Data scientists, marketers and their advisors gathered at NYU’s Stern School of Business yesterday to discuss “The End of Intuition,” the increasing role played by data analytics in the advertising business and what role humans play in making decisions in the industry.
“Your intuition is wrong and you need to trust machines to do the marketing for you,” said Tom Phillips, CEO of media6Degrees, the analytics firm that co-sponsored the first annual ADS-CON (Advertising in and Data Science Congress) event with the NYU Stern Center for Business Analytics, in his deliberately provocative introductory remarks.
It was generally acknowledged that analytics are opening up vast new ways of understanding how people are consuming media, in ways that sometimes are counter-intuitive. But, in spite of that, computers can only produce good data when humans ask good questions, the experts said. Finding the balance between man and machine is not easy, however.
Foster Provost, Professor and NEC Faculty Fellow at the NYU Stern School of Business, spoke of the need for causal inference-the reasoning process in which we come to a conclusion that something is, or is likely to be, the cause of something else-to properly interpret data on social commerce.
Marketers assume for example, that if people “like” them and tell their friends about a product via social networks, they are influencing them to purchase the product. However that doesn’t take into account homophily, or the tendency of people to cluster with others who share their own values and interests. There are also confounding factors to take into account that may influence people to take the same action at the same time; for example, if it’s raining, many people pull out an umbrella.
Identifying which of these factors is actually responsible for a person taking an action is important in knowing how to approach them with advertising, Provost said. One requires targeting, to give the right ad to the right person, whereas the other is based more on giving people incentives for peer referral.
NYU worked with Facebook to determine whether virality could be built into products. It sent out different versions of product endorsements; one featured a personal invitation from a friend to check something out, another was merely a targeted advertisement on the side. Overall, the targeted ad reached a far greater number of people than the personal invitation.
However, the latter was found to lead to a far higher engagement rate among those who received it. “Big Data can answer many questions, but not without causal inference,” said Provost.
Randomized experimentation is another technique necessary for correctly assessing the role of social influence in marketing, said Sinan Aral, Assistant Professor of Information, Operations and Management Sciences, at the NYU Stern Center for Business Analytics. The Center has looked at how current ratings can affect future ones, in a world where “there is an explosion of digital social signals and rating systems,” according to Aral. One such analysis showed for example that online reviews that were randomly manipulated to be more positive than they actually were experienced a 25 percent boost in their ratings, whereas those that were randomly changed to be more negative than they actually were showed no such impact. The center has looked at such topics not only in the context of marketing, but also to influence behavior in general, such as to encourage people to undergo HIV testing or to combat election violence.
Sometimes companies are slow to react to what the data is telling them, according to David Edelman, principal at McKinsey & Co, such as in the case of marketing budgets not reflecting the reality of how consumers make purchases today. Consumers today go through a process of researching, evaluating, gathering and then giving feedback on a brand, a process in which they may possibly bond with the brand after purchase, Edelman said. Yet advertising budgets are still mainly focused on paid media such as point of sale and brand advertising, and not enough on creating and curating content designed to get customers engaged. “Every brand becomes a publisher and companies need to put more money into creating better experiences for consumers,” he said. He also pointed to the need for improved tools that easily enable companies to pull data together in real-time.
More generally, Duncan Watts, principal researcher at Microsoft and the author of “Everything Is Obvious,” addressed the dangers of depending too much on human intuition to predict the future, whether in marketing or elsewhere. “Everything is obvious once you know the answer,” he said. Common sense is a way for us to provide explanations for the world around us, yet our understanding is often misinformed and leads us to thinking we know more than we do. Only by analyzing how and when common sense fails can we improve how to plan for the future.
Marketers create personas to better understand their target audience and what it looks like. If marketers can understand potential buyer behaviors, and where they spend their time online, then content can be targeted more effectively.
What’s behind a successful data-driven marketing strategy?
One of the major challenges in the martech industry is getting the attention of prospects in a world where they are bombarded by content and emails on all sides.
Facebook is addressing one of the biggest missing pieces of its chatbot offering: analytics.