You work for the company that has and on daily basis collects huge amount of data all over the place – in various different systems.
You’re aware of the fact that those data are not well organized.
Your boss wants more efficiency in data product development and delivery processes, especially in regard to the IT.
Your legal department wants better organization of data and more transparency.
Your IT and business colleagues want sufficient and understandable documentation in relation to the data and their applications.
You actually as well feel the hidden potential of those data, however, don’t know how and where to start.
If you face with those challenges, then you’re known you need something mostly referred as Data Architecture (DA) or Enterprise Data Architecture (EDA). But how do you start with DA? Believe it or not, but you start almost the same as with any other type or architecture.
This article has an aim to provide simple and understandable framework to start data architecture in your company and to develop appropriate data strategy as well. It provides seven steps (Figure 1.) you need to follow to solve “data challenges” at enterprise level. It is suitable for: (i) those starting or wanting a role of Data Architect / Enterprise Data Architect; (ii) IT leaders which company doesn’t have Data or Enterprise Architect; (iii) business stakeholders working on data relevant projects; (iv) and for management to understand to complexity of data based processes. So, here we are:
Figure 1: A framework to start data architecture and to develop appropriate data strategy
1. Define what do you want
First, you should define what your desired state would look like. For example, you want do have sufficient, transparent and understandable documentation about all data in your enterprise. You also want to have your data agile, meaning you want your data to be very flexible and optimal in supporting implementation of business requirements. This step includes speaking with various stakeholders, operative and management, about their wishes and requirements. It also includes documenting those aspects, organizing and refining them to define the desired state acceptable by all, probably through compromise.
2. Analyse what you have
Second step includes analysis of the current state of the data in your company. In this step, you should speak with every business/IT team or department to collect as much information as possible about the data in your company. Don’t forget to document everything you or others consider relevant. The best way is to start with creating appropriate data diagrams immediately after the first meeting and to update them through every subsequent meeting. In the case of the data, the most appropriate types of diagrams or models for the beginning are those documenting data flows at high-level. They would help you to understand the way that company does its business, and it would help others as well to better understand the whole concept.
3. Identify issues, problems and challenges of the current state of your data
Third step could be actually integrated into second step as it includes speaking with IT and business teams about problems and challenges they’re facing with when working with data in the current state. You ask them simple what are the biggest challenges and problems they had, and in addition, what are their ideas to make the systems or environment better. Also, don’t forget to document everything, because you can use information about their problems and challenges to justify the need for data architecture with other, more sceptical stakeholders. You can also use their ideas for improvement to develop appropriate and acceptable data strategy. The people that actually work in the same company sometimes can provide the best ideas – you just need to motivate them to speak.
4. Organize all information you acquired
Fourth step covers analysing, organizing and clustering all possible information about the data you got in first three steps. So, you defined what you want to have, you made analysis of the status quo and you actually know what you have. At this point you should have some ideas about how to get to the wanted state, at least based on advices you got in previous step. You should also speak with other domain experts about your ideas, just to get some additional insights and prevent omitting something very important. If you don’t know any domain expert personally, use appropriate LinkedIn groups, or similar, to exchange your opinion. Consulting relevant scientific and industry based literature (such as articles, conference papers, case studies, white papers, etc.) is a huge plus as well. At the end of this step, you should have a clear idea what needs to be done – a some kind of data strategy.
5. Presenting your strategy to management and stakeholders
Now, as a next step, you should prepare and present your data strategy to the management and other decision-making stakeholders. This is really important step! So, don’t forget to describe the problems and issues caused by the current state of the data. Identify and explain benefits your data strategy brings. For example, how many personal days it would save for an average project, or how will it support legal compliance, transparency, sharing know-how, etc. Try to separate your strategy into maximum of five understandable steps that actually reflect the actual tangible products. If management or other decision-making stakeholders have some additional ideas, they consider relevant and important, you should try to find the compromise and incorporate them into your strategy, and then actually start with conducting your strategy. If you got asked how much time do you need for your strategy, an average time to apply whole data strategy is somewhere between 3 and 10 years for large enterprises in Europe. However, it would be very good to discuss this issue with domain experts before presenting your idea in the front of
management.
6. Translating your ideas to functional implementations
Step number six is to translate every single step from your data strategy into products or projects that are implantable. For example, let’s image that introducing a Centralized Master Data Management System (CMDM) is the one step in your strategy. So, in this case, you have to translate the idea of CMDM into actual implementations. You actually define what needs to be done to achieve the state of functional CMDM in your enterprise. You define projects, products, steps, milestone, sprints, or anything else that is appropriate according to the culture in your company. In this example, you must define those aspects together with relevant IT domain experts.
7. Hand over, supervise and improve
In ideal circumstance, step seven would include handing over the projects or EPICs defined in previous steps to respective project managers, EPIC owners, product owners, etc. However, as a person responsible for conducting data strategy, you should always advise, supervise, take the part in important meetings, and try to further improve your data strategy. You’ll definitely get some new ideas or new requirements which you have to integrate as effective as possible into your existing strategy. Even you might think you produced ultimate and perfect strategy, you’ll be surprised how many challenges and changes you’ll be faced with. So, prepare for compromise. The patience and communication are the keys of the success.
So, if you intend or have a requirement to do something with your data, I really hope that this little guide helps!