Data doesn’t help you make the right decision; it helps you make informed ones.
In 2017, the Economist magazine proclaimed that “Data is the new oil”. When the phrase was coined, I am not sure if anyone fully understood the implications of it all. In this day and age where we are literally drowning in data, we can better appreciate it in hindsight. While data is like oil in the sense that it needs to be collected and refined before it can be deployed to create value; but unlike oil, it appears to be an inexhaustible resource, where the world is constantly posting and consuming content on social media and streaming services while buying products and services online.
However, many misunderstand how data creates value. People are interested not specifically in the data, but rather the insights hidden within the data; insights that could become the basis for a competitive advantage in this highly competitive data driven world. Many believe that insights mined from the data would lead them to the “right” answers for their problem; but that is not how it works, it is an oversimplification. Data really creates value through “informed” decision making; “informed” decisions are not the same as the “right” decisions.
How data creates value
Without getting too technical and getting caught in the technical jargon, data creates value through the following cycle. Some of you might recognize this as the scientific or research process, the words and the number of steps may be different, but this is the process by which civilization has progressed over the past few centuries and will continue to leverage into the future.
Contrary to popular belief, the process doesn’t start with data, it really starts with asking the right question; followed by considering a comprehensive set of potential causes of the problem (hypotheses). The hypotheses will provide guidance as to what data to collect and process for analysis. Analysis deploys quantitative methods to objectively prove or disprove each hypothesis with data. Bear in mind that hypotheses are never proven with absolute certainty; they are often proven with a predetermined level of confidence (for example 95% or higher).
A series of proven and disproven hypotheses provides insights into the causes that matter, matter less or don’t matter at all. In most cases, it doesn’t offer a clear decision nor action plan to create value. It needs to be interpreted by the right people with the right experience and seen in the right context to develop an executable action plan for the organization’s consideration. The effectiveness of the outcomes depends more on the interpretation and action plans than it does from data alone. Finally, the analysis, recommendations and trade-offs are presented to decision-makers for their approval before committing resources and action.
Data analytics or Data science is trying to bring the same rigor into organizations to help organizations make informed decisions through objective insights supported by data experiments. While informed decisions may not be the “right” decisions, but it does set the organization in the right direction towards it.
Learning from Data
If you look carefully at the process, there are many places where things could go wrong or be missed. Some examples could include:
What if leaders are asking the wrong questions?
Despite the best efforts in building a comprehensive list of hypotheses, some pivotal ones could be missed, which could have a significant impact on the recommendations.
Translation of the hypotheses to data for analysis is more of an art than a science, especially as it pertains to complicated issues.
The perfect data field may not exist, and analysts are forced to use a proxy data field instead; which could lead to misleading findings due to bias and/or the fact that it doesn’t represent the idea of the original hypothesis
Where data may not exist, it may be collected anew. Data collection is a whole topic by itself, to ensure that the data is representative of the population and biases are managed.
The analysis process is not infallible. Biases in the data aside, analysis could suffer from biases through conscious and unconscious biases of the analyst themselves. Analysis could also bring up red herrings. Correlation is not causation, but all causation is correlated with the outcomes; but how do we tell the difference?
The same proven and disproven outcomes can be interpreted into many different insights and action plans. The effectiveness of the action plans depends a lot on the analysts involved; specifically, their critical thinking, experience and creativity.
Decision makers also have a part to play as well. They need to know enough about the process, statistics and math to have confidence in the process; and be open minded enough to let the data sets change their mindsets. Analysts also need to build the right soft skills to convey a strong business case for the recommendations and associated trade-offs, supported by the data in a concise, coherent and compelling fashion.
If taken all together, some may perceive the process to be broken or at least unreliable. But in reality, only some of the gaps would appear at any one time; it would be unusual if they should occur all at the same time. One of the most often overlooked features of the process is that it is a cycle and not a process with a starting and an ending point.
The cycle is intended to depict the continuous improvement aspects in the process; more specifically the process of learning and building on what has been learnt. Regardless of any potential gaps in the analytical process, the measure of the success of the decision lies in the outcomes resulting from the decision. If the decision was correct, it should produce the intended outcome. If it did not produce the intended outcomes, then something might have been missed in the process and the resulting decision, which leads us back to step number 1 – asking a better question and the cycle starts again allowing organizations to build on their past knowledge and experiences.
Key Takeaways
The scientific process has a proven track record of making meaningful discoveries that impact human civilization. Organizations can leverage the same process to gain objective data driven insights about organization issues and challenges. While the process cannot guarantee the right decision is made; if consistently applied over time, data can help organizations gradually nudge their way to the “right” decision one informed decision at a time.
If you want to learn how to go from data to insights to informed decisions, check out FYTs upcoming workshops.
Comentarios