Карта сайта
Версия для печати

Пять рекомендаций при формировании успешного проекта MDM

30 октября 2013 За последние 12 лет понятие «управление мастер-данными» (Master Data Management) эволюционировало с расплывчатой абстракции в хорошо известную и прикладную дисциплину, практикующуюся во многих организациях по всему миру. Читайте рекомендации ведущих экспертов, какие аспекты нужно учесть при формировании плана проекта по внедрению мастер-данных. (Материал опубликован на английском языке)
Since the emergence of MDM as a popular discipline, hundreds of articles and white papers have been published that call out critical considerations for organizations that embark on an MDM journey. There are slight variations in the recommendations of industry experts, but most fall along the lines of:
  1. Make sure you secure strong business sponsorship.
  2. Make sure a clear business case is defined.
  3. Do not treat MDM as a technology-only project.
  4. Establish a data governance program with clear roles and responsibilities.
  5. Remember that MDM is a journey, not a destination.
While the above recommendations are certainly valid and important guidelines to follow, we have observed a number of other key considerations that our experience shows also have a significant impact on the success of an MDM implementation project. These additional points can take an MDM project from a mere milestone to a lasting and impactful cornerstone of an organization’s information technology and business process infrastructure.

1. Carefully find the appropriate balance between master data control and business process impact.
Find the right balance and be prepared to accept that central execution of master data maintenance may not always be acceptable from a larger business process perspective. To be in complete control of the maintenance process of, for instance, customer data, it is recommended that the data is managed in a central repository from where the data is distributed to ERP and other applications where the data can then be used. An architecture like this is known as “central maintenance” or “centralized MDM.” After implementing this centralized architecture, users who enter orders in a downstream ERP system will no longer have access to create customers directly as they may have been accustomed to. From their perspective, having to go through a centrally controlled process to request a new customer record may be considered bureaucratic or a hindrance to effective order entry for instance. Finding the appropriate balance between central control of master data and the need for speed and flexibility will require weighing the conflicting interests, considering allowing additional users groups to create customer data and considering if some fields of customer data can be managed locally, while other fields are managed centrally. Do note that the right approach will differ across organizations and across master data domains.

2. Be mindful of conflicting business requirements.

An MDM solution is often aimed at supporting multiple business functions, which sometimes have conflicting requirements. These can range from how data should be structured and how certain fields are used, to who should be allowed to edit which fields. With customer data, for example, what may be a great solution for finance may have a significantly negative impact on order management or vice versa. It is not uncommon either to come across requirements from a specific user or business unit that may undermine the greater objective of MDM for the organization as a whole.

3. Define detailed use cases and process flows for data maintenance and integration.

An MDM solution typically interacts with multiple other systems through tightly integrated business processes. Detailed design and visualization of how different user roles are responsible for each step across master data maintenance processes are essential to confirm that the design and all impacts for how master data is captured and synchronized across other systems are thoroughly understood. After each test cycle, revisiting these processes and use cases is critical because, as the project progresses, the requirements and constraints evolve and further increase the complexity.

4. You need to have dedicated tactical business resources on your team. 

The business resources on the project need to truly understand the business and help make the right design decisions for the organization. Too often, MDM teams are staffed primarily with IT resources and just a few businesspeople — perhaps even some new to the organization with limited understanding of how the business works. This can be a profound mistake. An MDM project needs resources who understand how the MDM solution will impact various business functions and can help test the solution, negotiate compromises on conflicting requirements and see everything through beyond go live. Examples of roles that are may make be good candidates as business resources on an MDM project include financial analyst, process engineer, quality assurance specialist and product manager.

5. Plan and execute multiple test cycles with thorough regression testing.

From an IT solution perspective, MDM projects are often high-complexity projects with many integration points between business-critical applications. As a result, multiple test cycles and thorough regression testing are a necessity and should include business users across all of the impacted business units from the earliest test phase possible. Each cycle should be prefaced with a core team review of the requirements, constraints, processes and use cases so repeat tests of outdated scenarios do not hide critical defects from discovery and resolution. Again, the dedicated business resources play an essential part in making sure the various business testers are focused on the most important regression use cases and where the new MDM solution has the most impact.

While the first five common recommendations are an essential part of an MDM project, the five additional recommendations outlined here are equally important for an effective MDM implementation project.