The obvious abilities of a data scientist center around technology, statistics and mathematics, but the very important talent of any practitioner in this field is storytelling. Cold hard data are just that, observations and features in machine learning parlance. The best way to convey the results of a data science project is not necessarily through a series of tables, statistics and plots. True discovery is found through a story.
As one simple example, last year I worked on a data science project for a marketing VP at a SaaS company. He had a good sample of leads data and the goal was to determine what sales process would most efficiently yield a close on a deal. I was able to provide a number of answers to this question in my final presentation, but after only a few minutes of working with the data in R I could tell him a quick story that he found illuminating. Through use of a distribution density plot, I determined that the sweet spot for number of prospect “touches” was 4.5. The distribution had values for “Number of contacts” well beyond that point, in fact there were some outliers exceeding 35! Apparently, the point of diminishing returns was well less than what the sales team was doing. As my story went, it would behoove the sales people to understand the lack of value of prolonging the sales process and instead moving along to fresher prospects. It was just a matter of finding that story deep in the data and being able to convey it effectively
Here is an article that discusses the importance of storytelling in data science further.