Many factors can drive a manufacturer’s need for ERP integration with a third-party application: the need to create a supplier management program, for example, or to better manage inventory across the supply chain. Process-oriented initiatives often require cross-platform data, making integration necessary.
But when it comes to ERP integration wish lists, CRM takes the top spot, according to Joshua Greenbaum, principal at Enterprise Application Consulting in Berkeley, Calif. “CRM is the single most important application that you can deeply integrate. CRM owns the customer record. Everyone in the organization needs a reference point to the customer -- sales, marketing, procurement, accounts payable, even product design,” Greenbaum said.
Whether integrating your ERP with an on-premise application (such as Siebel) or one hosted in the cloud (such as SalesForce.com), your ERP systems integration project will typically follow these steps, according to Gerardo Ramon, process manager and practice lead for SAP Process Integration at itelligence, a Cincinnati systems integrator:
1. Discuss at a high level the depth of ERP integration you will need. The basic question: How big is the integration problem? As a first step, you will sit down with your systems integrator, independent consultant, ERP vendor, IT leadership, and business sponsor -- anyone and everyone who could possibly be involved -- and determine in general the type of integration to be implemented.
Will you use middleware or an enterprise application integration (EAI) platform? Do the two vendors have enough experience working together to take the project and run with it? If one of the applications has a service-oriented architecture (SOA) foundation, will you be able to use Web services for easier integration? If you will have to hard-wire a connection between a legacy application and another application, will internal or external people do the coding? Beyond the technical integration questions, the team should also consider people and process-integration issues.
2. Chart your ERP integration vision and strategy, from a scope, technology, capabilities, timeline, personnel, and budgetary perspective. It is important to identify at this early stage the challenges that you might face down the road, Ramon said. Integration is still expensive, even with tools and techniques that speed up the process. As with all large projects, it is better to prepare a timeline showing tangible chunks of progress instead of the proverbial Big Bang.
3. Develop the architecture. A number of key questions arise here. Where are the points of integration? Where do legacy apps play a role? How will you connect with them? How will you synchronize the data? How will you clean and prepare the data to get ready for the integration? Are you integrating on-premises software with Software as a Service (SaaS), and if so, what does that mean?
The integration team must also ask two more questions about the architecture:
One, does the integrated application require real-time or near-real-time data? Many processes are event-driven as opposed to real-time (for example, a sales rep places an order, and the warehouse fulfills the order). Event-driven monitoring demands more control than monitoring of real-time events.
Two, will the application pass a high volume of data continuously (i.e., synchronously), or is it more of an asynchronous batch process? Asynchronous event notification allows the application to constantly monitor events (such as inventory updates) without monopolizing system resources, while preserving throughput. Synchronous events require more bandwidth and will therefore affect the integration architecture. For example, a real-time synchronous application such as e-commerce will need real-time log-on capability and account verification, which would typically require linking to the CRM system to access customer master data.
4. Do the data cleansing and mapping in tandem, if possible, to save time. Data cleansing is generally the domain of IT and tends to take much longer than anticipated, though extract, transform and load (ETL) tools can help speed the process. Data mapping, meanwhile, must be led by business users who have deep understanding of the business processes involved. “You need a business process owner to define the process and the master data types you will use,” Ramon said. With this step complete, the stage is set for the applications to be connected according to your chosen architecture.