Developing Conceptual Data Models (CDM) and assessing their use and adherence to standards is an excellent way to advance the understanding of data in any organization
The Conceptual Data Model (CDM) is a technology and application neutral data model that is developed to help ensure that IT understands the business – before data stores and applications are designed and developed. The CDM is a diagram identifying real world concepts/objects/things (entities) and the relationships between these to gain, reflect, and document understanding of the business to:
- Foster semantic reconciliation
- Improve business / IT collaboration and expanded data management capabilities
- Serve as a framework for the development of high quality information and systems
- Improve project scoping (for data and application design)
Conceptual Data Models and Disruption
Why should CDM’s be disruptive? The goal for using a Conceptual Data Model is to challenge the status quo by delivering higher quality information and applications on-time, within budget. Additionally, a frequent result of developing Conceptual Data Models is the assistance the business receives with the insight into its data and relationships, through a process called “disruption.” Disruption means being “forced to think of another solution, a fresh idea, which in turn forces a positive reassessment of a company’s offering”
The Conceptual Data Model (aka Semantic Model, Business Information Model, etc.) usually is the first data model developed in a phased modeling approach, but all too often it is not performed due to factors such as lack of modeling experience and training, time pressures, emphasis on quantity versus quality, and mistaken notions about what a CDM is and its benefits.
Due to the outstanding benefits that can accrue when development of a CDM becomes a part of an organization’s data modeling standard it is important to have guidelines for the development of the CDM and a way to ensure that the CDM will be understood and utilized effectively. The Conceptual Data Model Assessment (CDMA) is a tool for this purpose.
Conceptual Data Model Assessment
The CDMA is comprised of a tool for measuring a CDM against objective criteria, along with an assessment methodology to ensure the assessment is performed effectively, that the results are properly delivered to the right audience, and used effectively to influence not just the quality of data and applications, but to help the business understand itself in different ways. The business is used to process models, organization charts, and other business models, but often does not have exposure to a model that models the business from a “data” perspective – a “what” not “how” model.
The Conceptual Data Model identifies the key business objects and how these objects relate to each other. The Conceptual Data Model can be disruptive because:
- Business concepts need to be named and defined. These names often have to be resolved through coordination between multiple business units, which leads to improved communication, understanding, standardization, and collaboration.
- As a result of the improved understanding of data (captured as business metadata and incorporated into or linked to the CDM via a metadata repository tool), data quality can be improved. If the organization cannot define the data element, how can anyone know whether the data is of high quality?
- The CDM helps to align IT to the business. By taking the time to understand the business via the CDM, the business feels it has been understood properly. The CDM is used to validate and correct understanding – before solutions are developed, thereby saving time, effort, and money. Misunderstandings can be cleared at the start of a project, rather than at the end during implementation.
- Information is a key asset – however, often it is not treated with the same care that financial, human, or other resources are managed. Conceptual Data Models are a key enabler of Data Governance to ensure that information assets are understood and utilized effectively. The Conceptual Data Model helps an organization identify and understand its information assets and helps to ensure that highly complex information systems are aligned with the business vision.
- Application design is affected by the data model, whether the application is an Enterprise Data Warehouse, ERP, SOA, OLTP, a Master Data Management hub, etc. For example, if a many-to-many (M:M) relationship is missed (e.g. a CDM was not developed), this can significantly affect application design. A common way to resolve a M:M relationship is with an associative entity which means that one more table and one more relationship has to be taken into account in: program code, screen forms, reports, SQL, interfaces, testing – significantly affecting project schedules and costs.
- More important are the costs associated with the data duplication or missing data that can occur because of an incorrectly modeled relationship. A large bank recorded all customer information at the account level. When there was more than one customer on the account critical information (e.g. identifying information, FICO score) about the secondary customers was not recorded. As a result, the bank did not have a true picture as to the number of customers it had and the true level of activity for its customers.
The measurement tool consists of a major checklist spreadsheet for scoring, and an associated chart for visualization of the results of the CDMA. The checklist is broken into the following six (6) categories:
- Descriptiveness
- Presentation / Clarity
- Model dissemination and usage
- Model metadata
- Lexicon / Terminology
- Model maintenance
Criteria for Assessing Conceptual Data Models
To derive the score for each question, the following criteria are evaluated (criterion 1 is a yes or no answer; criteria 2-4 are each given a score from 1 to 10).
- Defined as an enterprise standard? Each question represents a functional consideration (“consideration”) for delivering outstanding Conceptual Data Models, and this criterion determines whether the consideration is part of the modeling standards for the organization being assessed.
- Degree of tool capability to meet desired consideration? Not all modeling tools are created equal. This criterion helps determine whether the tool supports the functional consideration and the degree to which it does so.
- Degree of importance? How important is the consideration to the organization?
- Score. Objective score from 1 to 10 assigned to the question judging how well the Conceptual Data Modeling effort addresses the question.
The scores for all of the questions in a category are averaged to derive a category score (Combined Score). The combined score can be graphed to provide an overall picture of the Conceptual Data Modeling efforts to identify strengths and weaknesses.
Figure 1: Conceptual Data Model Assessment Results Example
The measurement tool is one part of the Conceptual Data Model Assessment. An accompanying methodology enables the CDMA to be conducted in a professional, effective manner, ensuring that Conceptual Data Models meet the needs of the enterprise or project. The assessment methodology includes considerations for:
- Information gathering (Interviews, documentation review)
- Architecture review
- Modeling standards review
- Modeling tool capability review
- Modeling notation review
- Presentation of findings
To be complete, the CDMA requires a summary report and executive presentation which identifies:
- CDMA definition and purpose
- CDMA approach
- Current modeling methodologies / standards in use
- Strengths of the Conceptual Data Models, and CDM practices
- CDMA areas for improvement
- CDMA recommendations, including:
- Specific modeling recommendations
- Ways to improve the use of existing modeling tools, or recommendations for new modeling tools
- Training required in data modeling, data architecture, etc.
- Data analysis techniques, e.g. data profiling
- Data Quality issues caused by poor or missing models
- Recommendations for improving business / IT alignment using CDM
- Standards to implement, why and how
- How to realize the benefits of Conceptual Data Modeling (business and IT)
- Project plan for implementing recommendations
The CDMA assessor will present and review findings to appropriate executives, management, and staff.
An example of a CDM assessment’s value can be shown by an insurance client that changed its practice of product development because of the CDM effort. The product subject area was modeled, but the model did not deliver a clear picture of the many types of products and classifications due to conflicting terminology and a lack of product development standards. After the assessment, an enterprise product taxonomy/categorization was developed as part of the CDM, became the standard for classifying products and helped the business to improve its understanding of products and their organization.
Conclusion
All organizations should develop a conceptual data model at the start of every project and as part of their enterprise data model programs. In addition, every conceptual data model should be assessed against objective criteria to determine if they were developed according to industry standards and processes, and to improve the practice throughout the organization.
