Five Crucial Soft Skills for Successful Data Leaders

Data and data-backed analytics, including artificial intelligence and machine learning, are experiencing exponential growth. Businesses are deemed conservative if they have not yet started embracing new trends with data-supported analytics and machine learning. Executives encounter rising pressure from competitors to differentiate based on these contemporary trends. According to Business Insider’s 2022 jobs report, hiring for Artificial Intelligence specialists grew by 74% in the last four years. Harvard Business Review proclaimed data science as the “hottest job of the 21st century.” This trend has also revealed a dire need for a robust infrastructure built over cloud and big data platforms, causing a corresponding growth in jobs in these related areas.

With so many related job profiles and needs, the natural query that comes to mind is: Who are these people? Job descriptions are vivid and vast. We expect data scientists to be data engineers, data engineers to be data analysts, and data analysts to be data scientists. Feedback often received is that we not only need them to be technical but also business savvy. As a leader trying to keep pace with the business while building quality data technologies to scale, I frequently wonder what to anticipate out of all our different but similar data professionals. The similarities that come to mind are not simply being business savvy but having five crucial soft skills that lead to successful outcomes.

1. Thinking and Acting in HIIT mode

According to The Nutrition Source at Harvard, HIIT is a type of interval training exercise. It incorporates several rounds that alternate between several minutes of high intensity movements to significantly increase the heart rate to at least 80% of one’s maximum heart rate, followed by short periods of lower intensity movements. It is essential to keep a tab on the heart rate not to overdo anything. HIIT produces incredible endurance in the body and clears up the mind.

There are numerous occasions when data professionals experience the need to peak their thinking to 10,000 feet to understand the goal of their data exercise, and then bring it down to 10 feet of specificity, which many times involves interacting with real data. Thinking at a high level and acting at a low level should become muscle memory. The most crucial aspect here is that thinking at 10,000 feet and working at 10 feet is best done by the same person. Repeatedly, we see failed use cases where data scientists rely on data engineers even for exploratory data. A HIIT mode of thinking and acting will ensure success and pace.

2. Comfort In Ambiguity

There is seldom a data set that will give perfect results. The size of the data does not matter for the majority of the use cases. Many business processes produce incomplete or inaccurate data, and require making assumptions or cleaning it up before it can be useful. Every data professional must understand the need to clean the data quickly, then make a few valid assumptions in conjunction with business, and move on with analyses instead of going into analysis paralysis.

Selecting and interacting with business data is an art in many cases before science can be applied to it. If you are doing location analytics, ask yourself:  Do you need to see perfect zip codes, or is the county name better? What is the grain of your data set that can sustain the assumptions you are making and answer your analysis goal? Can you live with a few thousand rows with county names, or do you need a million rows with an exact address? I say it depends!

3. Resident Consultancy

From leaders to analysts, every data professional should be a “consultant'' in their minds, even if they are full-time employees. Consultants are results-oriented, fast-paced opportunists. For data use cases, business leaders are not always available, so taking the initiative and seeking opportunities to work top-down as well as bottom-up is key to adoption of these use cases.

The entire data department should be on the lookout for opportunities to find new avenues to make that one small incremental outcome and continue their storytelling. A significant difference between real consultants and resident consultants is that the latter are true subject matter experts. With that skill, they can present without preparing for months together. They have the power to continue their storytelling and always tie threads.

4. Curiosity and Graceful Questioning

Working with complex data from complex business events and operations could mean continuous clarifications and deeper inspections. Compared to app development, where requirements are usually robust, data professionals often face the need to continue asking follow-up and curious questions to build a powerful story or trusted data science model. It can be tricky to keep everyone interested while asking. Here are a few tips:

• Get access to the apps generating the data you are analyzing.

• Support your questions with your preliminary findings.

5. Persistence Over Perfection

No data science models are perfect; some are better than others. No visuals tell the whole story initially, but they tell better stories over time. Seeking perfection is a downhill slope and one that stops data innovation. Walking out of presentations that lead to curiosity and follow up questions generally result in incremental actions with solid wins over time.

About the Author

Deval is Vice President of Rocket Data at Rocket Companies, leading cross-functional teams that are responsible for data product development, thus allowing confident decision making and frictionless client experiences. Deval is a talented leader with a career encompassing 22 years of experience in product engineering and data analytics. She has worked across finance, insurance and manufacturing industries. She is a regular speaker at CDO Magazine and STEM events and at the National Diversity Council. She was named a 2020 “Woman of Influence” by Venue Cincinnati and a “Global Data Power Woman” by CDO Magazine for the past two years. She holds an electronics engineering degree and an MBA.