Insights on Data and Analytics by Mohan Sushantam

 


What are the leadership skills needed to drive effective change in organizations to achieve data and analytics excellence?


Storytelling is a key skill required to drive transformation using data science, superior analytical skills notwithstanding. Highlighting the insights backed by data in the form of a crisp story goes a long way in achieving analytics excellence. Most of the business leaders don’t appreciate the nuances of the model building but they do like to focus on the value created by a bespoke model. Taking a simple example of whom to sanction a loan to? 


Traditionally credit teams focussed on both quantitative and qualitative metrics and used manual judgement to decide on the application but with the advent of analytical models which can look at hundreds of data points and identify the trends which end up increasing both speed and accuracy of decision making. The change from manual to an automated decision-making process has human inertia involved which is resistant to change and feeling of losing control to a machine. Stakeholder management becomes key here on how you can to manage various participants so that the solution becomes win-win for everyone.


Can artificial emotional intelligence be a substitute for real emotional intelligence?


No. Real emotional intelligence is kosher. Only livings beings should have exclusive rights to it. One of the biggest drawbacks of machine learning or AI is its dependence on training datasets. So a machine builds its “emotions” on the basis of whatever has been fed to it by the programmer. Obviously, it can capture new “learnings” but how to interpret the new information is again driven by the programmers. 


This builds a huge bias wherein a machine can keep taking wrong decisions just because it has been done so in past which end result being perceived as successful. A good example to explain this is the automation of the hiring process using AI. Assuming an organization suffers from unconscious bias against female employees and has been hiring predominantly males. The “success” criteria as defined by the programmer is a good performance which again is biased because of organizations unconscious bias. The resulting algorithm using this data will predominantly select male candidates. Here a real emotional interface is required to appreciate and correct a “historic” wrong.


What does sales analytics mean?


Sales analytics has multiple facets. Since it’s a critical function is most of the organizations, it generates a maximum amount of data as well majority of which is not being captured properly at this point in time. Segmenting a customer group and targeting them with the right products is being done since time immemorial. Where analytics can add value is creating a granular laser focussed segmentation strategy instead of having broad criteria, besides building the right time to pitch and even creating the right “pitch” based on customer preferences. This will not only help in driving higher sales at a much lower cost but also help the organization to drive experiments which can capture new trends and help to build new products.


What role does analytics play in measuring the engagement level of employees?


Measuring human engagement is tricky because a lot of it is qualitative and answers are dependent on the mood of the employee at the time of answering a survey. So a better way to capture this would be through surrogate questions e.g. participation in-office events, offering suggestions during feedback sessions etc, and it should be continuous throughout the year instead of capturing the responses at a given point in time. 


There are plenty of use cases where analytics has been used to capture engagement levels using voice modulation and sentiment analysis has been carried out after appraisals or performance reviews. The engagement level is also subject to interpretations and hence objective rules should be defined (beyond business impacts) to quantify employee happiness. The approach has to be fine-tuned after numerous experiments over a long period of time to achieve significant benefits.


How can data scientists and analysts improve their communication skills?


As already pointed out earlier communication is a key skill needed to become successful data scientists. How to explain something complex to a layman through charts and figures needs deep subject matter knowledge. Talking to a person sitting in front of you in the language they easily understand is critical. So when talking to Risk head a different approach is needed compared to when conversing with a marketing head. Also, data science should be sold as a tool to help do better business instead of selling it as a disruption which can make existing business team redundant. At the end of the day, even data scientists are salespeople who are trying to prove their product is better than the existing ones.




Bachelors and Master in Technology from Indian Institute of Technology Kharagpur. Started working in analytics start-up before moving to HSBC Analytics. After spending 3.5 years at HSBC with stints in Bangalore and Chicago, I came back to India to work with Citi Risk Analytics team in Mumbai. Currently employed with M&M Group Strategy Office leading the data science team for Mahindra Finance.

Post a comment

0 Comments