Manager – Synechron Business Consulting, Hong Kong
For organizations to be in control of their enterprise data, not only today but also 10 years from now, a new vision on data management is key. When data management is embedded into change programs as an intrinsic value, organizations are set to be futureproof and in control.
Data management and change programs: common challenges
When multiple change programs are simultaneously impacting the context of enterprise data on a daily basis, keeping track of your data can prove to be a complex exercise. In practice, organizations face a significant disconnect between data owners and data consumers. Data (model) changes are often not communicated effectively and timely or do not reach the data consumers at all. This complexity increases exponentially with the number of change programs.
With legacy systems being decommissioned and replaced by target enterprise solutions, change program efficiency and progress are often hindered by poor data lineage documentation and limited data traceability, outdated data dictionaries and other shortcoming data management practices. With changes in both data definitions and data lineage being poorly agreed on, documented and communicated over time, the effect of changes on the integral reporting chain is not overseen. Timely delivery of change programs is therefore endangered.
Hence, for organizations to fully reap the benefits of being data driven and to be in control of their data, not only today but also in the long run a new and futureproof data management strategy is needed. Our approach aims to tackle the beforementioned issues by addressing three pillars: centralized data governance strategy, decentralized embodiment and innovative tooling.
Facilitating change: centralized data governance versus decentralized data management initiatives
Data governance and data management initiatives are often perceived as top-down measures to be implemented ex post. In our approach for data management, one could wonder, is there any room for centralization and top-down communication?
Indeed, data governance strategy and underlying principles are to be set centrally and flow into the organization unilaterally. Having a clearly defined and agreed upon framework set from a central unit such as the Data Management Office makes perfect sense: change programs tend to impact the entire reporting chain rather than only the functional domain in question. Hence, a common way of working which is supported by all stakeholders decreases the likelihood of delays.
In example, pre-defined agreements on 1) the involvement and 2) the ownership of different teams in the functional acceptance test process enhances the chance of success.
Furthermore, setting uniform and clear principles regarding data documentation such as data dictionaries and data lineage contributes to a common understanding and therewith to the likelihood of adequate program delivery. Too often, knowledge concerning (legacy) systems is concentrated in persons rather than in clearly and uniformly structured functional documentation.
But how do data governance principles translate to data management operational practices?
When data management measures are embedded into change programs from the get-go, operational teams are forced to think about and embed data management in their daily practices. On top of that, when change programs include the aforementioned principles intrinsically, data management efforts are embraced by the entire population. Concluding: even though principles are defined centrally and constitute the data governance framework, ownership of the data management measures and other operational implications are owned by change programs situated in the federated domains of the organization.
Embedding data management into change programs
For organizations to be truly data driven, data management efforts should be perceived as value adding rather than a top-down measure that is forced upon the organization. Therefore, successful data governance is to go hand in hand with data management principles being adopted by (data) domains change programs and their operational teams. With successful adoption, day-to-day data management activities are owned and executed by and within the domains, fully enabling a federated governance approach.
Depending on the size of the change, a change program could decide to include a dedicated data management lead or workstream. This lead would be responsible for ensuring that business requirement documentation, data definitions and data lineage meet quality standards expectations of stakeholders in the reporting chain.
In practice, documenting change is perceived as an ex-post activity. Too often, the party responsible for keeping the documentation up to date is not informed in time or not informed at all when business requirements change, or new functionality is implemented. This hinders program delivery and frustrates stakeholders in the (reporting) chain.
When clear data management principles are embedded into the definition of done at the level of project teams, change documentation efforts are 1) distributed over a larger and better equipped workforce and 2) executed during the change process rather than after. In this context, the definition of done is meant as the set of deliverables that is owned by the project team.
Project teams are responsible for documenting and communicating changes (i.e. ‘push’) to the accountable owner rather than the owner having to request said information after the fact (i.e. ‘pull’). Furthermore, assessing where and how to implement data quality (DQ) checks prior to finalizing the design contributes to the effectiveness of said checks. This is due to involving the right people (individuals involved with the actual change) at the right time (timing element). Also, proper implementation of DQ checks ensures timely alignment with data owners and consumers downstream in the chain. Furthermore, well designed and effectively implemented DQ checks enhance the business value of the data sourced for reports while contributing to BCBS 239 regulations.
Innovative tooling as a driver for efficient and compliant data management
Day to day data management activities can be time consuming. As new technologies arise, organizations are looking for ways and tools to be cost-efficient in their data management organization. Using (centralized) tooling for data management offers a number of practical benefits and contributes to 1) compliance, 2) being in control of data and 3) up-to-date data models. For example, when data models are administered and maintained in a uniform way using common central tooling, change and decommissioning programs are more likely to deliver on time. Also, knowledge retention and efficient onboarding of new resources is ensured.
Furthermore, such tooling can act as a bridge between the logical and physical world: not only can logical data models be administered and maintained, actual changes to databases can be triggered automatically within the same tool.
An example of innovation in the data management landscape is automated data lineage. Even though there are limitations, automated data lineage tooling can help to unravel data flows and intermediate transformations in a time efficient way. Automated data lineage is especially helpful in environments where futureproof technology is used, such as data virtualization.
Organizations are challenged by the increasing number of change programs that are impacting the context of enterprise data on a daily basis. In order to be in control of data, not only today, but also 10 years from now, a new take on data management strategy is essential. Three fundamentals are underlying to this strategy:
1) Centrally set data governance principles;
2) Embedding of data management practices by (data) domains and their change teams;
3) Uniform and innovative tooling.
A clear and agreed upon data governance framework improves consistency and cohesion between apparently separate change programs. Working in a uniform way results in a more efficient staff onboarding process, which tends to be troublesome in many change programs.
Furthermore, when data management practices are embodied by project teams, these teams are responsible for accurately documenting changes (business requirements, data definitions, interfaces) and timely informing the data owners as well as stakeholders downstream the reporting chain. This increases the likelihood of timely program delivery. Also, with data management ownership embedded into the definition of done, the hand-over from projects to business-as-usual teams is significantly enhanced – breaking down silos by design. Hence, well implemented data management practices minimize the effect of changes on the integral reporting chain.
To facilitate the abovementioned steps, teams should be facilitated by innovative tooling that makes life easier for both project teams as well as (risk) managing stakeholders. Centralized and digital tooling to capture data (model) changes, rather than using end user computing tools like Excel, contributes to being in control of your enterprise data.
Sparked your interest?
At Synechron we are more than happy to explain how your organization can benefit from our experience in assessing and (re)designing processes. Over the years we have helped numerous clients with the change to more resilient reporting processes and tighter data control.
About the author
Manager – Synechron Business Consulting