‘Big Data’? It’s Really Just About Listening to People

“Big Data” is a popular buzz-phrase in the online world these days, tossed around in settings ranging from brand marketing to web analytics to grassroots field organizing. Much of the time, it’s being used as a sales hook — i.e., “our product helps you leverage the power of Big Data for your campaign” — which makes plenty of people suspicious. Is “Big Data” the latest vaporous marketing trend that’ll wash across the land, leaving little of substance behind?

At CampaignTech last week, former Obama analytics guru Amelia Showalter used a great definition for “Big Data”, something that cuts through the bullsh*t to get to the heart of what we’re talking about. In her eyes, data is all about listening to people. Ideally, if we were trying to find out what would motivate customers (for a brand) or political supporters (for a campaign), we’d ask them one-on-one what they care about. But that’s not always practical, so we use data as a stand-in.

Sometimes we’re talking about data from polling or focus groups, where we’re using a small sample of our target population to get a sense of what the larger group is thinking. Other times we’ll use direct information: where do they live? What primary elections have they voted in? What campaigns have they donated to? What answers did they give to a canvasser’s questions? If they’re already on our list, where did they come from (i.e., through what channel did they join)? Which emails are they opening? What action links are motivating them to respond? Which website stories are they reading? Which are they ignoring? Which social media posts are they sharing, and which are falling flat? All of these data points are proxies for the bigger questions: what do people think? What do they want? What can we do to fire them up and get them moving?

Of course, the danger is that the data becomes more important to us — more “real” in some sense — than the people whose activities and interests we’re trying to use it to measure. This can lead to bad decisions, particularly if the data isn’t as representative as we think (one of the oldest ideas in computer science = garbage in, garbage out). As former Obama data manager Ethan Roeder said in the CampaignTech panel I moderated, “no targeting model is as good as asking someone how they feel.” And listening — really listening — to the response.

– cpd