An increasing number of administrators see data-driven work and data-driven decision-making as a strategic focal point. But in practice, important managerial decisions are still too often based on gut feeling and hierarchy. In doing so, data is primarily used as confirmation of management intuition. The solution lies in the realization that data-driven work is an organizational issue and therefore lies on the director's plate and not with IT.

Data-driven work is a leadership issue, not an IT problem

Behavioral scientist and Nobel laureate Daniel Kahneman, in his famous book "Thinking, Fast and Slow," described the following statement: "There are some conditions where you have to trust your intuition. My general view, though, would be that you should not take your intuitions at face value." 

His evidence for this claim comes from years of research on human decision-making and cognitive biases that he conducted with Amos Tversky. Kahneman shows that intuition is often based on recognition and experience, which can be reliable in some situations. But in more complex or unfamiliar situations, intuition can be misleading due to biases and heuristics (rules of thumb and mental shortcuts). Therefore, he emphasizes the importance of critical thinking and not blindly relying on intuitive judgments.

Many organizations that want to extract value from data invest primarily in Data architecture and IT infrastructure. This creates an important precondition to indeed work in a data-oriented way. But in practice, we often see that it is approached as an IT problem. Without looking at the managerial and cultural aspects of data-driven work, and without taking into account the subjective bias we all suffer from.  

The heart of the problem: Management biases and a culture that ignores data 

When data (actually "information") turns out to be incomplete or qualitatively questionable, people quickly fall back on ingrained patterns of choice based on "experience" and intuition. This becomes stronger as the organization has a hierarchical structure/culture. A variety of human biases then becomes visible at both the top and the shop floor, where Confirmation bias (confirming one's own judgment) and Authority bias (if the boss says so) are common. 

Poor data quality as an excuse 

Instead of working with and improving imperfect data, the lack of perfect data is too often used as a reason to make old-fashioned decisions by feel again. But working data-driven is not a habit; many managers are not trained to interpret data independently, or more irreverently, not sufficiently numerically trained. The Birthday paradox often cited among statisticians is a case in point.

This paradox shows that in a class of only 23 students, there is a greater than 50% chance that students have their birthdays on the same birthday. This seems intuitively surprising because the number of students is relatively small compared to the number of possible birthdays (365 days). Our brain is not used to dealing quickly with such numerical issues so we fall back on our (incorrect) intuition that this probability is much lower than 50%. Also in organizations, such intuitive interpretations regularly occur whereby we too often rely on our supposed insight and insufficiently look at the underlying data. As a result, the company's data maturity remains stuck at the operational level.

Of course, it is easy to see what needs to be done differently, but what can we do to effectively make data-driven decisions?

Create a culture where data-driven work is truly embraced 

The most essential step is the realization or recognition that we are now subject to our human weaknesses when it comes to decision-making. Without awareness, no improvement is possible. And intentions to work more data-driven do not then lead to actual results. Therefore, we have the following five practical recommendations:

1. Make data-driven leadership the norm and develop it 

• (Management) training in data literacy and fact-based decision making; 
• Create role models within the top who make (and explain) data-driven decisions. 

2. Introduce a "data challenge culture. 

• Allow employees to question decisions critically (obv data);
• Organize "challenge sessions" where decisions are discussed and substantiated with data.  

3. Improve data quality, but don't let it be a barrier

• Start with the best available data and improve it iteratively; 
• Create transparency about data uncertainties and how to deal with them; 
• At the management level, appoint data domain owners (often linking to process owners). 

4. Make data understandable and applicable 

• Invest in storytelling and visualizations to make data insights appealing and relevant; 
• Pay attention to biases in visualization as well. The form helps determine the level of acceptance of the information provided; 
• Use a limited set of relevant dashboards that directly support management decisions.

5. Measure and reward data-driven decision-making 

• Link leading KPIs to the use of data in decisions;
• Make data-driven decision-making part of performance reviews. 

Conclusion: Data-driven work requires a change in behavior, not just better data 

The problem with not working data-driven lies not with technology or data quality, but with the way managers make decisions. By changing leadership, culture and incentives, data-driven work becomes the norm rather than the exception. This approach will help your company truly transition to a data-driven organization, rather than just using data as hype.

For more explanation of the process of good decision-making, read the continuation of this blog here.

About Jeroen Kuijlen

This blog was written by Jeroen Kuijlen, responsible for the Data Strategists business unit within Valid. Data Strategists advises and assists organizations in developing and realizing their data strategy and applying data innovations. Starting from the organizational strategy and goals, this involves looking at issues such as information policy, data governance and the translation to architecture principles and the design and implementation of data platforms. In addition to his leadership role, Jeroen is also partly active as a strategic advisor in the field of information policy, data management and innovation.