This week, we continue with our “Ask the Experts” series with a focus on master data. Understandably, master data is not the “sexiest” of topics and isn’t hyped on media headlines like AI and VR, but it is arguably one of the most critical areas to comprehend, define, and manage appropriately—particularly when embarking on a systems implementation. Here, Parker Avery experts Heidi Csencsits, The Parker Avery Group, and Marty Anderson provide their perspectives on the importance of proper handling of master data.
Q1. During system implementations, it is critical that the foundational data – master data – is clean, complete, accurate, and consistent. What are the top 2-3 roadblocks that implementation teams run into when attempting to get a company’s data to this state?
Heidi: Change management is often a roadblock. If a company has not had a governance model in place and data is classified inconsistently, when the organization moves to standard definitions there will be business resources who feel they’ve drawn the short straw on this (by not doing things “their way”)—and they’re not going to like it. At a recent project, the client implemented standards in their merchandise foundational, hierarchical data; in doing so, they removed an aggregate from their product hierarchy that was pervasive across the organization. When asking what attributes the business team wanted to plan, the first thing they said was that missing aggregate. A second challenge is when there are diverse sources of what can be considered as foundational data in multiple systems that use the data in different ways. Another Parker Avery client initially had four people typing the same content into four different applications—we’re human, we make mistakes, so this leads to inconsistency.
Lynne: Stakeholder engagement is critical to ensure that governance is a priority and data is considered an asset. Foundational data owners should be identified to steward the data. Data definition and cleansing are business-owned tasks that must be done in a collaborative and communicative manner to garner as much acceptance across the board as possible.
Marty: Key misses include: a lack of understanding of how the different types of data need to flow through each system “end-to-end,” which source is best or most accurate for each data type, and exclusion of important roles or business functions when reviewing final data needs.
Q2. What organizations and roles from the business are typically required to ensure data validity, and what specific skills or knowledge do they bring that ensure good master data?
Heidi: A neutral role, which has no vested interest in the assignment of data. Meaning, they don’t worry if the color is “Blue” or “Sky Blue” as long as data is consistently defined. This role should also have a general understanding of retail merchandising. Understanding the “why” behind someone asking for a new attribute and not fighting every step of the process is a way to ensure the data truly meets the needs of the business.
Lynne: Subject matter experts (SMEs) who use the data every day to drive business activities. Each functional area must be represented in order to ensure data validity—they know the data necessary to manage their business. For example, both sales operations and finance departments require the customer data to be “clean.” Sales operations needs data to enable selling, whereas finance requires data for determining payment terms and credit limits, but they both need all of the vendor foundational data.
Marty: Usually anyone who manages a particular process or system that requires basic transactional data to function properly. For example, admins and process owners from the following:
Additional roles include merchandising data support, data stewards and process owners (MDM, PIM, etc.), business intelligence/reporting leads, and IT roles such as database architects, business analysts, and ETL specialists. Further, if you are looking to tie other areas of the business in with additional MDM modules, you may also want to explore data needs for human resources and marketing, but during data standardization and cleansing activities, we advise including these roles in separate breakout meetings.
Q3. There are many “best practices” for master data management (MDM). What MDM best practices have the most impact on ensuring master data continues to be appropriately managed after the system goes live and the implementation teams disband?
Heidi: Make sure all fields are clearly defined and training for new team members is in place. Many retail organizations do not have standard, structured onboarding training for new hires in their corporate offices. Continuing to rely on “tribal” knowledge transfer can easily cause incorrect data setup in the future. Also, having refresher courses (e.g. brown bag lunch sessions) that include friendly reminders of the correct process and setup procedures can help existing team members continue the appropriate setup and maintenance. But here’s a word of advice (based on experience.): if you’re going to do these, make sure the presenter is vibrant and charismatic, so the audience remains engaged in the topic. Also, when setting up the system, if at all possible, be sure to limit the number of fields that are completely free-form text—these are a gateway to lack of standardization across the organization.
Lynne: It is very easy to implement a different data set for each system that is deployed within an organization. The idea of a single version of the truth stored in the ERP or core merchandising system and shared to the rest of the organization is critical. The entire organization should be working from the same data sets which are all part of the company’s data assets. The existence of a master data organization or “merchandise information office” is also important in ensuring that the foundational data is accurately defined and maintained.
Marty: There are several that I consider high priorities:
Clearly defined data maps that show how data flows through end-to-end systems from creation to archive/purge.
A well-defined and consistently executed archive/purge process.
Establishment of a data steward role that will manage the ongoing data governance processes to ensure that data integrity is preserved, and new data needs, requests, and issues are appropriately handled.
Q4. On the flip side, what are the top 2-3 challenges companies experience when managing their master data post system go-live?
Heidi: Continuing to use and maintain legacy systems that have not adapted to the new master data results in having one foot in the old world, one in the new world, and a pretty good chance that data will not match between the two systems (especially at aggregate levels). This environment also hinders adoption since the business community will still be using older terms as opposed to the new terms.
Lynne: Loss of executive sponsorship and therefore enthusiasm for master data governance enables the return to the “same old” work methods that create disparate data, multiple versions of the truth, and inconsistent data across the organization.
Marty: This goes back to implementing some best practices and include system performance degradation due to lack of consistent archive/purge process and data duplication, proliferation, and inaccuracies as a result of poorly defined data governance standards and/or lack of data stewards.
Q5. Do you recommend other specific opportunities/projects that necessitate master data cleanup, or does it always need to work hand-in-hand with a systems implementation? Is it ever a standalone initiative?
Heidi: It actually can be a standalone initiative. If no guardrails have been in place, data definition can be highly inconsistent, often resulting in the business relying on faulty information for decision-making. A thorough data cleansing can provide consistency to an organization before a systems implementation takes place. It is time consuming (and can be painful) but will benefit in the transition to a new environment, as well as continuing use of existing systems.
Lynne: Yes, it can be a standalone activity used to prepare for future initiatives such as an ERP implementation, planning and allocation implementation, or BI/analytics implementation. However, generally speaking, MDM by itself, does not drive revenue or increase efficiencies enough without a system implementation that delivers ROI. Regardless, it is important to establish master data governance and management to ensure ongoing focus and progress across the organization, channels, and functional areas. Using the same set of foundational data speeds decision-making and the ability to quickly address business challenges.
Marty: Any new system upgrade/implementation should start with a data review whether or not it is an MDM implementation. Consistent archive/purge processes should be executed at least annually if not quarterly. Many times, report standardization (core reporting) necessitates a data review/cleanup as reporting and analysis becomes more and more critical to run a successful and profitable business.