The explosion of ‘Data Science or AI as service’ has brought all organizations to the forefront of the AI world. However, many organizations are unable to generate economic impact despite leveraging these emerging technologies. Today, a huge number of papers and companies talk about the success of models etc., however, the economic value remains elusive.
Satyam Priyadarshy who is the chief data scientist at Halliburton with over 20 years of experience in academia and industry, shared his experiences on how companies are claiming they have achieved digital transformation but have essentially failed in generating impact. He takes us through the journey of some of the commandments of failure in this journey.
Digital Transformation In The Current Age
Satyam believes that while the industries are trying to be data-driven and are using various tools to bring it into action, people are taking too much about it without showing actual results. He says that for companies to show results in continuous transformation, it is required for them to work in the real-time. Stressing on the need to have a digital maturity, he says that people are keen on using these trending terms but what we should be actually looking at is if it creates economic value and the required impact. “Measuring the value is important and if people cannot do it, then it is not worth it,” said Satyam.
Stressing about the fact that digital transformation is not just a buzzword but a technology very much in use, he said that it is being used in all the aspects of businesses. “But there is a huge digital divide that exists. There is a gap between domain expertise, operations, opportunities and more,” he said.
While many companies claim to be using dashboards but they are not as useful as they seem to be. But there has to be a complete understanding of these challenges and adopting it for transforming the way companies work today. “Digital journey has evolved over the years from digitization to digitalization and now digital transformation,” he shared.
No Economic Impact Is Failed Digital Transformation
Satyam shared that big data analytics, integration of emerging technologies, automation and applying all of it in real-time are the four pillars of digital transformation. “In order to achieve this we have to look at data of all kinds, data science, emerging aspects and economic impact,” he said.
He added that data science has existed since years as people have been doing science on various kinds of data whether physics, chemistry or mathematics. Only recently it has become a marketing term and businesses are calling themselves data-driven without driving real value from it. The conversion of dark data to smart data in many industries is still quite low.
Digital transformation should show an economic impact which can be measured in term of revenue, cost, NPV and others. If the company is not able to show any of these, it is probably not even digitally transformed.
Why Digital Transformation Fails
Data cannot be explored: Lame fear is one of the biggest reasons for failure. There are many industries that have a huge amount of data stored in its repository but are not explorable. “Search engines at many companies do not even compete with the likes of Google as the data that you would like to search is still not explorable in those search engines. Most companies have data but they are not being put to use. In order to be highly effective, every data should be searchable and knowledge management systems should be made effective,” shared Satyam.
Low-quality data: The latest incarnation in the data world is that of data lakes. While there is a generous amount of data but data quality is not up to the mark. This essentially means that companies have a lot of dark data which haven’t been converted into smart data. Creating no value from data is another aspect. Moore’s law can generate a lot of data but creating moving data is tough.
ROI is another aspect: We often talk about what is the return of investment in data-driven innovation, whereas it actually should be analysed as the return of innovation. Data science applications may take time to show results and therefore considerable time should be spent in looking out for results.
Proof of concept: Proof of concept is another largely used term in data science innovation. If we are looking at the proof of if data has value, if analytics works or if AI works, it has already been proven. What we need to look at is Proof of Value.
Poor leadership: One of the reasons why these points go amiss is leadership failure. “I know many leaders who claim to know it all, but actually have no idea about these terms and how the tech works,” shares Satyam.
Adaptability and agility: Companies are not adopting it as fast as they should.
Confusion about data: Another aspect is confusion about data and its use. While companies are looking to build machines with human cognition whereas they are not even close to building one with even animal cognition.
People do not know how to measure impact: While a lot of companies are claiming to have undergone digital transformation, they still do not know how to measure impact.
