Category Archives: Decision Optimization

Innovation in Sourcing

Today’s guest post is from Chetan Raniga, who is General Manager, Americas at Trade Extensions.

As someone who’s been in the strategic sourcing field for over ten years as a consultant and product manager, it’s been interesting to see the rapid evolution of sourcing solutions over the past few years. Leading solution companies now realize that users need solutions that feel familiar; that’s why Excel integration is common among leaders in the supply management arena. It’s the same reason Coupa has screens that almost mirror Amazon.com — providing an interface and workflow that’s both familiar and intuitive. Another example is the use of dashboards — the charts and alerts in Trade Extensions remind users of Mint.com, a popular personal finance site, though they show vastly different types of data!

Here are some other changes for the better:

Collaboration and Workflow:
The sourcing solutions of the past were extremely tactical (e.g., automating the process of running an RFQ or auction for a specific category), and therefore, didn’t give buyers the ability to share ideas, exchange documents, and view the real-time status of their sourcing initiatives. Now, platforms provide robust project management capabilities with Gantt charting, custom workflows (e.g., only have new suppliers go through the qualification step), document sharing, and Skype-like chat features. A buyer can see exactly which suppliers, team members, and stakeholders are online, and instantly communicate with them. Audit trail and logging capabilities have also gotten stronger, which is important for the confidence of both buyers and suppliers in using these platforms. Multiple teams are now using these platforms to share data. For example, a Direct Materials sourcing group can incorporate freight pricing from a tender conducted by the Logistics team. The group can use the platform to determine which items will use the suppliers’ freight (delivered pricing) and which will use the 3rd party carriers’ freight (FOB plus freight).

Flexibility:
Systems of the past didn’t provide the flexibility that we have today in collecting inputs. Labels such as ‘Price per Unit’ would be hard-coded or the cost formula would only support a limited number of operators and functions. Data entry was also cumbersome and error-prone since it involved either manually entering or copying-and-pasting vast amounts of data. Today’s solutions integrate with Excel, so that existing data and formulas can be easily leveraged. For example, item, demand, and cost component data stored on separate Excel spreadsheets can be uploaded with one click. Even better, some solutions allow users the ability to customize the supplier’s bid form. This is critical to change management since companies can continue to use their existing bid forms in the bid gathering phase but obtain the decision support and reporting benefits in the analysis phase. These improvements have led to even shorter RFQ/P creation times.

It’s also now possible to run auctions with optimization (a step forward in utility from the original concept of reverse auctions), and to run RFQ/Ps with feedback — blurring the line between RFQs and auctions but also going further by providing custom feedback (e.g., a custom message of “Not Competitive” is shown when the bid is x% greater than the median price).

Usability in the Analysis Stage:
The one sourcing area that has lagged in adoption has been the use of optimization (which the doctor has defined as the application of one or more rigorous analytical techniques to a well-defined model to generate the absolute best decision from a multitude of possible alternatives in a rigorous, repeatable, and provable fashion). It sounds complicated, and in the past it really was. For example, if a customer wanted to see what the result would be if all the business went to incumbents at their current proportions, then she’d have to create a rule limiting allocation for each item and affected supplier. That’s painstaking when you have a couple of hundred items — but most projects had thousands of items! Nowadays, in a modern optimization solution (which include the solutions by Trade Extensions and BravoSolution), the buyer just selects one rule, written in plain English (as shown below).


Even better, new platforms allow buyers the ability to create rules in Excel and then upload them. In the example below, the buyer is setting limits by plant and supplier simply by completing a table.


Reporting:
The solutions of the past didn’t offer much in terms of reporting. Most had a couple of pre-defined reports that exported to Excel. Buyers had to spend additional time modifying the reports — even changing labels and creating pivot tables — before they could present the results to their peers and managers.

Solutions today have made major strides in this area. Leading spend analysis tools (which include BIQ as well as Trade Extensions) allow users to create custom reports that can be saved as templates and re-used. The ability to choose specific dimensions (rows), columns (facts), and other information means that users no longer have to go outside the system for further manipulation. Some tools even allow the ability to drill-down/up on data (e.g., view allocation data by country first, then by region, then by state, and finally by city/plant).

We have heard buyers comment that their analysis time is shortened by three-and-a-half (3.5) weeks on average by using the new decision support and reporting capabilities mentioned above.

Thanks, Chetan!

How can we minimize our costs while maximizing the use of our current assets?

Modelling, Simulation, and, most importantly, Decision Optimization … and I’m very glad to see that the ISM recently ran an article in Inside Supply Management on “Improving with M&S”. Because, as the article points out, once the behaviour of a system is understood, supply managers can realign assets with the intent of finding the arrangement that will yield the most optimal measures of performance.

Furthermore, as the article points out, modelling, in its most basic form, is the creation and arrangement of representational elements that approximate reality. Simulation is the potential interaction, or playing out, of these elements over time. And optimization is the application of rigorous analytical techniques to a well-defined scenario to arrive at the absolute best decision out of a multitude of possible alternatives in a rigorous, repeatable, and provable fashion. Together, the technologies can be used to determine how to balance costs and resources.

The article gives some great tips for novices in MOS (Modelling, Optimization, and Simulation). Specifically, it notes that:

  • the goal is not to mirror the real-world system in all its detail but to represent those aspects that are expected to meaningfully affect the performance variables of most interest
    as it is impossible to model everything, there is a need to focus on what is relevant; for example, there may be 50 different cost elements in a breakdown of a component down to raw materials, but if 10 of them represent 80% of the cost, and only 5 of them are variable, focus on those 5 and wrap the other 46 into one
  • if the model and simulation are producing behaviour similar to the known real world, then the model most likely is an appropriate, or valid, reflection of that real-world system
    if the model works well, don’t second guess it; unknown is unknown and trying to guess the unknown doesn’t make it known
  • we learn through experimentation
    but experimenting with the real system is costly, counterproductive, and could meet organizational resistance
  • the process of modelling and simulating a real-world system assists in identifying major system strengths and weaknesses
    strengths which can be leveraged through targeted interventions and weaknesses which can be overcome through appropriate system redesign; the process of identifying strengths and weaknesses and virtually experimenting with changes to the system allows analysts and supply managers to see unintended second- or third-order consequences that might not have been readily apparent otherwise
  • selecting and analyzing various supply management activities and processes can be crucial in improving efficiencies and performance
    if the organization doesn’t know where its performance can be improved, then it will not be able to improve its performance

All great tips and all reasons why modelling, simulation, and optimization should be used by each and every Global 3000 organization.

What to Look for in a Strategic Sourcing Decision Optimization Solution

Once it is understood that Strategic Sourcing Decision Optimization is the application of rigorous analytical techniques to a well-defined sourcing scenario to arrive at the absolute best decision out of a multitude of possible alternatives in a rigorous, repeatable, and provable fashion, one can define some core capabilities of a strategic sourcing decision optimization platform. Then, when one is looking for a sourcing platform that includes decision optimization, one can determine whether or not the platform includes true strategic sourcing decision optimization foundations.

The following are core capabilities that should be present in any platform that claims to be based on strategic sourcing decision optimization:

  • Solid Mathematical Foundations
    LP, MILP, QP, and Convex optimization are good foundations. Random sampling, Monte Carlo simulation, and evolutionary / genetic programming are, on their own, not sufficient.
  • True Cost Modelling
    The models must be accurate and complete. Not “close” approximations.
  • Sophisticated Constraint Analysis
    At a minimum, capacity, allocation, (risk) mitigation, and qualitative constraints must be supported.
  • What If? Capability
    The “holy grail”, the tool must be able to generate, analyze, and compare multiple “what if?” scenarios in order to truly be useful to the organization.

In addition, the following capabilities are nice to have:

  • Constraint Impact Analysis
    Why is this solution “optimal”? Which constraints are driving the allocation?
  • Network Modelling
    for the analysis of demand across multiple categories and network (re)design
  • Automatic Scenario Generation
    that automatically creates “what if?” variants of a given scenario to jump-start analysis

For more information, see our recent article on What to Look For in a Strategic Sourcing Decision Optimization Solution over on the new Next Level Supply site. This article, that summarizes and updates some of SI’s best writings on Decision Optimization, including the Next Level Purchasing Optimization Interview and the e-Sourcing WikiPaper, is a good refresher for those of you looking to (re) acquire a sourcing platform based on strategic sourcing decision optimization.

Three Great Tips for Optimizing the Distribution Network

A recent article in Logistics Management on Warehouse and DC Management: 6 Tips for Optimizing the Distribution Network provided a number of great tips for optimizing the distribution network and reducing logistics costs which are expected to quickly surpass the 2008 high (on the US Freight Index) as oil prices increase and availability of qualified truck drivers decrease. The following three tips are particularly pertinent.

  • Ask the Right Questions
    What are the perceived service level requirements? What impacts will changes in delivery lead-time have on revenues in a given market? What are the baseline operating expenses, inventory assets, and capital investments? How do these compare to alternative scenarios? Should the company home-source, near-source, or global-source? Remember, Supply Management must deliver value and balance costs and lead times against revenues and demand times in its redesign.
  • Use an Effective Network Modeling Tool
    Effectively modeling a network in home-grown spreadsheets and databases is impossible for an average enterprise with more than half a dozen locations and global sourcing requirements. Not only does an organization have to effectively model a network and proposed alternatives to identify the best network designs, but it needs to monitor what’s happening with the network, which is not possible if all the organization has is a simple spreadsheet or database tool.
  • Perform an Inventory Optimization Study
    While adding more Distribution Centers (DCs) may reduce transportation costs, it will also increase inventory costs, on average, as more inventory will generally be necessary to meet default stock levels. However, if not all distribution centers service locations with the same service level commitments, then inventory can often be varied to minimize carrying and holding costs. However, a careful analysis, supported by an appropriate toolset, will be required.

Trade Extensions: No Rest for the Wicked-ly Powerful – Part II

As per yesterday’s post, it’s been less than five months since we last checked in with Trade Extensions, who had traded up to a Fact Sheet User Interface and added a slew of new features, including improved RFI support, multi-dimensional rankings in e-Negotiation, Google Earth integration, new incumbent rules, and an OLAP foundation to reporting, including the implementation of a new n-way comparison report. Since then, Trade Extensions has been on a tear to add new functionality as fast as it can to make the platform not only one of the most powerful expressive bidding optimization platforms on the planet, but also one of the easiest to use — listening to its users (which include the Fortune 1000) and adding features and functions that make an average buyer’s life easier, taking usability to a whole new level yet again. And while earth-shattering technology improvements are cool, it is usability that is the ultimate key to to adoption, use, and, ultimately, cost avoidance and reduction in your sourcing organization.

Scenario Creation & Analysis

Not only are there new rules that allow partial awards to be fixed based upon existing scenarios, but the number of constraint categories has doubled. While there were only general and incumbent constraints in the past, there are now an entire category of scenario reference rules and post processing rules. With respect to scenario reference rules, not only can allocations be kept, but bids can be favoured or penalized as well. The post-processing rules are also quite useful. Allocations can automatically be rounded and allocations that don’t meet a minimum number of units can be removed (or re-assigned to the supplier who meets a minimum allocation with the lowest total cost).

Feedback Mechanisms

The buyer now has fine-grained control over what the supplier sees, and can even mix feedback types. For example, if the buyer only wants the top three suppliers to know they are top three, but suppliers four to six to know their exact rank, they can specify that specific rank starts at bidder four, and the top bidders default to “top 3”. In addition, if the supplier does not meet a minimum bid increment, which can be defined in a number of ways (including, minimum dollar or % decrease over last bid), the supplier gets a nice red error that the bid is not acceptable AND a message indicating the minimum increment required. Finally, and this is really cool, the user can define custom color-coded bid feedback fields based on dynamic formulas that now only let the user know where they rank, but how competitive their bid is (against the current bids from the competition) in English using a buyer defined scale such as “Competitive”, “Slightly Competitive”, “Not Competitive”, and “Not Acceptable”.

Plus, the buyer can now chat with users online in an integrated IM client, and immediately see who is online when they log in as it is a widget on their project management dashboard.

Odds and Ends

The “dashboards” for RFX and auction phases have also improved. The summary, bidder summary, and lot summary are now completely customizeable by the user, support custom fields, and user-defined colour codings in the rankings. In addition, there is integrated show/hide, drill-down functionality, and customizeable pop-up (bid, trend, and bidder activity) charts where a user can select one, some, or all of the rows in each report.

They have also added a basic workflow engine that allows buyers to initiate rate requests, lot requests, and allocation publishing requests of project managers / administrators when new needs arise during a project. This allows managers and supervisors to maintain control and a complete project history to be maintained. The workflow is fairly basic at the present time, but I suspect it will mature and fill out quickly given Trade Extensions’ track record of rapid application development over the past two years. (Especially since the feature is being used by a couple of very large companies.)

All and all, it’s a lot of new functionality in a short time frame that makes the tool extremely useable by an average buyer.