Updated: May 29
Data Strategy: Unleash the Disruptive Power of Data
Because data sits in databases and files on some server somewhere, the information technology (IT) department was historically taking care of it. As a non-IT person, all you wanted was access to the data. Somehow IT would proceed to explain just how hard this was for them to do without requirements. You would retort that your needs were "I want access to everything." You know the drill. It involves circles and the expression "going around."
Times have changed. IT now manages infrastructure and offers services, and the business is being asked to play the role of "product manager" to guide IT in leveraging this data asset. You are not a data management expert, yet you are in the driver's seat. So how will you guide IT?
You are not a data management expert, yet you are in the driver's seat. So how will you guide IT?
Thankfully, you have us to guide you. The tool we will go through today is called the Data Strategy. It is useful to look at the Data Strategy as a contract between those who need the data, a group you are part of, and those who will make sure that it is available, of high quality, and relevant, according to your expectations (within reason, more on that as we go). Imagine you are a product manager, and this is your product roadmap.
What is a Data Strategy, and Why Does it Matter?
Scour the internet, and there are hundreds of definitions. We think that a picture is worth more than a definition. That data is an asset to be harnessed is an understatement nowadays; the Data Strategy documents how the organization will come together to ensure it optimizes what it will get out of data. It creates the plan, the roadmap to get there.
There are six strategies to define:
The data asset itself
How we assign value and set priorities
The platforms and digital infrastructure
The organization, the people and the culture
As the diagram shows, the Data Strategy will define what each component means and how we expect it to be delivered. Of course, the six strategies are very connected and must work together.
A Data Strategy will likely lead to a Data Transformation, in the form of a program that will last more than a fiscal year, delivering incremental business value and the underlying required sub-components, such as operationalizing a Data Governance Framework and a Data Quality Framework. There is a good chance that it could be self-funding.
The savings will come in a few different shapes:
faster, more accurate analytical models can create sizeable revenue uplift
reduced storage and infrastructure costs (the cloud.)
improved productivity among data resources
The business case will come from answering these few questions, upfront:
What features, based on data, can transform your business model or your market position?
What area of the business is most at risk of being disrupted by data?
How can data improve profitability?
How can we use data to reduce our business risk?
What is the opportunity cost of waiting?
Can you do it with in-house teams?
When we think of opportunities like data de-duplication, redundant or inconsistent data, and automation, it's easy to see how a Data Transformation based on a solid Data & Analytics Strategy can indeed pay for itself. And don't get us started on how projects with limited budgets often put in place solutions that are either not scalable or sustainable. The Data Strategy aligns not only imminent work but also how the future will evolve.
The benefits of a solid Data Strategy are:
inform decision making;
know customer trends;
smarter services and products;
better internal operation;
How to balance offense and defense
The questions above highlight another aspect of the Data & Analytics Strategy: Are we trying to change the business, or are we mitigating risks and protecting something? Most organizations need to do both, but depending on your industry, you may lean one way or another. As an example, the financial industry is very regulated and one company may opt to first focus on defence or prioritize some capabilities that cover such needs as compliance with privacy laws.
Leandro DalleMule, CDO at AIG at the time, and Thomas H. Davenport wrote an article for HBR.org almost three years ago about trade-offs between "defence," and "offence" when it comes to data, between control and flexibility in its use. The article says that in highly-regulated industries, like Healthcare, we typically see a more defensive approach., whereas, in Retail, we take a more offensive approach. And 50/50 is not a good idea, based on the fact that IT will need to execute somewhat differently depending on the orientation. Read the article for more details.
How to put the business first
These principles will help in setting overall objectives. You will go down rabbit holes during your work sessions, but these principles will help you see things clearly and get back to basics if you refer to them often.
Encourage collaboration and sharing;
Make data visible and accessible;
Dissociate the needs for protection, availability, quality, and relevance of data from the operating processes and practices required to fulfill those needs;
Establish guidelines for analysis and analytical applications;
Identify which data leads to customer lifetime value;
Identify which data leads to operational efficiency;
Identify which data measures "experience" with your products or services;
Be clear about the business outcomes you are driving for.
How to plan and carry out a Data Strategy
First, success requires some pre-conditions. To have the best chances of success, these will help:
Obtain buy-in. We have seen that organizations obtain the best results when they work collaboratively. Not all organizations have such a culture, unfortunately. No one said it would be easy;
Have a data management team that will drive the exercise for you; they will inherit much responsibility in carrying the work out, so they should be a big part of the process;
Involve all essential functions. Data can help save costs, increase revenue and change or improve your offering. So everyone can benefit and contribute. Eventually, many of those involved here will be part of Data Governance.
If you need a project plan, a Data Strategy exercise goes like this:
Assess the current state;
Define the future state (for all six components in the framework above);
Identify the gaps, especially what is preventing you from reaching the business outcomes you have identified;
Choices. The gaps may prove to be more than the organization can absorb, for valid reasons. Refine your target state so that filling the gaps become your objectives for moving forward;
Identify your transition approach based on the constraints and the business priorities of the organization;
Put together a program that incrementally transitions to your future state according to your defined approach;
Use the Data Governance function to guide the program and adapt, because one thing is a given, the plan will change;
As the program winds down, Data Governance becomes the driving force behind the evolution of your Data Strategy in the future. It continues to ensure the organization derives maximum value from data. Because it is business-driven and perfectly aligned to the business plan, it becomes easy to justify keeping all components optimized and up to date.
How to include all the latest Data capabilities
The Data Strategy should go into some level of detail, particularly conceptual architecture. For all six components previously discussed, think of a diagram that fits on one page that shows boxes and lines, sometimes with arrows, that shows what you aim to build in simple terms. A high-level data flow diagram or an organization chart both fit the objective. Here is an example from Snowflake for the Data Platform component:
As mentioned before, experts and your data management team should do this for you, with your guidance. But here is an important point: your Data Strategy must be rooted in reality. It must leverage available technology concepts. Therefore, it should include your organization's position (will you use them, why, how do they fit in your conceptual architecture and what business outcomes they will contribute to) on the following generally accepted, modern approaches to analytics:
Data ingestion, replication, and extraction-transfer-load
Master & reference data governance and data quality
Data pipelining and orchestration
Distributed big data processing
Big data databases and data storage