It’s getting harder and harder to discern what your customers are thinking because traditional forms of market research are losing their effectiveness:
- Fewer than half of all households have landlines, which means random dialing is less representative of the general population.
- Consumers and B2B buyers are inundated with surveys, both on-line and on the phone. They are less like to participate.
- Time poverty affects everyone, making it less likely that people will participate in surveys.
A decade ago when market research response rates were 35% to 40%, we could assume away non-response bias. But today, with response rates as low as 5% to 10%, we need to question whether we are getting results that represent the population.
Further, consumers and executives are getting more sophisticated when they do respond. They are more likely to be politically correct or disguise what they really do or how they really feel. Why did Nate Silver miss predicting the election of Trump? It may be that survey respondents did not want to admit, even to someone they would never talk to again—the market researcher—that they were going to vote for Trump. This seems to explain part of the gap between poll results and actual voting.
A possible ray of hope is presented by Seth Stephens-Davidowitz in his book, Everybody Lies. His thesis is that Google Trends analyses of internet searches provide a much more accurate picture of what people really think and feel than traditional phone surveys or questionnaires. He cites how Amazon and Netflix use search behavior analysis to create much more accurate customer behavior predictions than traditional research methods. And the digital world of the internet makes A/B testing and verification of search analysis findings much more feasible.
Netflix knows which movies you will watch before you do.
Remember when Netflix had you create a list of movies you intended to watch? You put down highbrow films—black-and-white documentaries, the winners of the foreign film Oscar. Well, they learned you wouldn’t actually watch them. So, they used the millions of clicks from other viewers and identified the ones who watched the movies you watched. They then recommended you watch the movies your mirror image viewers (the term is Doppelgängers) watched that you haven’t watched—often lowbrow comedies or romcoms. Turns out you typically viewed them. And with these suggestions you visited Netflix more often and watched more movies.
Stephens-Davidowitz goes so far as to assert that marketers should ignore what people tell them:
- Consumers say they don’t want to eavesdrop on their friends. But Facebook gets more and more popular.
- Consumers say they don’t want to buy products that are produced in sweatshops. But these products get more and more business.
- People say they want politicians to state their positions. But Donald Trump has ignored or reversed many of his stated positions—and his popularity goes up.
The internet gives you the opportunity to track what people actually do.
Randomized experiments have always been the marketer’s stock-in-trade. Marketers have designed test market experiments to evaluate new products, new pricing, new messaging, and new channels. But these experiments are very expensive. The internet makes these kinds of experiments inexpensive and fast.
Stephens-Davidowitz says, “In the era of Big Data, all the world’s a lab…. Facebook now runs a thousand A/B tests per day, which means that a small number of engineers at Facebook start more randomized, controlled experiments in a given day than the entire pharmaceutical industry starts in a year.”
A caution about measurability
There’s an old saying about measurability: “If you can’t measure it, you can’t manage it.” Stephens-Davidowitz cautions this is not strictly true. He cites the emphasis in education on testing. “We can measure how students do on multiple-choice questions. We can’t easily measure critical thinking, curiosity, or personal development.”
Socrates solved this problem with his questions. Robert Penn Warren had his poetry students memorize verse, so they could learn the sound and meter of poetry. Harvard Business School’s case studies enable students to learn strategic thinking.
So, while Big Data solves many of the challenges of learning what your customer thinks, it’s not the panacea.
“A combination of curiosity, creativity, and data”
In order for a researcher to claim causality, that is, that variable X causes variable Y, three conditions must be met:
- X must occur at a time before Y
- There must be statistically significant correlation between X and Y
- There must be a cogent theory that explains why or how X causes Y
Stephens-Davidowitz points out that Big Data provides many examples of the first two criteria. What is needed, he observes is that the researcher has “A combination of curiosity, creativity, and data,” that is, a plausible theory to go along with the millions of data points.
Some politicians oppose the use of science to predict things like elections or climate change, perhaps because the results can be inconvenient. The marketer, however, drives results with accuracy and therefore takes advantage of anything that generates insight about the customer, their needs, and what will influence them.
This kind of intuitive theory development, by the way, is what artificial intelligence (AI) in marketing is about: connecting disparate dots to better understand the customer.
And while it’s not a panacea for understanding what your customers are thinking, Big Data analysis of millions of internet searches, of Wikipedia, Facebook profiles, et al. combined with curiosity and creativity in intuitive analysis offers a new approach to improve our understanding of our customers, what they are thinking, and how they will respond to our marketing programs.
These new approaches couldn’t be coming at a better time.
See how data from Google Trends can be used to inform the creation of an effective branding strategy that includes some form of diversity messaging in this Google Trends Study.