Data Modernization: The Big Hurdles — and How to Clear Them
Businesses know the future is increasingly data-driven, but many are surprisingly early in the journey, according to our recent research. We’ve identified the most common challenges and provide advice on surmounting them.
Data can either be the heartbeat of today’s enterprises or the Achilles’ heel, holding back potential innovation. Without strong data management, a modern data architecture and data governance foundation, businesses will have a difficult time layering on advanced analytics capabilities or implementing emerging technologies. In order to enable new, artificial intelligence-based insights for the most competitive results, they need to modernize data, unite applications with existing and new data sources, and enact strong data governance protocols.
We recently teamed with Forrester Consulting to evaluate analytics platform use cases, features and functionalities that support business needs, as well as the underlying data management and governance technologies that assist those use cases. We found that while organizations are forging ahead with their data-driven strategies and initiatives, some have not done the work necessary to ensuring a solid foundation is in place. (For more information, see the full study, “The Road to Data Modernization.”)
Here are the major challenges of data modernization, as well as advice on clearing them.
Clearing Data Modernization Hurdles
Regardless of whether they build or buy a solution (or choose a combination of the two), all businesses struggle to get their data management infrastructure right (see Figure 1). Whether early in the process or further along, they must contend with organizational gaps and improperly curated data, which can weaken the data foundation and lead to snowballing problems down the line. Our research found these common issues:
Data quality is a major barrier. When implementing an enterprise-wide advanced analytics platform, more than half of organizations noted data quality as a challenge; in fact, it was the No. 1 problem. Nearly half of respondents (48%) said data quality also impedes scaling AI to an enterprise level. This was cited as the No. 2 challenge here, just behind a lack of employee talent.
Employee skills are misaligned with data management needs. While all organizations struggle to some degree with workforce talent, attracting and retaining people with the right skill set is a defining issue for companies seeking to purchase a solution. This, in turn, impacts how well they can implement solutions and manage their data. Half of all respondents said employee talent was a challenge when implementing an advanced analytics platform, and 41% said data science talent was a challenge. The same holds true for scaling AI to an enterprise level, with 49% citing employee talent and 41% specifying data science talent as major challenges.
For those building a solution, open source is another hurdle. In organizations that have chosen to build a data management and analytics platform, the use of open source tools and technology introduces new challenges. Indeed, 58% of these companies are challenged by the speed of change, 55% by maintenance challenges, and 55% by a lack of the right implementation skills. Additionally, the use of open source is a non-starter for some enterprises — 64% have security concerns that keep them from even considering this option.
Basic solution implementation is in the early stages for many. Over half (57%) of businesses are either planning to implement or currently implementing enterprise data lakes, while 44% are planning or implementing data governance and management tools. Considering how foundational these solutions are to a data-driven business, these findings are problematic.
To date, solution implementation has been dissatisfying. Simply possessing a tool doesn’t mean you have a full solution. Roughly a third of organizations are dissatisfied with the implementation of their data governance and management tools, analytics platforms and data lakes. Not surprisingly, more advanced solutions such as AI and deep learning tools are even less successful; 39% of respondents are dissatisfied with the implementation of these tools.
This dissatisfaction likely stems from the failure of these tools to meet business needs. While traditional data governance tools were built for data warehousing, data today can be sourced from everywhere, and the data ecosystem is more abstract. Companies need data governance tools that work with these new architectures.
Data lakes failed because they were initially implemented to reduce the total cost of ownership (TCO) of the warehouse environment. Once the lakes were built, the business didn’t always know how to use the new environments and couldn’t find needed data when required, turning data lakes into data swamps.
Similarly, we find that AI failures occur because data is unprepared or unavailable. Additionally, the lack of skills beyond advanced analytics puts organizations at a disadvantage. In addition to statistics capabilities, data scientists need to possess the business expertise to build the right corpus.
The bottom line is that business users need systems and capabilities to come online quickly and show results immediately. While vendors often position their tools as silver bullets, internal challenges remain, solutions fail, and expectations remain unmet. To succeed, organizations must understand that technology alone cannot solve their problems.
Here are some recommendations we developed after analyzing the results of our study. While these recommendations apply to enterprises that choose to build, buy or partner, several focus on relationships with service providers, as we found this to be the preferred course of action for most companies:
Actively test and adopt emerging technology.
Data and analytics technologies are evolving at a rapid pace. Technology purchased and deployed today may be obsolete in three years’ time. Service providers that stay on top of changes in data and analytics technologies will bring in the best solutions and platforms at the start — and will provide a path of evolution that helps enterprises future-proof their investments.
Put cloud at the center of your data strategy.
Cloud data platforms are designed for analytics and operational workloads. By abstracting storage, compute and state, the variety of data and use cases, workload demands and data volumes are managed elastically at a greater scale than in traditional systems. Security and governance are also better served, as cloud service providers embed stronger encryption and detection, improve privacy protection and address data quality.
Prioritize service providers that map to business outcomes.
A shiny new data platform is a failure before deployment if the objective is only to stand up the technology. Seek a service provider that understands business initiatives and priorities, and has the vision to change how the company operates when running on data. Such partners will be able to link the road map and future iterations to tangible business results.
Ensure data governance is by design.
The biggest threat to data is a lack of offensive and defensive policies that remove risk, drive compliance and deliver value. Too often, governance is executed after the fact as a cleanup operation. By selecting a service provider that weaves data governance into the fabric of the platform and plans for sustained governance, businesses can get on the right path from the start.
Understand what it takes to get the job done.
Having the right skills and project leadership is only the tip of the iceberg. Service providers differentiate themselves when their teams also operate under the best practices of engineering and development. Not only does working under agile and continuous development frameworks drive faster value; these same practices should also be stepping stones organizations to a new way of building capabilities. A service provider should offer training and change management programs to help organizations make this pivot.
Partner with a service provider that can navigate buy, build and hybrid strategies.
Each organization’s data and analytics maturity and objectives will determine which existing technologies meet their requirements and what they either need to build or wait for. Additionally, future-proofing data and analytics investments should always be top-of-mind to avoid vendor lock-in. Service providers should be experts that organizations can partner with to balance these priorities.
Lifting the business up to pour a new data foundation is complex and disruptive. Most enterprises don’t have the luxury of starting from scratch. Our research shows that partnering with a service provider offers an insurance policy for the strategy, design, development, implementation and culture change needed to sustain and extend the value of the new platform.