Why should companies start with questions rather than data in their analytics?
My point today is highly associated with what I wrote about asking questions.
Data is today’s diamond; the most valuable product of any companies.
According to recent research, the analytical market size in the US was valued at $4,75 billion in 2018, is expected to reach $22,4 billion by 2025, with a CAGR of 25.2% from 2019 to 2025.
According to IDC, by 2020, there will be around 44 trillion gigabytes of data (44 zettabytes).
Our new data production speed is 2.5 quintillion bytes per day.
Obviously, data science is a term that has limited usage in business. With the rise of social media and AI, companies are increasingly relying on data to gain deeper insight into their customers and markets. Many companies invest in stellar employees by enrolling them in data science programs, recruiting brilliant engineers with Ph.D., hiring industry experts to formulate a data strategy and rearrange their decisions around the data. I am following all this effort and investment with great admiration. My only fear is that, in regard to the budget allocated, often those attempts would be unsuccessful, as in the case of digital transformation.
We solve everything with technology, do we?
As the amount and complexity of data produced each year increases, so does the demand for tools such as augmented analytics, which enhances analytical efforts with machine learning and natural language processing, to gain valuable insights. This technology helps data scientists accelerate the data analysis process by handling and receiving key data sets. With the huge volume of data and limited lapse to gain competitive advantage by increasing efficiency, there is a rising demand for intelligent business tools in the field of data analysis and this need accelerates the growth of the market.
The article of HBR in 2013 highlighted the importance of utilising machines to get enormous data insight. It was a pioneer invitation. However, what I disagree is that these machines positioned enough to give the right and perfect insight. The article said: By embracing the power of the machine, we can automatically generate stories from the data that bridge the gap between numbers and knowing.
Wow! Very bold, isn’t it? If there is data, then we can generate any relationship for any output of those machines. This process was called “generating a narrative from data” in the article. Wouldn’t that any random conclusion be misleading?
If we try to interpret the data with Aristotle’s syllogism, the result will be poor, just like in this example: The truth is bitter. Pepper is bitter. So pepper is real.
The problem is definitely not the lack of data
There is a lot of data for individuals and businesses to make a clear decision. The abundance of data provides unparalleled opportunities to empower the idea and product by understanding human needs and providing real-time visibility of any target audience. Imagine going into a parking lot: there are plenty of places and you don’t know which one to choose out of these abundant options. As with the choice of location in the car park, a search for your needs — we can call the hypothesis here — will lessen your options. Therefore, it is very important to know how to approach the problem of data science where there are unlimited data sources and powerful technologies, and to understand what kind of questions data science can answer.
Data science is not magic.
Statistics and data analytics can’t magically solve all the problems of a company.
The problem is that companies can not turn their data into actionable insights that can be used to make better business decisions.
I know many business owners are trying to figure out how they can use data to grow their business. Many of them do not know how to ask questions.
It is not magic. But just as indicated in the famous best-selling book The Secret, there should be a clearly defined way of asking.
What can I not ask?
I can’t identify general problems like what our customers want.
As if staring into a tarot reading, I can not see how I can leap at the next phase in my business in future.
I can’t foresight what to do to double my profit.
Although these questions seem logical, they can not provide a sufficient starting point to examine the data.
What can I ask?
To take action, you need to define a hypothesis. It is critical that asking who or what by limiting them into a context of where, how, and when. By then,
I can predict the likelihood of something happening.
I can detect anomalies in an existing past.
I can comprehend if there is a connection between different incidents.
I can classify the data.
What Einstein and Socrates knew
The acquaintance of statistical science with the business world was first by the field of Operational Research, and then by developing web-interfaced algorithms we built Business Analytics units, and finally, with the rapid growth of the Internet, it gained its popular name Data Science. There is a very important emphasis on this easy-to-read and cool study: Science!
Where there is science, any attempt must be scientific. Its first condition is to ask questions, the bold and right question. This is the most important technical qualification of a scientist. The question forms the hypothesis and through experimentation, the hypothesis is tested.
The technical aspect is not enough. The most important competence of a scientist is to be a sceptic to the question and the answer. That scepticism allows the data scientist to succeed because the outcome is not greater than himself or herself. He or she will update the question in an iterative manner. This method is called the Socratic Method (critical thinking).
I witnessed such analyses that the data scientist was so in love with the output that he never had any doubt about his questions or revisited the business owner to discuss for iteration. Most of the time the aim was not to turn data into insight. The aim was to try to show that they did a very good analysis. Those data scientists were treating the business owners as frauds and applauded by incompetent executives who are ready to penalize the business owners. Unfortunately, while these people have the set of behaviours that can not even fill the title of the analyst, they carry the title of data scientist just because the company follows trends.
Scientists manifest themselves in behaviour, not just the expertise.
To have successful results, do keep in mind the human factor* while designing your organisation.
Good scientists have a set of traits to make observation, experimentation and iteration: curiosity, logic, creativity, scepticism, and objectivity.
Best companies which make decisions based on data prefer recruiting academics with PhD. Because they are supposed to be scientists and have traits of scientists.
To succeed in gathering valuable insight through your data science projects, please take not only the expertise but also the traits of your data scientists into consideration. And do not forget to integrate them into SBUs.
* The human factor is the strategic approach to design organisational systems focusing on how employees behave psychologically and physiologically in particular environments and situations.
