Serving Up World-Class Analytics


By Ben Olsen, CEO, Analytics Guild

Cooking world-class food takes both teamwork and technique. The dishwashers, the line cooks, the executive chef, “everyone is part of this team, part of the success,” says Susan Feniger, chef and Top Chef Masters Alumna. The creative execution of the food is a “huge part,” she says, “but the people coming to our restaurants and the people working with us side by side create the dining experience.”

Analytics involves both sides of an “analytics experience,” too: the creators and the consumers of data products. In order to create data products, ingredients must be gathered (data sourcing), cleaned (data cleansing/manipulation), combined (aggregation, fusing/joining), plated (dashboarding/reports/analysis/presentation work), and then given to your users to consume (delivery/communications). Your users consume the dashboards, analyses, and scorecards that your team produces. And even then, the experience isn’t over. Users decide when to “eat” or not, managers give feedback to development, and repeat visits to your data product prove that you have something special—maybe even something world-class. 

For professionals seeking mastery it’s not just about creating products that are good, but the best possible representation of the ingredients in taste, look, and feel, and furnishing each dish with the proper timing, lighting, and context. Every detail is accounted for.

All these details related to data consumption can get overwhelming for teams creating data products. That’s why applying good decision-making frameworks not only to the consumption of data (see our prior post about the origins of the framework and the consumption side here), but to the creation of data products is key to cutting through the noise in data preparation and delivery.

Widen your perspective:
The dishes that you and your team can create are as varied as there are regions, cultures, continents. So this exercise is best done in the first phase of creation--prototyping, investigation, experiments--to avoid analysis paralysis. Once you have a minimal viable product, continue on to the next step.

Questions to ask:

  • Is the data the only data you need? 
  • What chart types could display your data? Would another chart type be more efficient in communicating results?
  • How will you deliver your data? Which medium is the best, even if it’s different than those used in the past?
  • Are there other metrics that are better than the ones you already have?
  • Does your team have alternative recommendations?

Reality-test your assumptions:
Chefs create and then test before adding an item to the main menu, in a test kitchen, menu specials, or test nights. Ben Shewry, chef at Attica, one of the top 50 restaurants in the world, has “Experimental Tuesdays” to test new recipes with actual users. Get your dashboard out there to a small and friendly group--maybe it's a trusted co-worker or someone in a different function. Get their true opinions.

Questions to ask:

  • Do the data say something contrary to your intuition or experience? Why? 
  • Are there any users you can test out your product with before going to market? 

Attain distance before deciding:
In restaurants, the job of working expo (food expediter) is critical. They are the last line of defense before the food goes out to a customer. They don’t cook the food, but they make an assessment before it goes to the table. Make sure that you perform a similar “expo check” on your data products.

Questions to ask:

  • Have you found and built the most important information?
  • Would your audience want to see anything else? 

Prepare to be wrong:
Have a backup dish! And if your customer doesn’t like their food, make sure to fix it to their taste. Occasionally, warning labels are also good to have for that inevitable audit or tricky situation.

Questions to ask:

  • Do you have alternative reports? Supporting documents?
  • Do you know if your data pipeline is healthy? Should you write any disclaimers, make any caveats?
  • Did you show the unfavorable results with the good?  

Just as the best chefs in the world have mastered the fundamentals of cuisine, analytics professionals must master best practices in visualization, statistics and data science, business intelligence, and more. This framework is a container for solid decision-making throughout the fundamental processes, getting the team that much closer to analytics mastery: the means to innovate, delight customers again and again, and stay true to the deeper purposes of the practice.