SUBSCRIBE

Here Today, Gone Tomorrow

Dynamic supply networks are replacing the traditional static model, and supply-chain analytics will never be the same

By Seth Grimes

Don’t expect marketers at companies in thesupply-chain analytics business to admit this fact: What they’re selling is an outgrowth of business process reengineering (BPR). Back in the early 1990s, high-priced BPR consultants urged companies to refocus on products rather than business functions, to realign production processes in ways that clearly linked inputs and outcomes. In those days before the seemingly permanent bull market, BPR meant downsizing — cutting payroll fat personified by hordes of middle managers — in the name of eliminating production barriers. Cutting out the middle was also a goal of earlier stages of the Web explosion. In the Web context, this trend was termed disintermediation, the ostensible elimination of barriers between an organization and its community.

Cutting out the middle leaves a void that automated, Web-enabled, front- and back-office operations have filled. First out of the blocks were enterprise resource planning (ERP) packages covering manufacturing, operations, human resources, and the like. Importantly, ERP packages impose information standards and process models; organizations that implement ERP packages collaterally gain regularized collection of actionable business information. ERP’s success encouraged development of systems for customer relationship management (CRM), sales-force automation (SFA), marketing automation, and supply-chain management (SCM). All these areas, and the supply chain in particular, focus on production-aligned processes, and all have databases at their core.

If you work in an “intelligent” enterprise, your organization possesses volumes of structured data on its operations and exploits that operational data to tune your business processes. If so, you probably extract data from transactional systems and reformat and load it to a data mart or data warehouse for online analytic processing (OLAP)-style exploratory analysis. However, a lot of progress has occurred recently to bridge ERP and other operational systems with data warehouses. These enterprise application integration (EAI) advances facilitate ad-hoc performance measurement. Thus, more sophisticated organizations have developed or adapted formal models that allow more effective optimization and planning — such as comparative analysis, searching for trends, fore- casting results, and exploring “what-if” scenarios.

Measuring performance according to these models, which derive from industry best practices and benchmarking, is the analytic bread-and-butter in the supply-chain space. (I see little evidence of the direct integration of analysis into supply-chain management.) But a revolution is brewing in the supply-chain world as marketplaces promise to transform procurement and logistics, replacing the standard operational model — static chains — with dynamic networks that will require us to rethink analytic goals and techniques.

Modeling and Measuring

SCM took a big step forward with the formation of the Supply Chain Council (SCC) in 1996. While some industry organizations make industry and member promotion their primary goal, the SCC’s most notable achievement has been the release of the Supply Chain Operations Reference (SCOR) model. Although the major constituencies were manufacturers, suppliers, distributors, and retailers, SCC developed the model “free from the influence of existing software capabilities and limitations.”

SCOR concentrates on key production processes — Source, Make, and Deliver — and the overarching coordination process, Plan. The model works on four levels:

• A top level describing the model’s scope

• A configuration level defining the four core processes in terms of more detailed subprocesses, or process categories

• A process-element level detailing steps within a process category

• An implementation level describing procedures that form a process- element step.

SCOR includes benchmarks at the process-element level, defining them as “performance metrics and formulas to quantitatively evaluate the operations of each supply-chain process.” SCC designed SCOR as a reference model for organizations that want to assess their operations relative to industry- and process- related functional norms. Therefore, the benchmarks provide numbers that help assess the supply chain’s efficiency in light of past and industry-standard performance.

Through surveys, SCC establishes standards, discovers and documents best practices, and tunes the benchmarks to support the model’s effectiveness as a comparative tool. One survey of note, conducted last year by the Performance Measurement Group (a subsidiary of an SCC co-founder, consultancy Pittiglio Rabin Todd & McGrath), found that four indicators are especially effective for measuring resource utilization and supply-chain efficiency:

• Total supply-chain management cost

• Upside production flexibility, or the number of days required to achieve an unplanned, sustainable 20-percent increase in production

• Cash-to-cash cycle time, the gap between payments made for materials and payments received for products

• Delivery performance to request, the percentage of on-time or better order fulfillment.

You can calculate these indicators with relatively simple formulas that are well within the capabilities of typical OLAP tools, as are just about all SCOR-model measures. As a reference model, SCOR does not rely heavily on advanced analytic functions such as forecasting, which would normally contribute to planning, but does little to help measure the effectiveness of an organization’s Plan process.

In the four years since the establishment of the SCC, the council has sought to extend SCOR to cover global, multi-organizational, cross-functional supply-chain operations as well as new, non-manufacturing industries. (The SCC has not yet realized that “Source” is no more a process than “Eat” is. Fortunately, it’s a better modeler than grammarian.) These extensions are linked: Modeling and implementation by services industries are prerequisite to designing chains that cross old boundaries among manufacturing, supply, distribution, and retailer processes.

SUPPLY CHAIN OR VALUE NETWORK?

Why the term “supply-chain management” is bound for extinction The term supply chain stands out in the crowded lexicon of computing jargon as a poor descriptor. The supply chain is not self-contained: It is by nature outward-facing rather than inward-looking. A particular link in the chain represents supply for one partner and demand for the other. The “supply” in supply chain is adequate from only one point of view. The supply chain includes procurement, sales, and logistics and ideally would interact with supplier, shipper, and customer systems. When partners work together you get a collaborative supply chain, with joint planning and forecasting, but the added modifier doesn’t clarify the supply-demand confusion. Next, organizations seek to model and manage their own demand side — customer orders, fulfillment, shipping — and the production processes that transform supplies into products that meet demand. (This demand side, per the preceding paragraph, is the customer’s supply side.) In today’s world, maintaining excess inventory, whether of production inputs (supplies) or of products, is bad. Clearly the supply and production should be aligned with demand, and in fact, supply-chain products do consider this complete picture. The supply in “supply chain” has to go: Value is better. We’ve only done half the deconstruction job: The term value chain is still bound by the inadequate chain metaphor. A supply chain may have branches and redundant strands. For example, there may be more than one supplier for a given part or more than one part that can serve a given role in the finished product. There may be more than one shipping method available with each method having a cost, risk, and speed. There may be many distribution channels — direct-to-consumer, reseller, distributor — with varying benefits and costs. Thus, the supposed chain is really a web or network: a value network. The establishment of online marketplaces supports this perspective. The goal of a marketplace is to facilitate creation of ad-hoc supplier-customer relationships, to create dynamic trading networks that link the value networks of individual organizations, globalizing reach and speeding transactions while reducing costs. Online marketplaces are new and the focus right now is on positioning. As marketplaces enter the mainstream, the term supply-chain management will likely disappear.

Architectures and Implementations

One architectural approach is to provide tools that are tightly focused on the job at hand, analyzing procurement and delivery performance in a single-enterprise (noncollaborative) supply chain. For example, InfoRay’s Monitors and X-Ray distill information consolidated by the Business Context Engine into visual displays of categorized, linked business indicators. InfoRay’s tools are suited to a limited market niche but they can’t cover the full scope of a business’s operations. SeeCommerce (formerly VIT) similarly offers tailor-made analytic components — the SeeChain suit comprises supplier, demand, materials, production, inventory, and fulfillment components — but prefers to work with external SCM and ERP applications rather than interface directly to operational databases.

An alternative approach is to integrate best-of-breed OLAP tools into supply-chain applications, providing predefined data models, measures, and views. Here are some examples:

Our modern economy treats employees, contractors, and consultants as “human resources” (HR). Where you have resources and demand, you have a supply chain — called a “service chain” by Changepoint Corp., whose solution supports recruiting, assignment, deliverables tracking, time and expense management, and billing. Changepoint based its proprietary operational model on its own HR experience, but its decision-support components are strictly from off the shelf.

Changepoint provides analytic capabilities by integrating Knosys Inc.’s ProClarity OLAP application into the Changepoint Windows interface. Knosys has attempted to make ProClarity integration easy by offering a customization GUI as well as functionally equivalent ActiveX components that work behind the scenes using Microsoft SQL Server and OLAP Services. You can also extend ProClarity’s analytic capabilities for forecasting, advanced visualization, and so on through third-party components. Professional-services administration is a corporate front-office function — an area where Microsoft tools are pervasive— so relying on the Microsoft database, operating systems, and Component Object Model (COM) is a workable strategy.

Changepoint provides five OLAP cubes, which draw from operational data, along with reports and visual data representations — the latter including decomposition trees that segment values along dimensions and perspective views designed to show relationships among measures. These ProClarity interface elements are put in the domain-specific service of analyzing the sales pipeline, service delivery, resource utilization, and financials. Other vendors — most notably hosted-ERP vendor Infinium and manufacturing-chain company Manugistics — similarly embed ProClarity components in their SCM suites.

Hyperion Solutions Corp. is another analytic tools company that is eagerly supporting SCM, again via alliances and by extending the general-purpose capabilities of its OLAP database, Essbase. Essbase 6 sports greater capacity and new calculations and functions — including new trending and forecasting capabilities — that should make it more attractive in the SCM space.

Essbase gained market position because of the decision by Arbor Software, which was subsequently acquired by Hyperion, to promote Essbase as an easy-to-use, ubiquitous back end for analytic applications. Hyperion integrated Essbase with its own financial and budgetary analysis applications but has continued to make the database available through channels, including resale by IBM as an OLAP add-on to DB2 (DB2 OLAP Server).

SCM heavyweight i2 Technologies Inc. is Hyperion’s most prominent supply-chain partner. I2’s Rhythm supply-chain optimization suite uses Essbase as an analytic engine to enhance capabilities for factory planning, cycle-time optimization, and sales and operations planning. (Excuse the omission of details of models, performance indicators, and functions implemented by Rhythm with Essbase; they’re not especially different from what I’ve written about ProClarity.)

I2 is also one of the stronger aspirants, as gauged by considering technology and positioning, in the emerging network of online marketplaces. I2’s recent strategic moves — alliances with e-commerce platform providers IBM and Ariba Inc. and acquisition of Aspect Development and Supplybase and their complementary technology — will help i2 compete with ERP leader SAP and “if we can’t beat Microsoft on the desktop, we’ll do it everywhere else” vendor Oracle.

Marketplaces

Transforming traditional supply-chain models — SCOR’s Source, Deliver, and Plan processes in particular and their equivalents implemented in proprietary tools — to cover online marketplaces will be a challenge. The task is to model and optimize dynamic trading networks (see the sidebar, “Supply Chain or Value Network?”) where transactions are mediated by e-marketplaces, also known as exchanges, rather than solely through relatively static supply chains. Transformed models will require whole new sets of performance indicators; indeed, a different order of analytic capabilities will be required.

The big problem on the analytic side is the adequacy of current tools and approaches for new integration and analysis needs. Data in marketplace environments will be extremely dynamic as well as diverse: Marketplace participants, inventories, and prices can vary rapidly depending on market and external conditions. The road to integration has at least been mapped, even if the trip will be a long one: Marketplaces describe standardized nomenclature and operational data elements with extensible markup language (XML) but may also use traditional electronic data interchange (EDI) methods and protocols to conduct transactions.

The integration task is two-fold: An organization adapting itself to a world of trading networks will have to open its procurement, inventory, sales, and planning systems to one or more exchanges, and it will also have to work with traditional procurement and distribution channels. This integration is essential if you want to align procurement and production with demand, but these disparate marketplaces (and traditional channels) may have different trading conditions and provide data that may vary in reliability.

The first task in developing analytic capabilities here would be to either construct a single, logical view of heterogeneous, distributed data sources — that is, a coherent picture that would be easily understood by an analytic tool — or to extend such tools to understand XML dialects and work simultaneously with diverse data. I’m describing a federated database or “analytic portal” on the one hand — implementations of each are on the market — and tools that you can integrate into online trading platforms on the other. Integration requires an open architecture with an object model, and interfaces that are suitable for the public Internet and compatible with those used by leading e-commerce platforms. Microsoft’s COM and ActiveX are workable on high-bandwidth intranets, but have not been widely adopted in the nascent business-to-business (B2B) e-commerce world. Instead, integration requirements point to CORBA, Java, and JavaBeans. I expect to see tools that use these technologies do well: AlphaBlox Corp.’s Analysis Suite and Painted Word’s Mocha Blend Analysis Platform are two examples.

Vendors such as Informatica Corp., which recently acquired analytic tools publisher Influence Software, are well-positioned to handle both data integration and analysis. Influence Software, like other vendors cited, provides canned e-business models and measures. We should see later this year what comes of Informatica’s agreement with B2B platform vendor Ariba’s ORMS procurement solution. The goal is the same as that achieved for traditional supply chains: to source data marts with operational data and provide key performance indicators and analytic tools. Electron Economy is another company seeking to handle the data integration side albeit without Informatica’s analytic capabilities. Ironside Technologies Inc. is yet another, working to bridge exchanges built on disparate platforms; its work focuses on multi-exchange operations. That’s a big leap from modeling and analyzing information flowing through a single marketplace — and products aren’t there yet. The flow through a network of exchanges will be even more complex, and it may be simply too early at this stage to do significant analytic work.

Ways to Go

Organizations with which I’ve discussed extensions of analytic models have been guarded, unwilling to offer details of work in progress. My guess is that it’s not because they have a competitive edge to protect. Rather, it’s just too early to spend a serious effort on revamping analytic approaches when the quantities being measured — the performance indicators — are not yet well defined. Besides, everyone expects exchange-brokered goods to cost less due to marketplace competition, and significant transaction-cost advantages should result from the standardization imposed by exchanges. Performance tuning is often about incremental improvements that will be small relative to the gains expected by moving procurement, distribution, and planning to marketplaces where an organization can collaborate closely with business partners and align its own procurement and production with demand. Stated simply: it’s too early to think about serious dynamic-trading-network analysis.

When the time does come, I won’t be surprised to see techniques from logistics and communications network analysis applied in this area. The different types of networks are similar in that they contain many nodes with multiple paths among them, each path having distinct cost, capacity, and performance characteristics. Supply-chain analytics will have to evolve from interactive, slice-and-dice manipulation using bolt-on or embedded OLAP tools to integrated realtime, closed-loop analysis.

Rate This Article Rating: 1 (best) 2 3 4 5 (worst) Comments: Optional e-mail address: