Your Next Hidden Opportunity: Finding the Human Angle in Metadata
by Justin Massa
Spend any amount of time with a toddler and you’ll likely observe them sorting things. My daughter loved to do this, and the way in which she sorted her toys, markers, books, or random stuff she picked up was always fascinating. “Things that should be able to fly” and “things that would look better in purple” were among my favorites. She would line them up, inspect them, tweak their organization, and then eventually pick them all up, only to be re-sorted again.
Kids who creatively sort items are actually playing with metadata. They’re organizing toys through unique or invented types of attributes of an object, arranging them into neat rows and columns. “These are all dinosaur fingernails (aka semicircle blocks), from biggest to smallest.”
What does that have to do with metadata?
It’s long been my belief that you can tell the data expert from the casual data enthusiast by how quickly they ask for the metadata. Put short, metadata is “the data about the data.” It’s everything from the filename and extension to the “data dictionary” that explains a collection methodology or individual datum naming conventions. Column headers are the metadata, the value in each individual cell is the datum. (Side note: Using terms like “datum” and “these data” rather than “this data” are other surefire ways to identify data nerds.)
Metadata is how we know that an EKG reading is describing a human rather than a horse. Without metadata, most data is useless.
While most of the focus is on various ways of visualizing the 1s and 0s that make up a data set, there’s unexplored opportunity for designers to add, subtract, and create types of values that might be assigned to data points — the metadata. If we apply design thinking to how we sort and classify objects or behaviors, it opens up an entire world of insight that typical approaches to data and analytics will never uncover.
Taking It To The Next Level
Food Genius* has begun experimenting with with creating and assigning new types of attributes in the world of foodservice (restaurant) distribution. Rather than simply viewing products by the types of attributes that a manufacturer provides, such as “flavor” or “format,” they are combining multiple data sets to derive wholly new sets of attributes based on how products are actually used in the market.
Put short, they are merging product data with restaurant menus and re-attributing products based on their menu item application. Honey mustard dressings get new types of attributes (aka column headers) such as “primary application” or “specialty application” with values (aka individual cells) like “dip for chicken” or “sandwich wrap spread.” These new types of attributes enable you to make comparisons across all other products based on their utilization, which has implications for everything from suggestive selling to SKU rationalization to product design.
But imagine pushing beyond the explicit usage of a product. What if we were to start to assign different types of attributes to behaviors or objects — attributes derived through a human centered lens? What if we re-classified behaviors or products not by the type of action or product but by the need fulfilled by the action or product?
Paying your utility bill early because you just got paid is about “uncertainty in my financial skills,” or purchasing a portable power bank to charge your phone is about “fear of being out of touch.”
By blending how we classify qualitative observations with the metadata of quantitative datasets, it becomes possible to see the world in an entirely new light. It shifts how we leverage data away from optimizing many tiny interactions to serving as the foundation for wholly new products or experiences. It’s less about optimizing clicks and more about designing web services. Less about eye tracking scores on a planogram and more about creating an entirely new in-store experience.
Take, for example, the “childless adults purchasing children’s chicken nugget” behavior that fellow IDEOer Arianna McClain describes in What Chicken Nuggets Taught Me About Using Data to Design. With the right data and tools, you can explore the distribution, frequency, and penetration of the purchase of children’s food by childless households as well as its correlation with other behaviors. Extrapolating the unintended new market for marketing children’s food to childless adults isn’t too difficult.
But as another colleague, Matt Cooper-Wright, points out in Are You a Good Driver? How Designers Use Data to Get to the Truth, there’s a brutal honesty in quantitative data. If, for example, an analysis of the data suggest that the outlier behavior of childless households purchasing children’s food is also accompanied by the purchase of exceedingly decadent adult food items — such as Ben & Jerry’s Chocolate Fudge Brownie (my personal favorite) — then the story becomes even more nuanced. It’s exciting to imagine what happens when you iteratively experiment with metadata derived from human factor research to then highlight new areas to dig in.
No matter how skilled the researcher, there are inherent limits to human observation. Just as there are behaviors that can only be uncovered through human observation, there are behaviors that only data analysis can uncover. But to uncover these insights compels us to rethink the entire data “stack”: what types of data we use, how we visualize it, and — most importantly — its metadata.
While all of what’s described in this column is possible, much of it is still quite challenging. Sourcing the right mix of raw materials, talent, and tools along with the time and budget that such an iterative process requires, can be significant obstacles. And the allure of optimizing clicks or eye tracking scores frequently outweighs the murkier challenge of deriving metadata in service of designing wholly new products or experiences.
But a singular focus on optimizing tiny behaviors is a game of diminishing returns. Delivering products and experiences that satisfy deep human needs uncovered through an iterative process of qualitative and quantitative research is how you change the game.
*Disclaimer: I was the founding CEO of Food Genius and remain on the board of directors. IDEO hosted Food Genius as the first Startup-in-Residence and is an investor in the company.
