The (Statistical) Importance of Diversity

We often hear diversity as a requisite for modern business but why should it be more than a social norm?  Because it’s one of the best ways to build an organization that takes both a holistic AND agile approach in decision making.

Let’s take an imaginary interaction between Maria and Volkswagen.

Maria is in the market for a new car and heads to her local VW dealership to browse the new models.  She wants a safe, sporty drivewith storage space in case she decides to move across town, but nothing too flashy.  Oh, and it has to have the latest iPhone sync capability so she can jam on the road.  But today she’s in a hurry; it’s one o’clock, there’s a client call at two, and traffic was pretty bad.  The dealer has one chance to make a good impression.

Statistical Diversity copy.001

So he absorbs what he can with Maria’s vague requirements and picks up øor a segment of what she was trying to communicate.  Ømight be the bare minimal – a safe, sporty drive.  

Statistical Diversity copy.002

The salesman is also busy and has been hard pressed to sell more cars – sitting down with Maria to painstakingly examine every model’s pros and cons won’t fit his schedule and she may not be interested in that level of detail anyway. So, he calculates an offer with the first product that came to his mind when he heard sporty – the Bug.   

Statistical Diversity copy.003


Maria mentioned the other specs too, but the salesman misses these details because he’s biased to attach importance to some data points over others.  It’s worked in past sales, so why dig deeper?

Now this may not seem like a big deal at the moment.  But what happens if this sort of bias or gap occurs at scale, i.e. becomes systemic?  What if everyone at Volkswagen is looking at ø and responding around that finding, but not incorporating other sources of information?  

Statistical Diversity copy.004

If Volkswagen salesmen continue pairing women to Bugs, they might miss a better match (and higher value proposal) with other products.  A Passat coupe, for example, has more features, better storage, and maybe a higher safety rating, and still fits the desire for a sporty drive.  

Statistical Diversity copy.005

The true danger occurs when employees don’t know these blind spots exist.  Statisticians call this scenario a Type II error.  A false confidence prevails in your beliefs about the market and your understanding of consumers.  Since you weren’t including women’s opinions of the Passat into your model, it would be impossible to find you were off target.  The data points to the Bug, but that average understanding is flawed, it’s not the true mean!

Statistical Diversity copy.006

Incorporating a more diverse set of data would help the salesmen understand a more complete picture of market preferences and respond accordingly.  In statistics terminology, they would have a more representative sample of the (sub)population(s) which would include previously omitted, but relevant, information into their analysis.  The result would reflect a broader and more accurate view of possible solutions to provide consumers.

Statistical Diversity copy.007

But even this model is far too simplified.  In the digital era with thousands of touch-points across geographies and media there is an inherent mandate to embrace higher orders of complexity. One data point at the “new” understanding of “reality” isn’t enough – there is no longer a single, planned equilibrium in the market.

Statistical Diversity copy.008

Software could automate the parsing of different online communities or transaction histories, but ultimately the creative process of building and offering products for these clusters will rely upon human empathy and ingenuity.  In order to capture the variance of market characteristics, a firm needs a wide range of backgrounds in it ranks.  Diversity should expand beyond gender and race to encompass a broader spectrum of demographics including income, geography, and values. 

Statistical Diversity copy.009

An organization with a deep  portfolio of perspectives will innovate at scale  by engaging these groups with far more efficiency and accuracy than an org with a singular POV.  In the 21st century when market sentiment is changing at the drop of a viral video or a trending blog, business can be won or lost depending on how that information is absorbed by the organization.   But it’s hard to absorb new ideas and dynamic information if you only have one type of person working for you.

Diversity is more than a nice thought, it’s the new survival tool.