Today, you can’t help but run into the will of companies and public organizations to be data-driven and take advantage of the benefits of data science, artificial intelligence, robotics, machine intelligence, predictive analytics or just simple reporting solutions. While the benefits are undeniable, I feel like they’re talked about a lot more often than something is actually done about things.
The greatest obstacle is the same eternal even today; access to data. If data utilization has been identified as a necessary competitive advantage for today’s business, then why is it so difficult to access the data we need? So, why not start digital transformation by focusing on breaking down your data silos?
Data science has generally emerged in the field, with practitioners claiming that 80% of the work is data acquisition and preparation. Despite efforts by software vendors to create self-service tools and various low-code / no-code tools for data preparation, this share of work is likely to remain the same in the near future for few reasons.
Each solution requires data that is suitable and tailored to the intended use. Depending on the application you want, you need to collect, format, filter and process the data accordingly. Some data can be easily organized, but very often the most critical data is poorly available. Data is painfully often in silos that are very difficult to access.
A progressive and practical approach to removing barriers to silos is most effective. Select data that is critical to your business and start systematically breaking silos. In order for us to start breaking data silos, let’s look at a few aspects of why they actually arise. It is then easier to define the necessary practical measures.
1. Political
Knowledge is power, and certain groups in an organization may be skeptical of others who want to use knowledge. This often happens for legitimate reasons, as the potential for abuse, even accidental, is wide.
Data is not always a neutral entity - you have to interpret it knowing its history and context. This sense of ownership can work against the interests of the entire organization. I´ve seen this far too often, especially in public sector projects.
Very often, instead of doing it in practice, we start doing complex policies, operating models and research. Things could often be resolved by doing and boldly trying things out in practice. Powerpoints and vast PDF reports practically never solve real challenges. Especially when a common goal should be to breaking down data silos.
2. Structural
Software applications are written at a specific point in time for a specific purposes. In a world of limited resources, applications have been optimized for their main function. Incentives for individual groups are unlikely to encourage data sharing as a primary requirement. Data sharing and interoperability may not have been the primary focus. If it is almost impossible to break down silos, e.g. in legacy systems, try first connecting them and focus the value of obtaining data.
3. Vendor lock-in
Software vendors are among the first to know that access to data is power and their strategies can frustrate users’ desire to export data contained in applications. This is especially dangerous as software applications where the vendor builds solutions inside their cloud platform.
Some developers and vendors have worked hard to create complete work assignments and career paths around their software or application. Any hint of change from this world can threaten the livelihood of trained and certified software professionals. Fortunately, the attitude has changed recently. Data is wanted to be shared and seen as a fuel for business.
4. Company growth
When a company grows quickly, infrastructure and processes often fail to scale, individual departments may implement processes and applications in an ad hoc fashion. This produces data assets that are usable only by the teams that produced them, as well as a backlog of cleanup and integration work for data managers and IT.
5. Multiple and overlapping apps and sources of data
Nowadays many organizations use cloud-based SaaS applications to manage core processes and many of these applications do not integrate directly with each other. It can be very difficult to identify the basics of data usage when the data is not immediately under your control.
Another key challenge, on the other hand, is that an organization has multiple applications and data sources for the same purpose. Data management may become virtually impossible. This is especially true for large organizations.
Does this all seems challenging? Well, we can cost-efficiently help you in this as well, for example with the help of an innovation voucher.