As we continue to navigate these uncharted times, the promise of analytics to inform and intervene with disease progression becomes clearer. Mobility mapping from cellular data streams illustrates migration patterns and whether social distancing is being practiced. Diagnostic testing data and hospital admissions provide concrete evidence of positive diagnoses. Search information and social news analysis provides a third set of information to better understand disease progression and migration.
But an unlikely hero has emerged to help predict the future course of COVID-19: the survey. For the past several weeks, Facebook and Google have been assisting the Delphi Research Center at Carnegie Mellon in collecting millions of surveys from platform participants. Respondents answer questions about COVID-19 symptoms, with over 150,000 individuals in the United States participating daily. The resulting data are critical in helping to illustrate geographic disease progression as well as fuel predictive forecasts.
Talking to some unstructured data and AI vendors, though, you’d think the survey was dead … outdated … past its usefulness … and everything you need to know is already available in social and conversational data. Yet Facebook and Google wouldn’t be serving as community panel sources for individuals to complete surveys about the pandemic if that was the case. They’d already have the answers from their own incredibly robust AI efforts and data. But in this case, relying on behavioral and unstructured data alone isn’t enough information to understand COVID-19 patterns. There are a number of reasons for this: increased representativeness of survey sampling versus unprompted social activity, lack of available testing results, incomplete coverage of opt-in location tracking, etc. but the net result is the same. Surveys are far from dead; they’re a critical tool in a more comprehensive data strategy to predict the course of a disease.
So where do surveys fit today? How should you think about building a comprehensive data strategy to answer the customer experience questions facing your business? The easiest way of effectively mapping data for analytics initiatives borrows the classic ‘5 W’s’ and an ‘H’ from journalism. To create deep insights and meaningful solutions from data, you need to know who, what, where, when, why, and how. The reality is that different data sources serve different needs in helping generate insights; no one data source or analytic tool typically contains the answers to those six classic questions. I’ve used the graphic below in dozens of analytic consulting engagements to help source and organize the data available to solve a problem.
The goal for each question should be to answer it with data that already exists in your organization and applications. You shouldn’t ask your customers to provide information that you already have. We typically find that you can usually answer almost everything except the why of consumer behavior, with the information you already have at hand and today’s machine learning and unstructured data tools. The why is where surveys still shine: understanding why people think, feel, and behave in the ways that they do.
So before you tackle your next analytics initiative, be wary of the thinking that surveys are dead for measuring customer experience. As we’ve seen in this pandemic, sometimes it is the only way to get the information you need to answer the big questions. Never negate the importance of talking to your customers or prospects and asking them critical questions. It’s a good way to get to the how and the best way to get to the why of understanding their needs and behavior.
Image source: Getty Images
