This is the second article in a two-part series on data-centric projects and our Data-Centric Development Life Cycle (DCLC) methodology. (Read part one.) DCLC is an iterative methodology for data-centric projects that brings data problems and opportunities to the forefront before development begins — rather than during testing or after production implementation.
Data-Centric Development, Ready or Not?
How do organizations make the shift from process-centric to data-centric development approaches, like DCLC? It comes down to readiness — both at the organizational and project levels. There are certain capabilities that need to be in place to utilize DCLC — and the first is a genuine willingness to try the new methodology. Also important: A team with the appropriate skills needed for data-centric development work, which not all organizations have.
Another factor for data-centric project readiness is for the business to think in data-centric terms about the capabilities it needs, rather than remaining too high-level. For instance, source data analysis is performed on a data-centric project, and some companies may consider this just part of “analysis.” And, because they already have business analysts, they believe they already have the capability.
With DCLC, source data analysis is broken into several capabilities, including conceptual modeling, data quality assessment and data profiling. If an organization doesn’t have these data-centric capabilities, it needs a plan to acquire them. The plan should focus on identifying and describing the detailed data-centric skills that are needed. (For organizations that are new to data-centricity, a consultant may be needed to help do this.)
Once a company is generally ready to adopt the DCLC project methodology, it can then turn its attention to project-specific readiness. For example, the project may be in a business area where users lack a sufficient understanding of the overall data domain to be able to begin the project. (This is often seen in global projects where regions that have never worked together now have to rationalize their data.) In such cases, a significant amount of project time can get consumed with users just trying to understanding each other’s viewpoints.Readiness — both at the organizational and project level — will ensure success when the DCLC methodology is deployed on data-centric projects. Click To Tweet
Bringing a Data-Centric Approach to Projects
Organizations that adopt DCLC need to have a formal mechanism to deploy it on projects. This typically involves working with other business units that have responsibility for managing development projects. Often, it’s the formal project management office (PMO) that is staffed with project managers (PMs) for development projects. In other companies, PMs are appointed but aren’t part of a centralized team. With either structure, it is important to be able to introduce the DCLC methodology to PMs to have it deployed on data-centric projects.
In the case of a formal PMO, it is often the Data Governance (DG) team or a group with a similar interest in data that brings the DCLC methodology to the attention of the PMO. A first step is to get the PMO to understand what a data-centric project is, how it differs from process-centric development, and what the risks are of using a process-centric approach on a data-centric project. A good way to bring DCLC into relevant projects is for DG to be engaged in one or more of the pre-approval phases of a project, e.g., having a seat on the Architectural Review Board. DG can then help determine if a project is data-centric and a candidate for DCLC. The long-term goal is to get the PMO to formally adopt DCLC and be the group to determine which projects it is used with.
If your company structure uses decentralized PMs, you will need to introduce DCLC to them individually. Sometimes, the person may be a technical lead rather than a dedicated PM. In this situation, you should approach the PM’s business unit with a request to provide DCLC training. (Sidenote: If a PM isn’t a part of a PMO, DG may need to own a data-centric process initiative, rather than the PMO.)
Blending Data-Centricity With Process-Centricity
While the world has become increasingly data-centric, there continues to be strong demand for process automation. This means that the more traditional methodologies, Waterfall (a.k.a. Software Development Life Cycle or SDLC) and Agile, are still relevant. But we’re seeing more awareness that even process-centric projects have aspects of data-centricity to them — and certain projects will benefit from a blended approach.
Consider this example: “Increase the organization’s knowledge of data” is a deliverable in your process-centric project. The need is driven by your data scientists who need to analyze and understand transactional data. But the transactional application, being process-centric, isn’t “concerned” about the data scientists’ viewpoints. By adding elements of DCLC, this process-centric project can provide better outcomes.
Conversely, even data-centric projects have process-centric aspects. One example is data movement, where many elements of this data-centric project can be carried out using a process-focused methodology. Typically, data-centric aspects of the project are ignored. Project teams who understand DCLC and recognize its benefits can incorporate it, where needed.
It’s Time for a Data-Centric Approach
Process-centricity is deeply embedded in most organizations, and it can be hard to know when it’s the right time to make a change to a data-centric approach. For the governance and PMs who will likely introduce DCLC, they must do so thoughtfully. Their first steps should be centered on building awareness and socializing DCLC’s concepts.
Another early step should be to review past data-centric projects in the organization. Strangely (or perhaps not so strangely) there is a reluctance in many companies to do “postmortems” of projects that failed to deliver. Yet this is a healthy thing to do! It needs to be done in a professional manner, of course — otherwise it will be IT versus the business with the old refrain of “You didn’t provide the right requirements,” countered by “You didn’t deliver.” If you happen to find that all previous development projects were successful, there may be no need for DCLC. However, this is unlikely as you will certainly find several areas of improvement.
The next step is to gain consensus that DCLC’s concepts can improve future development projects. While this requires individuals to have some knowledge of DG and data management, this shouldn’t be a big obstacle with more and more professionals becoming data-savvy.
At this point, it’s time for your organization to get started with the DCLC methodology. Start with a small data-centric project, or carve out a portion of a larger project to test DCLC. As you bring the various stakeholders into the data-centric process, collectively establish measurement criteria that is meaningful to the group.
Once you find success with one data-centric development project, it will become easier to apply DCLC to other projects. Over time, this methodology will become institutionalized across your organization — and your data-centric initiatives will be stronger for it.
Interested in learning more about DCLC?
Ready to talk with us about how DCLC can help your organization?
Article contributed by Malcolm Chisholm. He brings more than 25 years’ experience in data management, having worked in a variety of sectors including finance, insurance, manufacturing, government, defense and intelligence, pharmaceuticals and retail. Malcolm’s deep experience spans specializations in data governance, master/reference data management, metadata engineering, business rules management/execution, data architecture and design, and the organization of enterprise information management.