Unlocking the value of data modernizations

If you are reading this now and are part of a data modernization in your organization, you might wonder why getting to tangible outcomes is much harder than you anticipated in the initial “sales pitch”. The rise of cloud provider mammoths and massive investments in artificial intelligence have brought a surge in compute power and paved the way to achieve the “art of the possible”. However, as in all things too good to be true, there is a catch. Cloud data platforms and solutions bring many intelligent capabilities, making best use of them in your organization is, ultimately, your responsibility. And unfortunately, in most cases, the shiny ROI from the onset of a data transformation is delayed or partially realized. This is because data transformations are, believe it or not, much more than technology. The “plug-and-play” myth, where you simply onboard a new cloud data platform in your ecosystem and start seeing the benefits of your work, is seldomly a given.   

There is another critical point worth discussing. History has shown us that the world likes to reinvent itself, with technology waves coming and going. Some technology waves are sticky, some disappear to reappear later because they emerged too soon for society to adopt them, and some are simply dropped after the initial market hype. Not surprisingly, the data transformation wave, with all the capabilities enabled by cloud and data management solutions, is a sticky one. Why exactly? Well, you might have heard it somewhere else, data is the most important asset an enterprise has and the backbone of past, current, and future technology waves. Just look at GenAI, in addition to its visionary potential, it already has a powerful and broad impact on organizations, helping them think differently about their business models and processes. However, to scale from experimentation to use cases in production, GenAI needs one fundamental ingredient. Not surprisingly, it is good data. And while structured data is, for most industries, enough to get a decent foundation for Machine Learning, GenAI extends the care also to unstructured data (i.e., images, documents etc.).  

Acknowledging the obstacles with data transformations, you might wonder what you can do to smoothen your journey. Unfortunately, as the experts like to say, it depends, as there is no single recipe for success. It eventually comes down to your own circumstances, business needs and readiness to adopt change. There are however recommendations to share from seeing what worked well and maybe less so. By no means complete, we will look at couple of these.     

A.      Get the most value from your combined Data & Digital transformations.  

What usually happens: many organizations started their data and digital transformations in parallel, however, frequently, decoupled from each other. The lack of integration across modernization initiatives could lead to missed value opportunities, where platforms are not used to their full potential. Take for example a company implementing a new cloud hosted CRM platform, oftentimes accompanied by its own data platform, while modernizing the enterprise data platform on cloud (i.e., AWS, Azure), in parallel. By not linking the 2 ecosystems to get to a complete 360 view of the customer, value from analytics use cases i.e., next best action, could not come to fruition as expected.

What optimally happens: A holistic modernization roadmap should include both Digital and Data initiatives, planning them jointly, based on value use cases for revenue growth and cost optimization. The intersection and dependencies between these should be captured early on and continuously refined, while identifying synergies to make best use of all available platforms.

B.      Inventory your data consumption needs and create a plan for the full transition

What usually happens: Many data transformations start with a “lift-and-shift” approach or expedite the raw data migration from legacy systems to a new data platform, postponing the true data modernization for downstream consumption to a later time. While this is one way to go about transformations, most likely it will create delivery challenges along the way and potentially force organizations to maintain multiple data environments in parallel. With potential cost implications i.e., multiple fees, doubling maintenance costs.

What optimally happens: Organizations should evaluate their current and future data consumption needs ahead, discover rationalization opportunities and create an initial plan for data consumption modernization i.e., reimagining data pipelines. It does not mean that the execution should happen overnight, however a plan helps the transformation program track its progress and define a clear message to the board level when the modernization efforts are expected to complete.   

C.      Tie your business case to the strategic goals of the enterprise.

What usually happens: most transformations start with a compelling business case, however, along the journey, it stops being followed through. This usually happens when business cases are defined with high level assumptions, not traceable back to the actual business value created for the organization. They become difficult to reconcile with the strategic goals of the organization and how the data modernization truly contributes.

What optimally happens: the business case should follow clear principles for design and ask for both bottom-up (impact on products, lines of business, business functions) and top-down (strategic enterprise goals) views, which ultimately reconcile. Furthermore, the approval of a business case to back a data modernization investment, should require a transparent framework how value is continuously measured, monitored, and reported to both business and technology stakeholders, across various executive levels.     

D.      Co-create the transformation with your business partners. 

What usually happens: data transformations start, in some cases, as pure technology initiatives and, ultimately, are “sold” to business stakeholders as a game changer. Imagine a data platform which is technically ready, but no use cases are onboarded to enable business to see the real value. The adoption by business users would still be possible, however more challenging.   

What optimally happens: a data transformation should have the “buy-in” from all key stakeholders, both business and technology, from the initial ideation phase. This also means full commitment and joint efforts to make the transformation a success. Furthermore, planning is a co-creation exercise, following the business needs and how this will be prioritized for maximum value.

E.       Don’t underestimate the power of Change Management

What usually happens: when a data modernization is defined, the immediate question is: why change? In the end, old solutions work already for many years, even with their many workarounds and manual interventions. A transformation also implies an upskilling of the workforce, which can be met with resistance. The change management aspect of transformations is, unfortunately, most often neglected.

What optimally happens: Change Management and the initiatives tied to this should be an integral part of the E2E data transformation roadmap. Such initiatives would ensure designing a realistic adoption plan, tailored to the circumstances of an organization, understanding talent gaps, proposing how to close them through internal upskilling and / or external sourcing and ultimately bringing the comfort level in the organization to embrace the change and be ready for it.          

F.       Collaborate with your internal partner network to ensure talent coverage.

What usually happens: One last point, however crucial in the day-to-day, is focused on understanding the strength of your internal ecosystem partners and making best use of their support. Imagine a data transformation starting without the buy-in of the InfoSec or Infrastructure teams. While a later resolution of all blockers is possible, this often leads to significant delays with the project and additional burden on the delivery teams.

What optimally happens: A strategic relationship map should be designed as a critical success factor for the program. This translates into incorporating members from various enterprise groups into the transformation delivery, combining different skills i.e., Data Engineering, Infrastructure, Privacy, Data Governance.

The above recommendations are just a small subset for success in data transformations. Putting together the assembly line into a factory value model for a smooth data transformation comes down, ultimately, to setting the right mindset across the organization.