Sourcing optimization is a process that uses a combination of mathematical models, computer software and computer hardware to analyze large numbers of supplier bids and business scenarios to identify "optimal" sourcing decisions. This powerful combination of mathematics and computer capability gives suppliers wide freedom to express their bidding preferences in response to a request for quotes and gives buyers the ability to analyze these disparate bids. The process also has been called "bid optimization," "decision guidance" and "expressive bidding." In this report we will use the term "optimization" to refer to this process.

Complexity in sourcing events is created from a combination of a large number of: suppliers/bidders, bids, bids on different parts of the total spend (bundled bids), tiered bids (volume discounts), specifications, business rules (scenarios or constraints), and locations to be served.

The concept of optimization is not new, but historically took a great deal of people power and time and required large mainframe computers. In the last two decades hardware and software advances have made this application available to more companies. The technology varies by provider, but in general it is based on specialized software that solves linear and non-linear programming models using extremely fast computers with lots of memory. By using these powerful tools, some large problems can be solved in less than one second, but a solution time of about 20 seconds is more common. However, extremely large, complex problems still may require a day of computing time to find a solution.

Early applications of sourcing optimization were concerned with buying transportation services for national and international shipping routes, which included all of the complexities identified above. Many suppliers were available, none of which could provide shipping for all modes and all routes. Also many shipping lanes needed to be covered, and the lanes could change from year to year. Furthermore, the type of transportation needed varied, including full-load, less-than-truckload, multi-modal, wide load and refrigerated.

For example, at one company, 150 suppliers bid on more than 7,000 shipping lanes. The number of possible bid combinations was in the millions, and the actual bids received numbered in the hundreds of thousands, which made evaluation of the alternatives a daunting task.

Using spreadsheets on desktop computers, it would be impossible to completely evaluate all of the sourcing alternatives and find the best combination of bids and suppliers. But optimization was used to help the sourcing team to quickly evaluate thousands of alternatives and identify sourcing solutions that offered the lowest cost based on the bids submitted.

Optimization Software Providers

Optimization services are offered by a relatively small number of providers for general purchasing applications. Providers include:

• Ariba /Procuri
• CombineNet
• Emptoris
• Iasta
• Manhattan Associates
• Perfect Commerce

For this study, executives from Ariba/Procuri, CombineNet, Emptoris, and Iasta were interviewed.

CombineNet offers a high-end, stand-alone solution that can handle the most complex sourcing problems. Manhattan Associates specializes in transportation sourcing. Dallas-based i2 specializes in factory scheduling and supply chain design, while offering a module for sourcing optimization. Iasta, Ariba/Procuri, Emptoris and Perfect Commerce specialize in purchasing/sourcing applications and offer optimization as part of their sophisticated suite of e-tools. In addition to optimization, the suites generally include spend management, RFx, reverse auctions, purchasing, e-catalogs and contract management functionality.

Applicability in Sourcing

Today, optimization is being applied to an increasing number of commodities and services, including such categories as:

• bulk fuel
• chemicals
• containers
• displays
• fasteners
• food ingredients
• hospital supplies
• janitorial services
• legal services
• lighting fixtures
• maintenance services
• MRO of many types
• natural gas
• non-woven material
• office products
• packaging
• pallets
• transportation services
• trash/recycle services
• vehicle washing services
• vehicles

The application is seemingly limited only by the innovation, imagination and energy of the sourcing organization. How much of a company's spend is applicable to optimization depends on several factors. Those factors include the composition of its spend, its complexity, internal users' preferences, support and acceptance by management, skill levels, creative thinking in application, and the structure of the supply base.

Applying optimization for the first time to spend categories often results in significant cost savings. For example:

• A consumer products firm stated that 10 percent of its $30bn spend was applicable to optimization with possible savings of more than $120m.
• An electronics firm reported using optimization on 56 percent of its total spend, resulting in a cost savings of $600m.

Users indicated they are increasing the number of optimization events held each year.

Optimization Model and Rules

Optimization allows buying organizations to build mathematical models that incorporate all bids from all suppliers. If a "feasible" solution is found, the solution will cover all of the requirements (i.e. all of the items are covered), meet all the business rules, and be the lowest-cost or nearly the lowest-cost solution. (A feasible solution may not be found if some items receive no bids or too many business rules are imposed.) Most importantly, the buying company can solve the models multiple times to test the effects of various business rules on the solution. For example, the company could test the impact of such rules as:

• Limiting the maximum amount of business allocated to any one company
• Requiring a minimum of business to be allocated to minority- or women-owned businesses
• Allocating all the business for a low-volume item to one supplier
• Only using suppliers who have adopted "green" or sustainable practices

Analysis of the effects of including these various rules is typically called "scenario testing" and is an integral part of the optimization process.

There are numerous reasons why optimization should be considered for sourcing events. The following provides a discussion of the most frequently identified benefits.

Companies and providers claim savings of up to 10 percent over other approaches (manual and spreadsheet-based) in a shorter amount of time. Furthermore, these savings are achievable while maintaining good relationships with the supply base.

As one respondent stated, "There are no simple solutions in an increasingly complex world." Internal, environmental and marketplace issues combine to make sourcing events very complex. The buyer's ability to manage this complexity directly affects the quality of sourcing strategies. For example, one organization, in sourcing lodging, had 15 major geographic regions and almost 750 hotels to consider. In the past the sourcing strategy would require negotiations on a region-by-region basis. However, with optimization, bids from all 750 hotels in all 15 regions could be considered simultaneously.

Traditionally, suppliers were restricted to bid on lot sizes and combinations constructed by the buying company. Optimization allows suppliers to bid on lot sizes and combinations of items/services for which they can provide the most value and to provide tiered pricing. The term "expressive bidding" is derived from this latitude given to the suppliers. Given this flexibility in bidding, superior outcomes for both buyers and sellers can be identified.

The sourcing team can test an almost unlimited number of business rules or scenarios. These rules may emanate from preferences of the buyer, internal customers, upper management or other parts of the organization. The results of these tests can lead to significant sourcing changes. Often the new suggested solutions represent a significant change from traditional sourcing preferences that were implemented for long-forgotten or perhaps even arbitrary reasons. The cost attached to these preferences is readily available to upper management for review, elevating the sourcing decision to a more strategic level.

Implementable Sourcing Awards

Through the ability to test many scenarios, optimization can provide visibility to the lowest-cost sourcing solutions. Supply managers can then focus on those that can practically be implemented. After reducing the attractive alternatives to a manageable set, negotiations can focus on key suppliers to work on the details of the agreements.

There is little supplier resistance to optimization because the process is virtually invisible to them. Indeed, the process can work to the suppliers' advantage by letting them bid their most competitive packages and not be constrained to bidding on lots designed by the buying organization. Re-bids may force suppliers to lower bid prices, but also provides the opportunity to "stay in the game" and often be presented with the opportunity to bid on additional business.

The time spent analyzing bids can be reduced by as much as 50 percent using optimization instead of spreadsheets. However, with spreadsheets, all of the bid combinations for a complex buy were never fully analyzed, so the reduction in analysis time with optimization does not represent the total benefit. Moreover, the time actually devoted to analysis is spent more productively by testing business rules, rather than just finding feasible solutions. So, using optimization, less time is required to do more productive work.

After the sourcing team analyzes the results from the initial round of bidding, the team often asks for re-bids from some or all of the suppliers. The company might suggest that a supplier submit a more competitive bid, bid on another combination of business, or consider bundling different packages. The re-bids can be readily incorporated into the optimization process to produce new sourcing solutions.

The approach to analyzing bids is systematic and consistent from event to event. Finding the best solutions no longer depends on the inspiration of spreadsheet users. The process is repeatable and teachable. And, as supply managers become more familiar with the technique, there is less need for outside assistance.

The costs associated with backup sources to mitigate risk can be quickly evaluated. This evaluation can help minimize the chances for supply disruptions while highlighting the costs of such actions.

Because the optimization process is verifiable and repeatable, proposed sourcing solutions can be readily explained and defended. Executive management can readily understand the cost impact of various business rules and decide if seemingly attractive polices are worth additional costs.

As is true of any application, optimization comes with its own set of challenges. First, there is a cost to acquire and use the optimization software from a provider. Most optimization users have a suite of purchasing tools that includes an optimization module. But optimization modules can also be secured without buying a complete suite of tools. For instance, CombineNet specializes exclusively in optimization services and does not offer a suite of e-tools. All of the providers interviewed for this research offer the service "on demand," with access to the application over the internet. Training is highly recommended, even for "self-serve" applications, and can be secured from the providers for additional fees.

Beyond the training needed to implement and run the software, consulting services are often needed to construct a mathematical model that corresponds with the sourcing problem to be solved. Constructing appropriate models is not straightforward or intuitive, and the skill level needed is not often found in purchasing/supply departments. For at least the first time optimization is used, specialized consulting help will be needed. For large, complex problems, sustained consulting help will be needed to construct the problem in such a way that it can actually be solved by the available software and hardware. Assistance may also be needed to create alternative bid scenarios and correctly interpret the results.

In general, sourcing optimization models cannot be solved by desktop applications on desktop computers. Servers with lots of speed, memory and specialized software are needed to handle the great number of variables and constraints and to present solutions in a reasonable amount of time. These services are made available "on demand" by the service providers.

"Paralysis through analysis" is a real danger with optimization, particularly for large problems with many stakeholders who want to test many different business rules. Buying companies need to have procedures for ending scenario testing and proceeding with the tentative awards and final negotiations.

Clean Data is Required

Optimization models need large amounts of data that must be cleaned and rationalized. This is true for both buyer and supplier data. In particular, suppliers often have to invest extra time and effort to collect and organize the cost data needed for responding to the RFQ.

Optimization can deliver real value in sourcing events and is another decision support system to help supply managers to reduce costs. It gives sourcing teams a tool to extend the depth and breadth of their analysis capabilities in a timely manner. Analyzing multiple scenarios gets the buyer closer to the "sourcing sweet spot," that is, those sourcing decisions that balance the needs of internal users with supply market realities at competitive prices. Finally, it enables a high level of communication between buyers and suppliers, and buyers and internal stakeholders. Optimization can lead to better relationships with suppliers because:

• Awards can be based on price, risk, quality, delivery and other key performance indicators, approaching a total cost solution.
• Suppliers receive feedback during the event and have an opportunity to submit more competitive bids.
• Training is conducted with suppliers on how to submit their bids.
• The bidding process is open and transparent.

Optimization is not the right tool for every category or for every buy, but it does provide decision support for a wide variety of complicated and complex buys and can lead to significant cost savings.

The full report is available at www.capsresearch.org.

Study authors: Larry C. Giunipero is professor of supply chain management at Florida State University; Phillip L. Carter is  executive director of CAPS Research and professor of supply chain management at Arizona State University.

Read Full Article