Learning more by asking less

Like everyone, my Insights team at PepsiCo is witnessing the disruption impacting billions of people around the world. We’re also trying to imagine how it may transform our industry—permanently—when the crisis abates. For now, as a F&B company, our job is to ensure we continue to meet the needs of families, whose eating and shopping routines have been completely upended. 

It’s counter-intuitive, but maybe some good may come out of this experience. The crisis is compelling companies to innovate and adapt in ways previously unimagined. This is happening all across the value chain; as my team strives to adapt real-time, we’re deploying a broad set of insights tools, from analyzing shifting sales data by channel to using empathy to understand how consumers are thinking and feeling as the crisis unfolds.  

Ultimately, we will need to look forward – to foresee what people will want and need, and how they will behave in the post-COVID world, to inform our company’s actions. This unique situation is shining a light on something we as researchers have known for a long time – asking consumers what they want or need or how they intend to behave is not enough.

For decades, market researchers were focused on asking. We asked questions. Tens of millions of them: in grocery store aisles, in focus groups, in letters, phone calls, in emailed and online surveys. 

Unfortunately, despite significant methodological leaps made since the market research industry began in the 1930s, the “ask” model has inherent limitations.

The first is bias. Despite efforts to scrupulously structure interviews and questionnaires, the fact is that every question introduces bias in some way. The way questions are processed varies from person to person. And then there’s our very human tendency to give opinions we feel compelled to offer versus those we genuinely hold, as in 2016 when US pollsters asked the same old questions about whom the public intended to vote for. 

Second, many of our most powerful questions simply can’t be answered accurately by people. Respondents don’t deliberately mislead surveyors, but much of what drives behavior lies below our conscious awareness. Our System 2 brain will do its very best to explain what motivates System 1, but most often it can only offer a logical answer, not a psychological one. Human behavior and motivations are complex and unpredictable—and do not yield willfully to surveyors’ models.

Third, when we ask, we put people into an artificial, out of context space. Even completing a survey in a home environment requires people to make predictions without knowing what will really influence them in the moment, or to explain their ‘typical’ behavior without remembering all the specifics that drove it. A lot is lost.

Fortunately, we’re finding new ways to find out what really makes people tick. Today, researchers are delving more deeply into the “say-do gap” where stated preferences are transformed into actual choices—between the first stirrings of an impulse (e.g., “I’m thirsty!”) and the final decision (e.g., “I want a Gatorade Zero!”). And they aren’t gaining these new insights by asking—but by watching. To that end, the 400-plus members of the Insights team I lead at PepsiCo are convinced that “seeing is the new asking.” 

PepsiCo and our research partners, with consumers’ (often enthusiastic) approval and cooperation, are using video technologies to show us how behavior unfolds, and decisions are made moment-by-moment.

Not only can we gain deep, unbiased insight by observing a single household, but increasingly we can in real time aggregate multiple disparate video inputs—mobile uploads, wearables, webcams, fixed cameras, social media—across households, and transform them into actionable, scalable data. In this sense, we’re scaling behavioral insight, and turning it into a new stream of big data. A relatively small sample of 200 households generates over a quarter million data points, and the ROI is expandable, as one project can be mined and re-mined across multiple brands, categories, and even markets.

With one of our research partners, @Big Sofa Technologies, we are aggregating video observation to “Quantify Everyday Life.” The experience is helping us in several ways. Most notably, we’re uncovering what we call “obvious invisibles,” wherein consumers (again, with their full permission of course!) reveal answers to questions we would never have even thought to ask as we watch them live out an un-voiced frustration with a product or implement a workaround that’s so habitual, they forget they do it. This is where we find the unvoiced need; this is where we see the true context that shapes behavior.

Take the example of Charley, who we saw bring a full lunchbox home to his Dad after school one day. If we had asked Charley’s Dad what he looks for when packing his son’s lunch, we likely would have heard some preferences around packing something healthy, something that will keep Charley full and get him through his day. But when we shared this moment back with him, we learned that lunch and recess happen at the same time, and the longer you spend eating, the less time you can spend playing. The ‘obvious invisible’ in this case is that Charley also needs something that can be eaten quickly and easily so he can get where he really wants to be.

By seeing instead of merely asking, we’re going to get a lot better at anticipating consumers’ needs and communicating and innovating in relevant ways.