The following article comes from a senior data scientist with an interesting and innovative mindset. Let's get into his thoughts, through his writing below.
As a data scientist, I was educated with two beliefs:
- Data as a means to achieve goals.
- Data in its final form and ready to processed
The first belief treats data as raw material: the data scientist prepares the data and business users will process it again for subsequent users such as clients.
The second belief treats data as material for decision making. The first belief positions the data maker as a neutral party, the data is adjustable according to user needs.
The second belief is more "interesting", the data scientist is the final result determinant of the data presentation that will be used by decision makers.
"Beliefs" are the methods (manhaj) created after consideration and research. Then, someone will go through it and make it a clear guide to its limitations (parts) which are built on principles and rules.
Which belief is the ideal one?
Having an audit process like in the first belief is ideal because there are two teams with two mindsets to adjust needs before reaching the decision makers.
This is very helpful because cognitive biases are filtered twice. The data interpreter would understands how the data generated has a certain bias towards the presentation of data, that's why it is sometimes insensitive to how other parties will react to the data.
The data team will have a contextual understanding of the displayed data and sometimes forgets that there is a background they forget to tell, the presence of a second team - for example, the Marketing Team - helps direct the key message to convey in a more conical manner and in accordance with the needs of the final decision maker.
This is a kind of Front-end versus Back-end pattern that will work well if both understand the knowledge of their respective sectors.
However, if one party does not understand the decision makers' wishes, it will be very difficult to reach agreement on the data to be presented, or worse, the data requested is "unseen".
Therefore, the second belief creates difficulties especially for decision makers who actually do not understand how to communicate what they want - this often happens. The result is an unproductive feedback that sometimes affects the color of the graph used.
This is different in the case where the decision makers already have concepts and key messages to convey, the data maker here acts as a consultant for decision makers. Helping them determine what data is needed and how to present it.
The most dangerous thing in this belief is the "yes man" data scientists, the data that follows the wishes and imagination of the decision maker will produce the same imaginative decisions.
If the data is meant to be displayed for public consumption or client pitching, it might not be a problem. However, if the data is used internally, obviously, don't claim that decision as data-driven.
Not only for tech companies, but if we aim to empower data as a determinant of company development, there must be awareness of data literacy from top to bottom, data-driven is an ecosystem, not just a linear process.
- Having a data team without having a decision maker who doesn't want to learn how to empower data is not enough.
- Having a data team that cannot educate other operational teams to use the data properly would hinder the data-based decision making process.
- Data-driven means willing to erase the culture of HIPPO (Highest Paid Person Opinion) or in the local wisdom of ABS (Asal Bapak/Bu Senang).
ACTIVE CONTRIBUTORS IN CMLABS, SINCE 2020
Andaru Pramudito Suhud is an alumni of the University of Indonesia with more than 6 years of work experience as Data-Scientist. Currentl,y Andaru is a Data Consultant and resolves the problems of multinational companies in various countries.
He is an intelligent professional with broad and deep insight. You will find his writing and thoughts again in cmlabs blog.