Large firms operate in complex environments and are therefore set up around specialised functions often referred to as silos. These silos group together specialised skills, knowledge and purpose. Separation of functions, and therefore duties, is also standard practice, ensuring those operating the business and those controlling the business are not the same people. In today’s increasingly regulated business environment Chinese walls are a business reality.
At the same time silos make collaboration across areas more difficult. Each separate area has its own particular business focus and perspective. Different areas can often also employ different terminologies, making it challenging to build a shared understanding. Different organisational reporting lines also mean there are fewer incentives to work with those outside your silo. In short silos making collaboration more challenging.
Although firms operate in silos, data flows across them and feeds those functional areas as blood in a living body. For example a firm’s clients might initially be onboarded by one function with several different functions subsequently selling them specialised products. Meanwhile support functions like the finance and risk departments will need to have a view of those clients and their transactions to execute their control measures.
The challenge for enterprise data management is to bridge, but not breach, these silos. The regulatory environment often means that these gaps can’t be physically crossed by staff or even by data. However data is in the unique position of being able to create connections between silos, in the process enabling collaboration across the whole firm.
Data integration efforts are often approached solely as a technical exercise. However, the key to bridging the silos and creating a lean data landscape is to first of all understand their specialised use of the shared data elements. How can we promote reuse of data if we do not have a view of who is using what? Unless we connect the data flow to its business usage, data will remain siloed and the firm’s overall effectiveness to create a more coherent and higher quality data asset hampered.
So what kind of understanding is required?
What is holding organisations back from becoming more joined up around data is a basic understanding of shared data items in terms of semantics, origination and business context:
Semantics: business understandable definition of the data item
Origination: where is the data item recorded and what is its provenance
Business context: what is its business context and relevance in terms of process, policy, regulation, change projects etc.
Making knowledge relevant also means being mindful of today’s messy business reality. This is likely to be well removed from the architectural ideal of clean lines and perfect standardisation. At most large firms the data landscape is fragmented, duplicated and littered with tactical and end-user-computing solutions (e.g. Excel, Access). At the same time lots of different terms exist for the same concept and up-to-date documentation or knowledge might not always be at hand. In short the data knowledge platform and framework capturing the understanding needs to be able to cope with this reality to ensure adoption and continuous support going forward.
Additionally the right tools and processes should be in place to allow for the knowledge to be built up collaboratively and progressively, allowing for any siloed understanding to be connected up without the need for any changes on the ground.
Surprisingly enough the depth of understanding required to foster collaboration is fairly shallow. Less is more here. The usual 20+ page document going to the Nth degree of detail is rarely a catalyst for collaboration; in fact this is a more often an obstacle to engagement across functions.
What brings people together is an accessible and thin layer of understanding focused on the essence and the interconnectedness of items. Lots of in-depth knowledge repositories often already exist in large organisations, be those data dictionaries, process repositories, project directories, data model repositories etc. However these are rarely connected up. In order to empower everyone to find answers about data, basic information on the organisation’s core data items and the connected business environment in which they operate must be made readily available.
Since shared understanding brings people closer together, one needs to ensure the lowest possible barrier to engage and participate as only then might people be inclined to at least have a look for what is already out there before creating their own solution. This means that one needs to present the integrated data view in an easily accessible and engaging format.
The data opportunity
Silos are a challenging part of the reality at large organisations and often perceived to be one of the main reasons why data is of poor quality. Rarely, however, is data viewed in terms of the opportunities it presents. We should recognise that data is the language in which we do business and, more than anything else, is the one thing that binds these silos together.
Understanding the data usage and flow across the silos creates a shared understanding of the business that allows each other’s information to be reused and connections to be optimised. A shared data understanding bridges the silos of the modern organisation to build a business-wide network, a lingua franca of data.
To understand your data is to understand your business.
If you are reading this then chances are you may be responsible for maturing the governance, quality or compliance of your data.
To help you implement some of the ideas in this article I’ve created some questions below that I personally use within organisations to help them gauge potential areas for improvement but please add your own thoughts and suggestions in the comments below.
Try to get sight of what is truly holding staff back to be more joined up around data. Is it really the quality of data or the fact they can not find, trust or understand what to use when?
Is there a business view on data being managed within the organisation which captures core data items and its business usage? If so, is it linked up with a data governance framework?
If there is an existing data governance framework what meta-data is available to data owner and stewards make sound decisions?
- When it comes to data integration is this seem as a purely technical affair? Here you might want to challenge senior management of the fact that if data is on expressed in business terms how will they ever measure its business performance.
About Patrick Dewald
Patrick Dewald is a Data Governance Architect and founding partner in Diaku. Patrick has a wealth of experience designing Master Data Management and Data Governance solutions for financial institutions. He has been heading up Data Governance initiatives, designing and implementing group-wide data services from the ground up for the best part of 15 years. Patrick is recognised by his peers as a thought leader in the field of data governance.