Data Analytics - How do I go about it correctly?

In the reality of companies and organizations, things quickly change: Different departments or site managers work with their own Excel files and individual definitions to calculate key figures relevant to the company — a company-wide comparability is therefore not possible.

In order to establish a data-driven culture, it is first necessary to define the key questions: Which KPI's ('Key Performance Indicators' or key figures) are relevant for my company or organization? What developments do I want or need to monitor and what goals have I set myself that I want to pursue?

These questions lead to the next step: Which data already exists in some form and which should or must I start collecting — regardless of whether it is internal sales data, market data, sensor data or budget requirements.

This questionnaire creates the company-wide data catalog: Here, the sources of trust and the validated definitions of calculations are documented, thus overcoming error-prone silos.

This data catalog can be divided into a data lake — a more or less loose collection of data that is valuable now or later for gaining knowledge or another type of monetization.

And the data warehouse — here the data is already being prepared by data engineers for specific questions and 'use cases” for defined analyses.

Companies often feel driven to introduce 'Advanced Analytics' or 'Artificial Intelligence' without a specific question. However, it is important to comply with the Pareto principle, which roughly states that only 20% is required to achieve 80% of the goals, while an effort of 80% is required to achieve the last 20%.

If an organization wants to get something out of the last 20% through artificial intelligence and predictions, the data basis in the company must first be created.

In practice, the opposite approach often fails and only disappointment remains, even if there is no lack of motivated and well-intentioned goals.

If a company or organization has not created the foundations for being data-driven, i.e. has established a data catalog and a (modern self-service) analytics platform as a 'single point of truth' in the company, implementing measures to meet buzzwords will only spend a lot of money and make little use of it.