Critical Success Factors for an MDM Initiative A Case Study of Cypress Semiconductor Mahender Bist Rapidflow Apps Inc. Jay Kilby Cypress Semiconductor Abstract Master Data Management (MDM) is becoming an increasingly important cross functional organization-wide discipline in small and large enterprises alike. The main reasons for heightened awareness in the MDM global business environment with information across multiple platforms and systems are spotlight on governance and compliance, and strong emphasis on data quality to improve operational and analytical intelligence. MDM solutions provide the means for consolidating, cleansing, centralizing and distributing master data throughout an organization to ensure consistency and streamlined processes to maintain data for organization-wide usage. Master Data Management is a vast area and can cover Product, Supplier, Customer, Accounting etc. This case study deals with the Product Master Data Management initiative at Cypress Semiconductor using Oracle PLM to address the following pain points: Cypress needed a single source of truth for Product information. The Product data in multiple systems was cumbersome and error prone to maintain. A limited parametric search capability in the legacy systems resulted in potential part duplication. Item creation took too long; Silo processes coupled with multiple system batch updates for part creation took 10-12 days/part Request There was no standardization in part numbering, descriptions and attribute definitions. There was little or no automated control over the Product Lifecycle; Cypress needed a phase-gate approach to update item lifecycle phases and automatic status changes based on lifecycle phases. Needed secured access to include role based attributes management. This case study walks you through the challenges involved in each step of the MDM initiative as shown in Fig.1 below: System Architecture & Rollout strategy Software Selection Oracle PLM Implementation Post Implementation Decision Challenges What should the entitled architecture be? What order should we implement the applications in? Challenges What is the best software to meet our business needs? Implementation Challenges How to streamline NPI and ECO processes? How to centralize and synchronize data to Multiple Legacy systems? How to standardize data & resolve data governance issues? Post Go-live Challenges Continuous Improvement BOM Creation Fig.1: Challenges COLLABORATE 11 OAUG Forum 1
This whitepaper uncovers the solution approach adopted in resolving the challenges, the benefits achieved and the lessons learned. Most importantly, it consolidates the case study experience into critical success factors for an MDM initiative. Company Background Cypress Semiconductor Corporation (Cypress) was established in 1982. It is headquartered in San Jose, CA and has manufacturing facilities at Cavite, Philippines and Bloomington Minnesota. Cypress has eight design center in the US, India, Ireland and Belgium. Cypress products are sold and shipped across the globe. Cypress is a broad-line semiconductor company and a supplier of programmable solutions in systems. Approximately 40 K+ customers use Cypress s flagship PSoC programmable system-on-chip platform and various end products that range from cell phones and MP3 players to washing machines, personal computers (PCs) and communications switches. In addition to PSoC, Cypress offerings include capacitive sensing and touchscreen solutions, universal serial bus (USB) controllers, and general-purpose programmable clocks. Cypress has four segments: Consumer and Computation Division Data Communications Division Memory and Imaging Division Emerging Technologies and Other Systems Background Cypress had a number of custom built, purchased legacy applications along with Oracle Financials applications categorized as below: Custom built applications to manage core part data and manufacturing Custom Manufacturing Execution systems for Wafer/Die manufacturing and Chip assembly Cypress custom built order management system to Order tracking system Custom built database for planning system Oracle Financials (Base Modules) iprocurement for Indirect procurement Limited Finished Good Items in Item Master Project Background To meet its aggressive growth strategy while maintaining high customer service with 95% on-time delivery to the original schedule, high quality and reduced lead times, Cypress had a significant challenge ahead of it of revamping the system quickly and its infrastructure. From the MDM perspective, some of the key challenges were: 1. Product Master Data Management s place in the System Architecture roadmap 2. Product Master Data Management software selection 3. New Part Introduction (NPI) and Engineering Change Order (ECO) process 4. Data Centralization and Synchronization 5. Challenges in data governance and data quality 6. Continuous Improvement after go-live COLLABORATE 11 OAUG Forum 2
Implementation Challenges, Solutions, Approach and Benefits 1. System Architecture & Rollout strategy - Product Master Data Management as First Phase Decision 1: The Enterprise Business Architecture team under Jay Kilby s (Vice President) direction along with the IT team finalized the future Oracle based applications footprint as represented in Fig. 2 below: SHOP QUOTE ORDER SCHEDULE PLAN F F MANUFACTURE FINISHED GOODS / SHIP FORECAST & SCHEDULE Forecast Plan PLANNING Plan FAB & ASSY MES CUSTOMER PORTAL 1 CRM ORACLE ORDER MANGAGE MENT SHOP FLOOR MANAGER BILL OF MATERIAL MATERIAL REQUIREMENTS PLANNING (MRP) 4 4 FINISH GOODS 5 SHIP PURCHASE ORDER 1 5 ORACLE PLM (PART MASTER) 4 INVENTORY 2 ORACLE FINANCIALS Fig.2: Integration Roadmap Approach: Selection based on maximum return on investment Areas where significant improvement could be achieved from deployment Feasibility study Risk analysis Integrated and scalable architecture Real time visibility of data across various applications Consistent and accurate data across the enterprise applications Reduced manual data entry Increased productivity Focus on Business Reduced Inventory Levels Reduced IT Support and associated cost Decision 2: The Cypress business architecture team phased out implementation to gradually adopt the end objective rather than the big bang approach. Product data was split across multiple systems. The Business Architecture team identified the importance of centralizing the Product data in Oracle Product Lifecycle Management as the first step. The goal of this project was to centralize input into Oracle and have Oracle back populate the other system tables. COLLABORATE 11 OAUG Forum 3
Approach: Phased out approach Streamlined information processes and master data quality Maximize the use of master data in downstream applications High rewards- significant potential upside by improvement in product data and process management Reduce system and business continuity risk Low risk of bad data in the downstream implementations 2. Software selection Challenge - Product Master Data Management Software selection is a painstaking exercise; however, Cypress proprietary ROI calculation matrix approach helped in the selection of the software. The Business Architecture and IT teams selected Oracle Product Lifecycle Management (PLM) 11.5.10 after reviewing various software packages. Approach: Cypress ROI calculation matrix. Key factors influencing the decisions for selecting Oracle PLM by the ROI Calculator were: Oracle PLM was completely integrated with EBS which cut the total cost of ownership drastically Functionality to meet the current and future requirements Intellectual Property protection Productivity Improvement Limiting the technology skills required by Cypress personnel by moving to a single but scalable EBS environment Improved data and process management Hierarchical cataloging Attribution capabilities Granular security Lifecycle based change controls Workflow based collaboration Productivity improvement 3. Oracle PLM Implementation Challenges NPI and ECO Processes Due to the complex nature of Cypress systems, the part setup process was a time consuming activity and thus, users who need parts setup had to plan accordingly (10-12 days). Processing time was drastically cut to less than 1 day in Oracle PLM by using well defined workflow based New Product Introduction (NPI) processes and Engineering Change Order (ECO) processes. The Product Management, Production Support, Finance, Marketing and Quality teams responded to workflow based notifications during part setup and changes. COLLABORATE 11 OAUG Forum 4
Approach: Streamlined Process design by Business Architecture Team Change control based on Roles as well as product lifecycle phases Workflow based approval and notification routing New product setup time reduced to less than 1 day Automated workflow routings Strong collaboration Audit trail of changes Significant cost savings 4. Oracle PLM Implementation Challenges - Data Synchronization With Oracle PLM implementation, all the products/parts were created in Oracle using workflow driven processes. Records updates from Oracle needed to be back populated into 6 different legacy systems (58 different file formats), Siebel CRM, and Cypress websites. Moreover, the scope of synchronization was a moving target due to the constant addition of part attributes in Oracle PLM and their interfacing to the legacy systems. The implementation team leveraged the inbuilt workflow based post attribute change events to synchronize multiple systems real-time whenever any product record changed. Approach: Leveraged workflow event functionality Dynamic PLSQL to auto-generate the synchronization extract code in case of configuration changes System highly automated Designed for scalability with minimum intervention Use of out-of-the-box functionality Approach reduced the time and cost of solutions implementation drastically 5. Oracle PLM Implementation Challenges- Data Governance and Standardization Data governance is the heart of MDM implementations and the initiative was well thought out to define and streamline: Policies, guidelines and business rules Accountabilities Enforcement of policies Due to the complex part data to satisfy the needs of multiple manufacturing, planning and other business systems, data governance was a significant challenge. The Cypress Business Architecture team outlined the data governance strategy well and laid out the execution plan to be successful. Approach: Design data policies for data quality and information security Data Steward team Adopted "Call to Action" model where team members were asked to address urgent issues COLLABORATE 11 OAUG Forum 5
Single definition of product data attributes Improved data quality o Validated fields prevented bad data entry o Easy to track inconsistent data with advanced search and display capabilities Post attribute change propagation Update of one field could be propagated to perform mass changes and to maintain data consistencies Pre-attribute change validation Validation rules Secured role based data access 6. Post Go-Live Continuous Improvement A number of continuous improvement initiatives were outlined and executed. Oracle PLM with unlimited attribute capabilities was able to expand simply by means of simple configurations to accommodate Logistics and other business functional area related part attributes. Additionally, business events were leveraged further to perform a number of other automation tasks including: Integration and automatic update of Data Mart Product Pricing and Cost Update automation Standardized part descriptions generation Lead time attributes update Phase out and obsolescence Data Quality improvement Bills of Material (BOM) data management Approach: Process design and approval from cross-functional teams Documentation and communication of who will be impacted by change Automated processes and productivity gains Cost reduction Data quality improvements Readiness of Shop Floor Management system deployment Final Oracle PLM System Architecture Following are the salient features of the final architecture with the Oracle PLM system in place as shown in Fig.3: Streamlined workflow based NPI and ECO processes Oracle PLM as Hub Business event based data synchronization to other systems Business Rules and Policies for product data governance Secured role based access COLLABORATE 11 OAUG Forum 6
Fig.3: Oracle PLM System as Information Hub Lessons learned The following aspects of the project were considered for scope of improvement for the ongoing improvements: Consistent Data Definition To manage the data quality, a consistent definition of the data and complete understanding of how and where it is used is a MUST. During the latter part of the implementation, issues emerged where some data elements were being used by different functional areas for different purposes. Since some of the usage was not system driven, the IT team was not completely aware of such details. A data steward team formation early in the project and their interaction with each cross-functional team can nip such issues as they arise. Issues appearing later in the project could have resulted in costly redesign activities and potential threat to project timeline. Start data cleansing early The data quality cleansing effort MUST start early. Part of the data cleanup activity in the source system was left for the last couple of months before the project go-live. This resulted in disparities during data migration. Therefore, data quality must be taken up as a separate effort prior to any MDM implementation. A dedicated effort is required to ensure timely completion of data quality activity. Avoid garbage in garbage out What data to migrate to a new system? It is critical that the MDM initiative have not only sufficient representation but also active participation from all business areas. Inaccurate assumptions made in the MDM design due to lack of input from COLLABORATE 11 OAUG Forum 7
business users in any specific functional area can be costly. All functional area representatives must review and take full responsibility for the MDM requirements. If necessary, a formal signoff process must be put in place to ensure accountability. Team and skills Ultimately, any project can t be successful without a winning team in place. Choosing the right consultants and execution team were critical and ensured the success of this project. Management must pay close attention to the skill set of internal resources who must be ready to take control of production support once the consulting team disengages. Management must make the commitment to hire and train the ideal people who bring the right technical skills and process understanding to not only provide post production support but also help in expediting the continuous improvement process along the way. Critical Success factors To sum up, the following critical factors were executed well at Cypress for the MDM initiative to make it successful: Management driven Master Data Management Strategy Management commitment is a MUST MDM Vendor Selection Selection to minimize overall cost of ownership with robust & scalable system; Minimize integration points; minimize technologies for a lean and efficient IT Workforce Self Service Architecture with Scalable Data Model Self-service reduces the burden on IT and empowers the business users; Important consideration criteria for Vendor selection Product Data Quality No MDM initiative can be successful without quality; start early; form data steward team/roles Data Governance Goal must include business rules and policies; implementation of policy is the key Integration/Synchronization Approach Use of business events/soa (Service Oriented architecture) approach to synchronize; select the right implementation partner with strong expertise Streamlined Workflows for Process Standardization Optimal use of the workflow to enhance productivity Ongoing Commitment to Data Quality Gradual controlled and reinforced Change Management Right implementation partner who has been there and done that Conclusion A company s survival depends on its products and services. Therefore, implementing a robust and scalable Product Data Management system in order to stay competitive is an obvious choice. MDM is a journey and not a destination. Constantly evolving business strategies to meet the market challenges call for a flexible MDM solution that can be adapted to meet the changing requirements. Whether a company starts the MDM effort with standardizing the marketing data or streamlining processes like new product introduction and change management, it makes good business sense to embark on the Product MDM bandwagon. Cypress has realized the significant gains from the MDM initiative and set a solid foundation for the future growth. The critical success factors defined in this case study are based on practical experience from the Cypress MDM initiative and can provide other companies with guidelines to succeed in their own MDM efforts. COLLABORATE 11 OAUG Forum 8