Most surveys indicate that more manufacturers are considering cloud applications to supplement their on-premises...
ERP. The combination promises the best of both worlds, but with the integration challenges of cloud computing increasingly coming to light, many might wonder how viable cloud ERP integration really is. Analysts say basic guidelines for choosing integration tools and a precise idea of the necessary touchpoints between on-premises ERP and cloud applications can lessen the uncertainty.
“ERP, at the end of the day, is a transaction system -- is an accounting system -- and there’s always a debit and a credit at the end of the rainbow,” said Bob Parker, group vice president at IDC Manufacturing Insights, a research company based in Framingham, Mass. “If what you automate [in the cloud] has a financial impact on the company, you need to integrate.”
Linking to Salesforce.com’s Software as a Service (SaaS) customer relationship management (CRM) application is by far the most popular integration target, said Mike Canniff, senior partner at the San Francisco-based integration consultancy Mercury Consulting and professor of management information systems at the University of the Pacific. “I also see a lot of Microsoft Dynamics CRM. It’s usually one of those two.”
Customer records in the SaaS CRM system -- typically accounts, leads and contact information -- must be mapped to equivalent records in the ERP system, according to Canniff. “Once a [CRM] account or lead has been closed and turned into an account, it becomes a real customer and that’s typically when the data needs to go over to the ERP system,” he said. “The actual orders themselves and order statuses need to be shared.”
The amount and type of data crossing over depends on how the company wants to manage an account’s entire lifecycle. In addition to managing some of the same customer and order data as the CRM system, the ERP software tracks the inventory needed to fulfill sales. Sales commissions might be calculated in CRM but executed in the ERP payroll module. “There are a lot of others, but that kind of hits the typical ones,” Canniff said.
Additionally, SaaS procurement and human resources applications belong with CRM in the top three of ERP integration targets, Parker said. Up-and-comers include global trade management services and order management, supply chain planning applications and warehouse management systems.
Where cloud applications, ERP meet
Several experts agreed that companies have three main technology choices for integrating on-premises ERP with SaaS: cloud-based integration services, on-premises integration middleware and the application programming interfaces (APIs) that come with many applications.
Before choosing an integration tool, Canniff recommends taking stock of the integration architecture that is already in place. Smaller companies tend to run lower-end ERP and accounting packages that have one-way flat-file or point-to-point integration to payroll, a setup that doesn’t mesh well with SaaS CRM, according to Canniff. “[CRM] tends to be a lot more bidirectional and dynamic,” he said, so integrating it with low-end ERP will usually require a separate integration tool.
For large companies, data cleansing is likely to be a major part of the project, and many cloud-based services and third-party integrators will do the work, Canniff said. Companies that have other on-premises applications to integrate should consider an on-premises integration product. And if the primary need is to integrate with an online business-to-business exchange to support large numbers of trading-partner transactions, cloud-based integration services may be the best option.
Parker said the cloud-based integration services offer significant network effects by multiplying the number of integration points an ERP system can reach and they sometimes provide performance management and business intelligence applications that run on top of the integrated data. But they also can overstate their capabilities. “The red flag for me is when they claim to have prebuilt whole processes,” Parker said, “when you get what I call the Lego pitch: ‘Look! It just plugs together!’ We’re not even close to that.” In reality, companies need mapping tools to translate the information from one application to the other. “There’s no such thing as turnkey integration,” he said.
For Parker, the integration’s complexity often hinges on whether there is a single, clearly defined handoff point between the SaaS application and the ERP (similar to the role the order plays in CRM integration), or if there are multiple integration points. The availability of standards often determines the eventual solution. “Even if you have lots of ERP integration points, standards can make it easier,” he said.
For example, if a manufacturer wants to use a cloud application that gathers data from point-of-sale terminals, the integration platform will need electronic data interchange (EDI) support, Parker said. But applications that are poorly supported with standards for describing data, documents or processes can present a challenge. “A good example is PLM [product lifecycle management],” Parker said. “There’s not a good standard for how [to] synchronize a bill of material”-- a key data type in PLM. In contrast, purchase orders are easily handled with a well-established document type called EDI 850 that is widely supported by cloud-based integration and procurement services.
The age of the ERP system is another controlling factor, Parker said. Newer ERP is more likely to support service-oriented architecture, or SOA, and “Web-intuitive APIs,” he said. “The degree to which the ERP package is customized will also affect your outside integration.
“Therefore, the first thing you have to look at is how modern is my ERP,” Parker said. The next factor is the degree to which the SaaS application has accounting implications. “The more the cloud-based application hands off at the normal starting point for ERP, the better off you are.”
But don’t be so mesmerized by data management issues that business process integration gets overlooked, Parker warned. A few years ago, business process management tools seemed like the ideal approach to high-level workflow integration across applications, but Parker now admits he and other analysts might have oversold the idea. “There’s a lot less of it than we thought there would be.
Instead, manufacturers are more likely to set up such cross-functional workflows by using APIs to integrate applications. The resulting connections are often asynchronous and lean toward data-level “publish and subscribe” models. “It’s not this grand [Microsoft] Visio diagram in the sky,” Parker said. “It’s more about taking advantage of the APIs we have.”