Master Data Management and Universal Customer Master Overview 1
MDM: Master Data Management Large companies often have IT systems that are used by diverse business functions (e.g., finance, sales, R&D, etc.) and span across multiple countries. These diverse systems usually need to share key data that is relevant to the parent company (e.g., products, customers, and suppliers). It is critical for the company to consistently use these shared data elements through various IT systems. In order to bring this consistency, we need to define the system of record, or the master system which will have the real-time true picture of the entity it is mastering. This not only improves the various operations of the client but also provides the basis for good input for BI reports / analytics. 2
Same Person Many Caps! Take the simple example of a automotive firm, say ABC Corp, having three different applications in it s enterprise to handle Marketing, Sales and Service. Now suppose that Joe is a prospective client and a marketing agent contacts him about a great deal that ABC Corp is offering on it s latest car model. A week later Joe goes to a dealer of ABC Corp and buys the same car. And a month later, he calls up to request his first free car service. Without mastering in place for it s customers, ABC Corp wouldn t know that these three Joes, in it s three different applications, are actually one and the same. ABC Corp. Marketing Customer #: Red Cap Joe Service Customer #: Green Cap Joe Sales Customer#: Blue Cap Joe 3
WITHOUT A CUSTOMER MASTER : Marketing App Cust #: RedCapJoe Name: Joe Smith Addr: 21 South St, SFO Income: $ xxx,xxx,xxx p.a. Hobbies: Reading, Music Sales App Cust Id: BlueCapJoe Name: Joey Smith Addr: 21 South St, SFO, 41181 Quote: 10 IBM PCs Order: 5 HP Laptops Service App Cust #: GreenCapJoe Name: Jo Smith Addr: 21 South St, CA SR: Printer not working. Service App Cust#: BlackCapJoe Name: Joe Smith Addr: 21 South St, SFO SR: LAN not working. WITH A CUSTOMER MASTER (E.G SIEBEL UCM): Service Cust #: GreenCapJoe UUID: JSMITH Name: Jo Smith Addr: 21 South St, CA SR1: Printer not working. SR2: LAN not working Marketing Cust #: RedCapJoe UUID: JSMITH Name: Joe Smith Addr: 21 South Street, San Francisco, CA, 41181 Income: $ 5,000,000 p.a. Hobbies: Reading, Music Sales Cust Id: BlueCapJoe UUID: JSMITH Name: Joey Smith Addr: 21 South Street, San Francisco, CA, 41181 Quote: 10 IBM PCs Order: 5 HP Laptops 4
WITHOUT A CUSTOMER MASTER : Applications out-of-synch wrt Customer Attributes Poor quality (un-usable) address data Duplicate records in an application leading to poor quality customer experience Not possible to correlate data in different applications to get a 360 view of the customer WITH A CUSTOMER MASTER (E.G SIEBEL UCM): All applications in synch w.r.t. core customer attributes Universally Unique Identifier (UUID) is published across applications enabling direct application synchronizing Good quality usable data (address etc.) Duplicate records in an application merged to improve customer experience 360 View of the customer using XRef through UCM Better analytics reporting possible 5
MDM: Master Data Management (Contd.) The most common functions of any MDM applications: Data Synchronization (acquisition & publish) Data Cleansing De-duplication Security / Privacy / Access Control Stewardship 6
MDM: Data Synchronization Comprises of receiving updates from other applications in the system and applying them to the base tables, and then publishing out the final Best Version back to all subscribing systems. 7
MDM: Data Cleansing Different Applications in an enterprise can have different system designs / limitations (e.g.: Address 1 can only be 20 characters) and sometimes data entry operators can use their own standards of entering data, to send the same data in multiple formats. MDM needs to collect this data and apply certain standardization rules to it. 8
MDM: Data De-Duplication Each application has it s own record identifier (ROW_ID, Customer#) by which it can uniquely locate each record. When the data from each application reaches MDM, it needs to identify the same record, resolve the duplication, and still preserve the external identifiers for publishing back future updates to this record. 9
MDM: Security / Access Control Access Control in MDM application is more in terms of which application has access to perform what all operations on which of the mastered entities. In a typical large enterprise with several applications, each application performs a certain role on certain entities (insert, update, query). In MDM, we need to register all these systems which affect any of our mastered entities, and give them rights to perform only those operations on the data as per their role in the overall IT strategy of the organization. 10
MDM: (Data) Stewardship An MDM application mostly does it s job in the background and publishes updates to all the end-user applications. Not many users will actually log in to it to see the data. Thus, from a user role point of view, the most important role in an MDM implementation is that of the Data Steward (or Architect or Administrator). He/She is the one who controls the flow, quality and security of the mastered data. A few others users might access the application to keep a check on the data. 11
UCM : What is it? So, what is Siebel UCM? A Siebel application which is the central master database/application for storing any and all customer information. In a typical large enterprise, you have hundreds of various proprietary and legacy applications which deal with different parts of the customer information, some deal with service data, some with sales data, some others purely with demographics and contact data. UCM has the capability to synchronize and bring together all this data to give a single unified view of the customer as a whole. From a Siebel application point of view, it s just another application like Siebel Sales or Siebel Call Center, built up on the same Data, Business and UI model. 12
UCM : What is it? (Continued) What are the similarities with other Siebel Applications? As told before, it uses the same 3-tier model used by all Siebel applications, the data layer, business layer and the UI layer. It uses the same party model, and most of the tables are also the same, S_CONTACT, S_ORG_EXT etc It makes use of the same Siebel EAI/EIM architecture. So how does it differ from other Siebel Applications? It extends the party data model a bit to include Source-Data-History tables (explained in the coming slides) It typically has very few actual UI users, mostly data stewards, and relies more on Siebel EAI / EIM for data acquisition / publish 13
UCM : What it does? (Functionality) Receives the transactions from external systems and applies them to it s base tables and publishes the updates back to all the systems. Publish frequency - real-time, batch or event-based (e.g.:on-update of Contact ) Siebel UCM Data Quality - in built component for data de-duplication Survivorship Rules - based on the defined confidence level for each system Easy setup for Auto/Manual merge Features to Un-Merge already merged records later-on if desired. Best of breed matching technology embedded from SSA Matching on name, address, and other attributes configurable for other attributes Configurable for real time and batch Suspect duplicates quarantined for review by the Data Steward. 14
UCM : What it does? (Continued) Shown below is a typical usage scenario for Siebel UCM. The UCM UI is used mainly for Administration, Access Control and Data Quality Administration. The Data Acquisition / Publish from / to other Siebel ebusiness Applications or other legacy applications in the enterprise happens through the Siebel EAI/EIM layer. 15
UCM : Deployment Models There are two deployment models for Siebel Master Data Applications: 1. Co-Located Model: In this type of a deployment, Siebel MDM application shares the same database and Enterprise Server as any other Siebel ebusiness Application. This deployment is used mainly for smaller enterprises. 2. Stand Alone Model: This is the recommended, and mostly used deployment model in which the Siebel Master is deployed on a separate box having it s own database, separated from other Siebel ebusiness Application. This is the preferred approach in larger enterprises where the volume of transactions is large as well as the integrating systems are many. (See Below) 16
UCM : How it does what it does? Typical data flow in UCM would be as follows: External System registration, preferences and setup Receive Inserts / Updates from external systems, cleansematch-merge them to produce best version record Apply the best version record to the UCM base tables. Create system audit trail (history) of previous version of each modified record Maintain External ID reference Publish the modified records back to the systems which have subscribed for receiving updates 17
UCM : How it does what it does? (Contd.) 1) System Registration & Privileges An MDM application synchronizes other applications in the enterprise. Hence, it needs to know which applications would be sending it inserts / updates while which ones would lookup information from it. UCM has screens where you can register these applications and give them privileges (insert, update, query, delete) and define publish frequencies (real-time, daily) 2) Survivorship Rules: With updates coming from different applications, Survivorship Rules define field level permissions for which systems data supersedes other systems. E.g. An email id received from the J2EE Web Application cannot be overwritten by the email id coming from a legacy mainframe system. 18
UCM : How it does what it does? (Contd.) 3) SDH: Source-Data-History Tables (also called UCM Staging Tables): These tables perform two functions: a) They act as a staging area which is required when records from all the interfaces will be stored temporarily, and checked for system privileges, survivorship rules are applied to them here, as well as data quality rules are enforced on them (cleansing and de-duplication). Once all this is done, the final best version of each of these records is placed in the base table holding your mastered record. b) The other function they perform is that of maintaining the System Audit Trail, recording which system sent what data when, and what was the previous values of the master data record in the base table before it got updated. Some of the different Type (Field: UCM Object Type Code ) of data stored in these tables include: History : These are the previous best version records in the base table before they got updated Source : These are the records which have been decided as the best version record and applied to the base tables. Merged : These are the victim base table records that got merged with other duplicate records in the base table. Unmerge : These are the records that got unmerged Cleansed : These are the cleansed records which were flagged as possible duplicates Batch : These are the records that have been recently loaded and are awaiting the BDM workflows to be run against them 19
The UCM Data Model (EIM, SDH and Data Quality/Governance) 20
Strengths of UCM Uses Siebel Architecture: Both for the application as well as EAI layers, UCM uses the industry leading Siebel Architecture, which helps connecting it to disparate systems with ease heterogeneous connectivity. Data Governance / Stewardship Model: The data steward model is extremely powerful and yet very simple to use for creating data governance rules. Data Matching: It has a built-in data de-duplication engine (SSA Matching Server) which can provide basic de-duplication functionality. Oracle suggests Siebel UCM to customers already having Siebel software in it s premises for Customer Mastering. To quote Forrester: The solution remains best-suited for service industries with global requirements and Siebel CRM. 21
Weaknesses of UCM Data Cleansing: Siebel Data Quality module has no built-in data cleansing engine without which the built-in de-duplication engine is not very effective. Thus, you need to purchase third party software to perform the cleansing functions. FirstLogic, Trillium and First Global are some of the software that perform this function. Firstlogic can be used only for ENU implementations (English Language) and has address verification capabilities only for North America (USA). Trillium reaches out to other countries also. Data Matching: The built-in SSA Matching server has limited capability when compared to third party software and cannot be configured much, has to be used as is. 22
THANK YOU 23