Digitization is all about connection. What cannot be connected creates a roadblock on the way to digital transformation. For companies that are on their way or plan to get going, it's becoming increasingly apparent that data silos are the biggest roadblocks to overcome.
Content:
According to a CMS Wire study, 47% of all surveyed customer experience managers said that data silos are the biggest challenge in their organization.
Meanwhile, a survey among data scientists found that 60% of all their work is spent preparing and organizing data (source: Forbes). Another 19% of their time consists of collecting data sets. In short, nearly 80% of data scientists' work is spent with sourcing and cleaning data, rather than actually analyzing it.
Edd Wilder-James explains in the Harvard Business Review that there are several reasons for data silos.
Software applications are often developed for a specific user group and a particular purpose. Sharing data is rarely the primary focus.
Data transparency is not always easy to implement in companies. In addition to compliance guidelines that necessarily restrict access, there are also quite often cases where data is deliberately withheld to reinforce power relationships. By being the main or even only person who has access to the information, they yield more power and are more difficult to replace. However, this, obviously, always harms the company's overall business intelligence (after all, this is not Highlander).
The older a company is, the more complex its data structures, applications and processes will be. Older (legacy) systems in particular are sometimes not as easy to connect to modern (cloud-based) applications. In addition, data has been collected in a wide variety of formats and taxonomies over the years, making it much more difficult to centralize and process.
When selecting software, companies are always scared that the chosen data processing/retention system is so unique that migration to another system is next to impossible. Whether deliberately designed that way by the vendor, or due to other circumstances, it can indeed be difficult to change systems and transfer data without loss or major operational disruptions (however, it is not impossible).
By the way, current developments regarding the so-called "super cloud" will be interesting since this new trend aims to get rid of exactly this issue.
For a large company with complex structures, it can be extremely challenging to identify existing data silos, collect and structure the data. Even a single app that processes data but is not linked to other systems can be its own data silo. Based on how many apps are used in companies - sometimes undocumented - it is easy to create a data labyrinth.
The results of a recent Mulesoft report show that the number of apps used in companies has grown by 10% in the last year alone. At the same time, only every third app is properly integrated in the IT-landscape.
In addition, a lot of data continues to be stored locally on individual computers or even on literal paper, so access is often only available to specific individuals. An Excel sheet that's not been shared in the cloud or a paper folder is even harder to locate than an app.
When starting a data project, the first challenge is to not only identify the existing systems, applications and data sources, but also to record the requirements of different business units. This is necessary to ensure that the new data system can be scaled properly and actually fulfills the needs of the company. If the new data system can't fulfill the requirements and lacks the necessary integrations, it's very likely that the users will use external solutions to do their work.
In order to introduce a central data platform, all company departments and processes must therefore be considered. This also includes third-party vendors, intermediaries, etc.
And don't forget potential future needs. In other words, the planned data platform must be flexible enough to adapt to changes if required.
Patricia Robles explains in her article on econsultancy.com that companies rarely allocate a budget for central data management. It's more common to provide data budget for individual business units or IT sections such as email, data protection, etc.
A central data platform, meanwhile, is often budgeted for marketing and sales. However, due to its scope, it either might succeed the existing budget limitations or be too small since it is only intended for one business unit.
But it is important that the elimination of data silos and the implementation of a central data platform is a priority for the entire company, which ideally does not have to be financially supported by one business unit but is budgeted and prioritized as a separate project.
Otherwise, the responsible divisions will - again - find alternative solutions that will further expand the number of data silos.
Overhauling internal data structures is costly and requires a great deal of internal resources. For many companies, this might not be worth it at first glance. A large sum for a data project is often more jarring than the many smaller, accumulating costs that are caused by unstructured data and lack of transparency. Additionally, these costs can be invisible and are rarely recorded, for example, when employees have to manually process data or manage data in several places at once.
This is why a comprehensive requirements catalog is important. It can indicate how costs and resources can be saved with a central data platform. It can also highlight where revenue can be leveraged through the use of automation or artificial intelligence, for example. If documented company-wide, it is possible to quickly visualize how much worth a seemingly costly project can be in the long run.
DIGITALL supports you in evaluating your data landscape and helps you to structure and link data on a central data platform. Whether smart interfaces, your own data warehouse or a central system for customer and object data - we accompany you on the way to digital transformation.