Majority of organizations outsource data science

One way to bring together information from multiple source systems is to use a data platform. This allows you to freely access all data. This goes further than, for example, a CRM system with customer information and sales data that can only be used for CRM functionalities. It's all about opening up the source data to other applications, for analysis and data science applications. A data platform provides a layered structure for storing, classifying, filtering and cleaning data. The right, high-quality data - the goal of data-oriented work - is the basis for the analyses that you can do with it and determines the extent to which you can start working in a data-driven way.

82% of the organizations indicate that they do not yet have their own data platform for data science applications. The 18% who do use it cite, among others, their own software, Azure, Power BI, Cognos, Google or Oracle.

Data governance

A data platform is a place where many different, sometimes sensitive, pieces of data come together. It is also the place where you determine who has access to which data: authorization. Or how long data may be retained: data retention. This ensures that the organization continues to comply with privacy legislation. For government organizations, there are also other guidelines, such as purpose limitation: the collection of data only for a specific purpose. When setting up a data platform, authorization and additional guidelines for responsible data use are laid down in a governance structure for the data platform as a whole.

Outsourcing data science applications

76% of organizations outsource some or all of their data science applications. That in itself is hardly surprising. It is mainly about the systematics surrounding it: setting up the system, opening up data and good governance. And that is not the core business of organizations and sometimes still relatively new to them. When organizations outsource data science, they prefer to leave it to a party with experience in the sector. Two thirds of the organizations that partially or fully outsource say they use an external party for knowledge and advice. Experience with data science is considered more important here (34%) than specialization in their specific market (19%). One reason for this may be that organizations expect to have sufficient knowledge in this area themselves. This can be disappointing in practice, as data science analyses and projects really require you to go into 'depth'.

As a rule, outsourcing generic issues is the obvious choice. Many tools and already developed models, for example for data storage, can be used for multiple situations. This way you don't have to reinvent the wheel and you can make the most of the advantages. Another factor is that organizations prefer to keep certain sensitive information in-house: for example, data about the organization's intellectual property. This data determines the distinguishing capacity in the market and therefore contains competition-sensitive information.

Challenges

Before organizations can get started with data science, there are a number of hurdles to overcome. Nearly 40% of organizations find "complexity" to be the biggest challenge for projects where data is needed for analysis. One of the factors that can make data science complex is that it takes many different types of data sources to build good analytical models. How do you bring all those sources together? How do you unlock the data? What algorithms do you run on it? But also: what legal rules of the game are there for these algorithms, such as privacy legislation? In addition to the complexity involved in collecting data, the presence and availability of knowledge in the organization in question plays an important role: which departments need to cooperate with each other when it comes to data-oriented work? Which disciplines and roles are needed? Who has the overview?