Storytelling

I love telling stories, and I'm learning daily how many fields and industries rely on compelling storytelling to succeed.

I studied marketing in grad school, but it wasn’t until after graduation that I really “got it.” Going in, I loved the story-based aspect of marketing, so that’s where I focused. Advertising, brand strategy, product marketing – they’re all different ways of telling stories.

I love telling stories, but I’m not always good at it.

For a while I wanted to write a book about marketing and storytelling. The idea was to examine the three dimensions of the marketing story:

  1. The story of your customers without your product
  2. The story of your product/business
  3. The story of you and your customers working together

The idea seemed sound at the time, but storytelling is far more nuanced than that. My exposure to data science has driven that point home because, at it’s core, data science and analysis is also all about storytelling.

Interpretation

One of my undergrad degrees is in political science. We studied a lot of history but often used historical statistics to describe events and motivations.

A specific professor confused me a great deal during that program. I took two different courses from her; though I can remember exactly which ones, I do remember two specific lectures she gave.

In one, she cited a statistic in a specific community in Northern Africa: the average age at which couples married. Her argument was that higher marriage ages implied people lived with their parents longer due to improving economic conditions. In her models, women stayed at home longer to focus on their education, deferring both marriage and children.

In the next quarter, tho, she used the same statistic to imply the exact opposite. People would stay with their parents longer due to economic uncertainty, deferring marriage and children because they lacked the means to start a family.

The same statistic. The same professor. Two different yet equally compelling interpretations.

Data Science

A similar issue came up when working with some data scientists to solve a problem. Due to conflicting guidance from business leaders we had only a fuzzy definition of success for a project. Rather than let perfect be the enemy of good, we proceeded with the fuzzy definitions and ran a few experiments.

The problem was that, depending on who we were talking to, the experiments were either a success or a failure.

The goal was to increase throughput of a specific system. The fuzziness in the goal was two-fold: over what time horizon did the increase matter and how did we measure throughput?

On the one hand, we had definitely improved overall delivery over the one-month period of the experiment. It was an overwhelming success.

On the other hand, long-term forecasting showed we’d sacrificed long-term delivery on a one year horizon in favor of immediate returns. The increased rate of short term delivery also amounted to higher volumes of work-in-progress for the system. In other words, it was an overwhelming failure.

Ultimately those in the second camp won out and we disabled the new system in favor of rolling back to a proven yet unreliable system and went back to the drawing board. The benefit of hindsight has taught us to clearly define metrics of success before proceeding, even if that means a project needs to be put on hold.

Data Storytelling

Whether it’s telling the story of a system experiment or explaining the implications of historical statistics around marriage rates, compelling storytelling is vital to communication. Statistics and raw data are completely objective, which is great. But objective truth tells us nothing without subjective interpretation.

Lies, damned lies, and statistics can be used to strengthen otherwise weak arguments. On the flip side, strong arguments are required to provide context and meaning to the statistics and data from which they’re derived.

Data science is something data scientists do. They study the data, explore the story it’s telling, run experiments to verify their hypothesis, and craft a compelling narrative around it. This is the crux of the field, and is a far cry more difficult than even the most sophisticated linear algebra or higher math that goes into the analysis itself.

I’ve spent most of my career learning to be a better storyteller. It’s not until somewhat more recently I learned just how critical that skill really is.