Offering vertical options

Getting the vertical expertise into a supply-chain app means taking business processes into account

The supply chain is a critical battleground for software vendors these days, as it begins to overlap with ERP and CRM and gains a greater share of otherwise shrinking enterprise software expenditures. So it’s no surprise that a lot of money and brainpower are going toward customizing supply-chain solutions for various vertical industries — on the part of both the larger ERP players and a host of smaller, specialist vendors.

Much ink has been spilled over the battle between these software behemoths and the smaller supply-chain vendors. And it’s certainly tempting to ask whether horizontal supply-chain solutions, customized with vertical templates, can possibly serve customers better than can vertical solutions that are focused and purpose-built for the business processes of various target industries.

But I’m going to take a radical stand on this one: Supply-chain verticalization isn’t about architecture, it’s about business logic. It doesn’t really matter how you architect this logic into your code base, as long as it’s the right logic. The task of modeling and optimizing vertical supply-chain processes can be so intense and complex, and there’s so much leverage in minor process variations, that the vendors who sweat the details best for any given industry vertical — whether they be large companies or specialized vendors — will come out on top.

At the end of the day, effective SCM (supply-chain management) is about analytics, workflow, and problem solving — and getting those things right means understanding the customer better.

Steve Williams, senior vice president of enterprise product development at Cambridge, Mass.-based Aspen Technology, notes that supply-chain software that is customized for a vertical industry must excel at real-time modeling, problem solving, and integration with various operational systems. Aspen Technology provides supply-chain applications to the process-oriented vertical industries for petroleum, chemicals, and pharmaceuticals.

"You've got to have more sophisticated modeling technology in your supply-chain toolset to take account of the specifics of the industry,” Williams explains. “You’ve got to be able to pull up the constraints that apply in the plants and reflect them in the overall supply-chain optimization problem.” 

Pallab Chatterjee, president of solutions operations at Dallas-based i2, concurs: "If you take a whole refinery system, it’s a mammoth problem to optimize.” And, of course, this level of specialization isn’t unique to the oil and gas industry; in textile manufacturing, fabric dyes from different dye lots cannot be mixed because the final product wouldn’t have a uniform color. In the semiconductor industry, where a typical factory produces thousands of different products at once, certain product assemblies require pulling inventory with specific quality characteristics, such as chip speed. These complex “share mix” constraints around how inventory supply and subassemblies can be used to create finished products is found in many vertical industries, affecting their supply chain requirements.

Integration issues

Another challenge for vertical supply-chain solutions is integrating with existing operational systems, including transactional and accounting systems, MES (manufacturing execution systems), and shop-floor control systems. Solid integration allows companies to reap the full benefits of end-to-end, extended-enterprise planning, execution, reporting, auditing, and performance management.

"You've got to connect the supply-chain software into the plants to get effective integration into those business processes,” Aspen Technology’s Williams says. “That’s a difficult exercise when a lot of important data is trapped in a spreadsheet on Joe Blow’s desktop. We need to get to an enter-data-once regime."

Most vendors use traditional asynchronous EAI messaging buses along with batch data transfer but are trying to move to XML, SOAP, and Web services to capitalize on the flexibility of these technologies. “The ability to use a services-enabled platform to deliver composite apps — I think that’s the next generation of verticalized solutions,” predicts Everett Plante, senior vice president of product development at Manugistics inRockville, Md.

In a factory setting, for example, supply-chain software must act as the integration engine across ERP solutions and third-party logistics software on both the UI side and the data side to optimize factory capacity and manage inbound supply, customer orders, and demand on a real-time basis.

“We see [the potential] to bring that together under a common framework, to manage and provide visibility into [vertical solutions]," Plante says. But, he notes, a whole range of technologies under development today must first come together, including portals, BPM (business process management), data integration, content correlation, and Web services — a situation that affects all supply-chain apps and not just those specialized for vertical industries.

In many cases, the visibility this integration provides will be mandatory, not optional. New supply-chain-related compliance regulations, such as the Food and Drug Administration’s 21CFR11 standard for tracking pharmaceutical products, require a high level of transparency with regard to the processes involved in creating a product. And for performance management purposes, customers want to be able to align business objectives with performance at all levels of the organization.

"We’re sort of closing the loop on the biz performance" with the new generation of supply-chain applications that are highly customized for and tightly integrated to vertical business processes, Aspen Technologies’ Williams explains.

The configurability debate

So how do these complex vertical supply-chain challenges get solved? Through a combination of customized code, configurable code, templates, and the aid of domain experts, who can support the process and the software with years of hard-earned experience. Take i2, which offers 10 tailored templates for 18 vertical industries, ranging from consumer electronics, durables, and packaged goods to automotive and industrial OEM, metals, and paper. i2’s Chatterjee claims there are 700 different configuration “flags” that can be set in i2’s software, of which 300 may be defaulted for one industry.

More importantly, the templates include workflows, modeling constructs, and “solver” optimizing engines, which represent best practices in each vertical industry but can be modified by customers who don’t necessarily follow industry-standard business processes or workflows. “You have to be able to represent the customer’s business processes, then set it up so he can optimize,” Chatterjee says. “The way Dell does work is very different from the way Compaq does work.”

Some disagree strongly, however, with the configurable template approach. “When I hear the term configurable software, to me that says software written the old way,” says Steve Banker, SCM service director at ARC Advisory Group in Dedham, Mass. “To the extent that you try to have features and functionality across all industries, you get this incredible tree structure; you end up with software bloat, version release becomes more and more difficult; it just becomes very unwieldy.”

Banker argues for a more granular, purpose-oriented approach to vertical supply-chain software development to avoid extensive customization of generic code. “The problem is, you want to do it differently with each customer, and configurable software doesn’t particularly handle that well,” he says.

But Udo Dengler, CTO of Los Angeles-based supply-chain vendorAdexa, says that you have to give experts in the field something to start with. "Put as much in the software as makes sense as a default," such as configurability steps that the field expert can implement, he advises.

According to Dengler, Adexa is focused on finding the commonalities between vertical industries to provide tools that its consulting organization can leverage. He calls these “generic, user-definable attributes” that are based on a “unified data model” — a language of data objects and “plug-in rules” to represent business problems and common attributes in the different verticals. “It’s object-oriented programming,” he explains. “You create a generic attribute object, and specific data objects like 'demand objects' make use of them.”

"Vertical solutions of the past were basically hard-coded solutions, nonconfigurable," Manugistics’ Plante says. UI and database configurability and an application’s capability of triggering industry-specific algorithms, even switching back and forth between algorithms on a per-SKU or demand-item basis, has made vertical capabilities more flexible. “It’s fundamentally Java built over a relational database that’s allowed us to do that," Plante adds.

The template debate may not be new, but Aspen Technology’s Williams adds that he doesn’t think the market “is willing to accept the type of standardization SAP imposed in its space 15 to 20 years ago. … They're not willing to buy off on the fact that their business process is wrong." And that reaction is helping drive the push for vertically specialized apps to match existing market-driven business processes.

Copyright © 2003 IDG Communications, Inc.

How to choose a low-code development platform