would be if costs were reduced by 3%. Consider locking in 100% capacity utilization on a production line as a means of obtaining additional discounts from a supplier. Evaluate more efficient transportation options.
Target 100% of Spend
As was stated in section 2.2, Decision optimization is applicable to any sourcing problem – not just the large or complex ones.
Statements that decision optimization is only for “large problems” or “complex categories” or “strategic spend” are noting more then a myth. The effort involved, which is minimal with a well integrated sourcing suite, is always worth it.
A Selected Bibliography
(The) 12 Days of X-emplification Day 3 - Optimization by Michael Lamoureux, December 15, 2007
(The) 12 Days of X-emplification Day 11 - Supply Chain Optimization by Michael Lamoureux, December 23, 2007
(The) 6 Days of X-asperation Day 4 - Questions to Ask Your Decision Optimization Vendor by Michael Lamoureux, February 6, 2008
(The) Advanced Sourcing and Negotiation Benchmark Report: The Art and Science of the Deal by Andrew Bartolini & Vance Checketts of Aberdeen Group, 2007
Analytics vs. Optimization by Michael Lamoureux, March 27, 2007
Bid Optimization - Case Studies by David Bush, February 13, 2007
CAPS on Sourcing Optimization by David Bush, May 10, 2007
Changing the Game in Strategic Sourcing at Proctor & Gamble: Expressive Competition Enabled by Optimization by Tuomas Sandhom, David Levine, Michael Concordia, Paul Martyn, Rick Hughes, Jim Jacobs, & Dennis Begg, January-February 2006
CombineNet Communiqué I: The Story to Date by Michael Lamoureux, November 1, 2006
CombineNet Communiqué II: Comparisons by Michael Lamoureux, November 2, 2006
CombineNet Communiqué III: Differences by Michael Lamoureux, November 3, 2006
CombineNet Communiqué IV: BoB's Unique Talents by Michael Lamoureux, November 10, 2006
CombineNet Communiqué V: Expressive Bidding by Michael Lamoureux, January 15, 2007
CombineNet Communiqué VI: Strategic Sourcing Decision Optimization by Michael Lamoureux, January 16, 2007
CombineNet Communiqué VII: BoB's Power Source by Michael Lamoureux, January 17, 2007
CombineNet Communiqué VIII: CombineNet Energy by Michael Lamoureux, April 1, 2007
CombineNet Communiqué IX: Interlude - We Can Optimize Anything by Michael Lamoureux, May 25, 2007
Decision Optimization Defined by Michael Lamoureux, June 15, 2006
Embracing Complexity by Michael Lamoureux, February 25, 2007
Landed Cost Optimization (LCO): The Next Wave in Supply Chain Cost Reduction by Infosys, January 2005
Minimizing Procurement Costs for Strategic Sourcing by Jayant Kalgnanam & Ho Soo Lee, September 2001
Optimization I: A Powerful Tool by Michael Lamoureux, August 25, 2006
Optimization II: Why it was Relegated to the Shadows by Michael Lamoureux, August 26, 2006
Optimization III: Why it’s time is finally here by Michael Lamoureux, August 27, 2007
Optimization IV: POE or BoB? by Michael Lamoureux, August 28, 2007
Optimization Based Procurement for Transportation Services by Chris Caplice & Yossi Sheffi, 2003
Optimization is the Future ... And the Future is Now by Michael Lamoureux, October 8, 2007
Optimization Moves into the Sourcing Mainstream by David Bush, March 9, 2006
Optimizing Your Sourcing Decisions by David Bush, September 19, 2007
Questions to Ask Your Optimization Vendor by Michael Lamoureux, October 16, 2007
Science Based Optimization Engines Drive Next-Generation Retail Application Suites by Paula Rosenblum, January 2004
Seven Risk Mitigation Strategies You Can Do With Smart Optimization by Michael Lamoureux, February 6, 2008
the doctor Goes Mental on Optimization Myths by Michael Lamoureux, December 10, 2007
What's Involved in SCNO (Supply Chain Network Optimization)? Part I by Michael Lamoureux, June 27, 2007
What's Involved in SCNO (Supply Chain Network Optimization)? Part II by Michael Lamoureux, June 28, 2007
What's Involved in SCNO (Supply Chain Network Optimization)? Part III by Michael Lamoureux, June 29, 2007
What is Supply Chain Optimization? - Part I by Michael Lamoureux & Charles Dominick, July 24, 2007
What is Supply Chain Optimization? - Part II by Michael Lamoureux & Charles Dominick, July 24, 2007
Michael G. Lamoureux, Ph.D. of Sourcing Innovation