Insight mixology

November 21, 2016

Designers, researchers and strategists are on a constant quest to gain insights and are increasingly leveraging big data and artificial intelligence (AI) to help find them.  

These methods are screamingly fast and can seemingly identify the “what if?” and “what the?” in mere hours.

Just five years ago, the standard methodology was to pack our bags and travel the country or the world to conduct a series of interviews or observations of target users or stakeholders. Observe, collect, analyze, synthesize – it could take weeks or even months to identify the big “aha.” Can we do it faster, cheaper and, of course, better with data? To answer that, I would like to look at a case study from Google.

Google now offers research services that leverage its data to identify insights so quickly it seems almost instantaneous. With its seemingly endless data stream, not only can it identify a trend through scrubbing users’ search patterns, it can pinpoint where and when the trend emerged.  

A recent case study features the tulle skirt. In March 2015, Google published a blog that chronicled several up-and-coming trends including the tulle skirt. It tracked the beginning of the trend to the Western U.S. in early 2014. The data revealed that Google searches of “tulle skirts” increased 34 percent from early 2014 to early 2015. The data was undeniable: there were increased keyword search rates for tulle skirts and the corresponding increase of 20,000 skirt-making videos posted on YouTube.  

Can identifying a trend be considered discovering a deep insight?  If I was a buyer, fashion blogger or boutique owner, knowing that there was increased interest in tulle skirts would be invaluable information. It would enable me to make informed decisions for the upcoming season. But if I want to really understand the “why” behind the increased searches of tulle skirt, I need to dive deep into the data or try to unearth the latent meanings manifested within the tulle skirt.  

I am a strong proponent of the Research Cocktail, a careful blend of data-based and human-centered research methods.

I need to leverage a mix of different methodologies to start unearthing some of the “why.” There is no secret recipe for finding the right mix of research methods. Many factors impact a project’s research mixology including skill sets, budget, schedule and objectives.  Many times we start from within, leveraging our empathy to relate to our users. As a female researcher with a bulging walk-in closet filled with all of the latest fashion “must haves,” I can easily self-identify with the “Female Fashionista.” I can relate to the allure of tulle, which evokes images of little girls pirouetting in ballet class, dreamy princesses and brides gliding down the aisle in clouds of gossamer. The tulle skirt becomes a metaphor for three stereotypical feminine ideals: Ballerina, Princess and Bride.

Even if we agree with the metaphorical significance of the tulle skirt, we may say, “So what?” If our desired outcome is just to understand how the trend emerged, we can return to the data and look for:

  • Other similar patterns
  • A singular significant event (such as a celebrity wearing a tulle skirt)
  • “Metaphorically” related events happening in the area at the appropriate time such as a visiting ballet troupe or bridal show.

Some careful research and data crunching may identify a probable cause or causes. We can then leverage this knowledge to help us identify future trends and build algorithms to help find them.  For example, in an article exploring Google’s trending research, the New York Times claims it tracked the tulle skirt trend back to a pair of Utah-based sisters with an Etsy storefront. But if we want to understand the significance beyond that of a trend and uncover the societal or psychological factors behind why women in the Western U.S. identified with tulle skirts, we need to dig deeper.

How? Here are some possibilities:

  • Go back to the source and leverage our data, using it to screen and identify our user interviews and observations. So instead of going to six states, we can identify two or three or design a more effective research protocol or write a better survey.
  • Dissect the demographics of those who like tulle skirts and use more traditional methods to probe on the allure of tulle and its role in today’s fashion.
  • Dive into the deeper meaning/metaphors of Ballerina, Princess and Bride, trying to understand why women continue to identify with them.  Can we find information on their historical, societal and philosophical significance?  On this quest, we can leverage past research on feminine metaphors, which will help us link them to our high-order emotional needs. For example: Ballerina > Dancer > Dance > Courtship > Joy.

Today’s research mixologist is blessed with an abundance of tools and methods. The sheer volume of data and the speed of analytics opens up immense possibilities – they are fast, vast and accurate. But I am a strong proponent of the Research Cocktail, a careful blend of data-based and human-centered research methods.  Exactly what methods you put into the shaker depend on your desired outcomes, but we must fight the impulse to lean on the immediacy of data and remember to add a dash of the bittersweet of human-centered research if our goal is to uncover the deeper meanings and user needs behind the “what the?” that big data can reveal.