6
NEVER ASK, UNEARTH
Bill Bernbach was a gifted, intuitive persuader and the father of the creative revolution in advertising. He said, “At the heart of an effective creative philosophy is the belief that nothing is so powerful as an insight into human nature, what compulsions drive a man, what instincts dominate his action….”1
Whether we are trying to get the public to change, trying to get voters to change, trying to get shoppers to change, or trying to get a family member to change, we need an insight into the compulsions and instincts that drive their current behavior.
But what is an insight?
Though we are not sure what an insight is, we sense its absence. In any organization, the surest way to undermine a proposed persuasion attempt is to complain that “It contains no insight.” The criticism is both deadly and irrefutable. Explanation can’t make an insight insightful any more than explanation can make a joke funny.
Jokes and insights, it turns out, have a lot in common. We can learn something from the comparison. Both jokes and insights are defined by their effect. A joke makes us laugh. Otherwise, it’s not a joke. An insight gives us the pleasure of a surprising truth. Otherwise, it’s not an insight.
A formula for jokes or insights doesn’t exist, but every good joke and every good insight has three qualities: They are unexpected, provocative, and true.
Unexpected
We’ve long known that jokes involve a setup and a surprise. A traditional definition of a joke and one that the noted humorist Sigmund Freud used is “bewilderment succeeded by illumination.”
A man gets a call from his doctor. The doctor says, “Thank god I’ve reached you. I have bad news and worse news.”
“Oh dear, what’s the bad news?” asks the patient.
The doctor replies, “The results of your tests came back saying you have only 24 hours to live.”
“That’s terrible,” says the patient. “How can the news possibly be worse?”
The doctor replies, “I’ve been trying to reach you since yesterday.”
Insights require the same setup and surprise, the same sort of “bewilderment succeeded by illumination.” I suspect that more jokes and insights fail from a weak setup than from a week punch line. Insufficient bewilderment diminishes illumination.
Provocative
Both a joke and an insight juxtapose the incongruous to make us stop and think.
An old man was sitting on a park bench staring at a teenage boy who had spiked yellow, red, green, and orange hair.
The young man says, “What’s the matter, old man…didn’t you do anything wild in your day?”
“Well…,” says the old man, “made love to a parrot once. Thought you might be my son.”
True
The element of truth gives a joke its edge. Truth seen from a new angle gives an insight its power.
A horse walks into a bar. The bartender asks: “So, why the long face?”
Persuasion requires an insight into the target, a fresh perspective that is unexpected, provocative, and true.
Where can we find that?
Language Camouflage
Let’s take a look at the rest of that Bill Bernbach quote. "At the heart of an effective creative philosophy is the belief that nothing is so powerful as an insight into human nature, what compulsions drive a man, what instincts dominate his action even though his language so often camouflages what really motivates him."
A person’s language hides his motivation. The camouflage isn’t meant to be deceptive. It’s not even voluntary. People just don’t know what motivates them.
To persuade, we need to know why people do what they do and what might cause them to change.
To find out, never ask.
Motivations don’t reveal themselves to frontal assault. People couldn’t tell us their motivations even if they wanted to.
Asking people why they do what they do, or how they choose, or what’s most important in their decision is a lousy way to find out. If we ask, we will get an answer, but we will likely get the wrong answer. People don’t lie. They just don’t know “Why?” But they think they do.
Because our decisions are made by or heavily influenced by the lizard inside, our automatic, nonconscious mental system, we simply don’t have access to “Why?” We are not conscious of “Why?” We cannot say how motives and perceptions that are invisible to us influence our behavior.2
Amazingly, even though we don’t know “Why?” we are sure we do know. If asked “Why?” we instantly answer. We are extremely good at making up answers to why we behave the way we do. We come up with the answers quickly and effortlessly and we believe them to be true.
We have learned that people can’t tell us why they do what they do from studies of brain-damaged people, studies of people under hypnosis, and studies of normal people.
The story of a patient called P.S. is one of the most fascinating.3 P.S. suffered from severe epilepsy. Because epileptic seizures can spread from one hemisphere of the brain to the other, doctors severed the right hemisphere of his brain from the left, a dramatic operation but one that improved his condition.
In a normal brain, there is a lot of cross-communication between the two halves. But the two halves of P.S.’s brain could not communicate. The surgery that severed the connection between the two halves of his brain diminished the effects of his epilepsy, but there were other consequences as well. For example, the left half of P.S.’s brain lost its link with language whereas the right side remained fluent.
These unfortunate circumstances did permit psychologists to do some interesting experiments.
Our right brain receives the information from light passing through our left eye. Our left brain receives the information from light passing through our right eye.
Psychologists rigged a system in which one picture could be shown to P.S.’s right brain (that is, shown to his left eye) and simultaneously a different picture could be shown to P.S.’s left brain (that is, shown to his right eye). The image shown to the right brain was a chicken claw. The image shown to the left brain was a snow scene. In a normal person this would be uncomfortable because the two halves of the brain would communicate and try to reconcile the different images. But with P.S., the two halves were not communicating.
Psychologists also gave P.S. a set of cards showing other objects. They asked him to pick the card that matched the picture he saw. His right hand, controlled by his left brain, which saw the snow scene, pointed at a shovel. His left hand, controlled by his right brain, which saw the chicken claw, pointed at a chicken. All of this made complete sense. However, when asked why he picked those two things, P.S. answered without hesitation, “Oh, that’s easy. The chicken claw goes with the chicken and you need a shovel to clean out the chicken shed.” His right brain, where his language capability resided, had only seen the chicken claw and did not see the snow scene. It did not know why one of his hands pointed at a shovel, but it immediately made up a rational explanation for his behavior.
People often don’t know why they do what they do. But, if asked, they will come up with an answer and believe it to be true. Unfortunately, the answer they come up with may have no connection to reality.
In a similar way, people do things under hypnotic suggestion and don’t know why. But these people quickly make up a rational explanation for their behavior and they believe the explanation they offer.
According to David Eagleman, “We have ways of retrospectively telling stories about our actions as though the actions were always our idea.” And, “We are constantly fabricating and telling stories about the alien processes running under the hood.”4
Richard Nisbett is professor of social psychology and co-director of the Culture and Cognition program at the University of Michigan at Ann Arbor. Nisbett and Timothy Wilson conducted a simple experiment in which normal people described why they made an everyday choice.5 The two scientists set up a table in the front of a Meijer’s Thrifty Acres just outside Ann Arbor, Michigan. A sign was on the table saying “Consumer Evaluation Survey—Which Is the Best Quality?” On the same table were four pairs of nylon panty hose arranged from left to right and labeled A, B, C, and D. Once people chose the one pair they felt was the best quality, researchers asked them why they had chosen that pair. In response, “People typically pointed to an attribute of their preferred pair, such as its superior knit, sheerness, or elasticity. No one spontaneously mentioned that the position of the panty hose had anything to do with their preference.” When asked, all respondents but one denied that position had anything to do with their choice.
In fact, all four pairs of panty hose were identical. Position was the only difference. As in previous research, people preferred the option on the right with 12 percent picking pair A, 17 percent picking pair B, 31 percent picking pair C, and 40 percent picking pair D. People chose one pair of panty hose out of four either on the basis of position or randomly. There was no factual basis to do so. They then immediately made up a rational reason for their choice, a reason they themselves believed.
Nisbett and Wilson found that people can only tell us what they think they know about how they think, but not how they actually think.
Market researchers are quite familiar with this phenomenon. Researchers distinguish between “reported importance” and “revealed importance.” Reported importance is what people say is important in their choice. Revealed importance is what is shown by analysis to be actually related to choice. What’s revealed to be important is often surprising.
Another experiment by Nisbett and Wilson concerns our ability to recognize why we are attracted to another person. If asked, we could surely say why we did or did not find the other person alluring. But do we really know why?
In a park in British Columbia, an attractive female assistant of Nisbett and Wilson approached young men and asked if they would be willing to fill out a questionnaire. The cover story was that the woman was involved in a class project on the impact of scenic attractions on creativity. After the young men filled out the questionnaire, the woman thanked them and offered to explain the study when she had more time. She tore off a corner of the questionnaire, wrote her phone number, and told each young man to call her if they wanted to talk to her about the details of the study. What researchers really wanted to know was how many men were sufficiently attracted to the interviewer to call her and ask her for a date.
The female assistant approached half of the men while they were on a narrow footbridge that swayed in the breeze over a deep gorge. She approached the rest of the men when they were seated on a bench after they had crossed the bridge. Which men were more attracted to the interviewer? Which men would be more likely to call her and ask her for a date? Would there be any difference? After all, it was the same interviewer for all participants.
Of the men approached on the bridge, 65 percent called the woman and asked for a date. Of the men approached while seated on the bench, 30 percent called her and asked for a date.
The experimenters theorized that the men approached on the bridge received the interviewer’s phone number when “their heart was beating rapidly, they were a bit short of breath, and they were perspiring.”6 The experimenters predicted that these men would misattribute some of their arousal to feelings of attraction to the woman. They were right.
If asked why they called to ask for a date, none of the men would have said “because my heart was beating rapidly when she gave me her number.”
The behavior of our body feeds back to our brain, but our consciousness is not in the loop. Our conscious mind does not know why we do what we do.
Okay, so people don’t know why they do what they do. What harm can there be in asking them?
Asking people “Why?” will send us in the wrong direction because bad information is worse than no information, a lot worse. We can’t seem to resist believing bad information even though we know it’s worthless. Kahneman’s research tells us that our automatic, non-conscious mental system will treat even bad information as if it were true.7 That is the way the system works and it works that way even among people who should know better. Even doctors, reporters, and scientists find it hard to resist information they know is worthless.
Despite the fact that “Why?” is a poor question to put to a respondent, “Why?” is still a common question in market research and political research. The boss wants to know the answer and those who are supposed to get answers can’t think of any other way to get them.
The answers pollsters get to questions about the qualities voters look for in a candidate are worse than useless. The answers are more likely to mislead than to illuminate.
In March 2015, PEW Research Center reported the results of their poll on the qualities important in a presidential candidate.8 By an almost 2 to 1 margin, Republican voters said that “experience and a proven record” were more important than “new ideas and a different approach.” But a mid-August poll by Fox9 found Trump to be the front-runner in the Republican primary, doing more than twice as well as any other candidate. Three candidates who never held elected office, Trump (25 percent), Carson (12 percent), and Fiorina (5 percent), were the preferred candidates of 42 percent of Republican voters. If Ted Cruz (10 percent) with his two-year Senate career is added, a majority of Republican voters preferred one of the four candidates with little or no political experience to any of the 14 candidates with extensive experience and records.
People don’t know what is important in their choice. Bad information is worse than no information. When the information is bad, we’ll end up trying to solve the wrong problem.
If we shouldn’t ask people why they do what they do, how do we find out the answer?
Unearthing True Motivations
We unearth “Why?” through observation or by simple analysis of people’s responses to other questions. We can unearth why people do what they do and what might get them to change if we come at the question indirectly.
We can begin by making a list of the most likely reasons why some people already do what we would like our target to do. These people may choose candidate A over candidate B, or use brand C over brand D, or recycle aluminum cans instead of throwing them in the trash, or they may have already stopped smoking.
Read about the issue. Collect the best guesses of others. Learn about research that others have done into “Why?” unless that research simply asked a direct question. Even if our target is one person or a small group of people and we have no money for formal research, we can still do some informal research.
Talk to the people who already do what you’d like. Just don’t ask them why they do what they do. Ask them what they think of the options. Ask them what they think of the people who do and don’t do what you’d like. Observe people who choose candidate A, or use brand C, or recycle aluminum cans, or who have stopped smoking. Observe with an eye toward understanding the desires that seem to underlie their behavior.
On the basis of what you have found, what do you think motivates people who already behave the way you’d like? What reward do you think they associate with these actions? At this point, your suspicions about their motives will likely be a good deal more accurate than if you had asked them directly.
If you believe the same reward that motivates people who already behave the way you’d like would also motivate your target, you can begin building that association.
Let’s say the action you seek is recycling aluminum cans and your target is a man who doesn’t recycle. You conclude from your informal analysis that people who do recycle do so because recycling makes them feel they are contributing to a cleaner environment. If you believe that same reward would motivate your target, you can start to build that association.
If the reward that others seem to get from the action you suggest won’t work for your target, you have to try something else. Continuing the recycling example, if the suggestion that he is contributing to a cleaner environment doesn’t look like it will motivate him, you have to find another reward.
Ask yourself what your target wants that they can get by doing what you would like. Look at those two circles—target desires and action outcomes—and examine the overlap. What does your target want that can result from the action you recommend? Don’t ask them what they want; observe. Ask others for their observations. Ask the target what they think of the options. Ask them what they think of the people who do and don’t do what you’d like.
If your non-recycler wants to lower the cost of government and you can make the case that recycling lowers the cost of garbage disposal, you may have found a useful approach to persuasion. Now the challenge is to associate recycling with lower cost of garbage disposal and raise the salience of both ideas and there is a good chance your target will become a recycler.
“Saving the earth” might motivate most people who currently use CFLs (compact florescent bulbs), but won’t motivate people who aren’t yet using CFLs. Look more closely at what those people think who do use CFLs, but aren’t interested in “saving the earth.” You’ll likely find that perceived efficiency is what motivates people who aren’t worried about “saving the earth” and efficiency is the way to grow CFL use.
Take a different case, one in which your target is many people. Let’s say you have a fast food restaurant called Deli Depot, and you would like your customers to recommend Deli Depot to a friend.10
You begin the same way. Make a list of the most likely reasons why someone would recommend Deli Depot. Read about the issue. Collect the best guesses of others. Learn about research that others have done into “Why?” unless that research simply asked a direct question.
As a result of all your background work (market researchers would call this “secondary research”), you have six hypotheses about why a customer might recommend Deli Depot to a friend. You hypothesize that the perceived perceptions that matter are 1) excellent food quality, 2) competitive prices, 3) competent employees, 4) friendly employees, 5) wide variety of food, and 6) fast service.
In a survey, ask a couple hundred customers about their perceptions of Deli Depot on each of those attributes, as well as how likely they are to recommend Deli Depot to a friend. Fortunately, that’s what the folks at Deli Depot did. In fact, they went further. In the same questionnaire, they also asked customers to rank the attributes from most important to least important in their fast food restaurant selection. The results are instructive.
Customers reported that “excellent food quality” followed by “competitive prices” are the most important criteria in their fast food restaurant selection. However, if you tried to guess how likely a person was to recommend Deli Depot to a friend knowing only their rating of Deli Depot on “excellent food quality” or “competitive prices,” you couldn’t do a very good job. You could do a much better job guessing how likely a person was to recommend Deli Depot to a friend if you knew only their rating of the restaurant on “friendly employees” or “competent employees.”
Reported importance—what people say is important—is very different from what is actually related to recommending Deli Depot to a friend. This is pretty common. When asked to consciously evaluate importance, people give the answer that appears most rational. But what really influences preference is often experiential qualities like friendly employees.
Of course, people care about “excellent food quality” and “competitive prices.” But Deli Depot and the other options people thought about were probably all good enough on “excellent food quality” and “competitive prices,” and small differences on those attributes were not very predictive of whether a person would recommend Deli Depot to a friend.
As we all know, a connection between a perception and likelihood of recommending Deli Depot doesn’t mean the perception caused the likelihood of recommendation. Correlation doesn’t mean causality. However, the absence of a connection, the absence of correlation, does mean that the perception in question has little persuasive value. You can use the lack of correlation to eliminate perceptions that don’t matter.
Look at the perceptions people hold that are related to whether they will or will not do what you’d like. If a perception is unrelated to what people do, that perception isn’t causing the behavior.
Take “speed of service” as an example. “Speed of service” is not directly related to the likelihood of recommending Deli Depot to a friend if:
1. The perception truly doesn’t matter. If people who say the service is fast and people who say the service is slow are equally likely to recommend Deli Depot, the perception of fast service doesn’t matter.
2. Few people hold that perception or its opposite. If almost no one says the service is fast or if almost everyone says the service is fast, the perception of fast service can’t explain why a sizable group of people are likely to recommend Deli Depot and another sizable group of people aren’t.
3. The perception has a more complicated relationship with likelihood of recommendation. A more complicated relationship would exist if people who think Deli Depot’s service is very fast and people who think Deli Depot’s service is very slow are both likely to recommend Deli Depot, whereas people who think Deli Depot’s service is neither particularly fast nor particularly slow are unlikely to recommend Deli Depot.
Whether (1), (2), or (3) holds, the perception of fast service is of little persuasive value. So the absence of correlation can be used to whittle down the list of possibilities.
When a perception, such as the perception of friendly employees, is correlated to likelihood of recommending Deli Depot to a friend, we still need to look further. If a perception is related to what people do, it might be the explanation. But you should ask yourself:
• Are both the perception and the behavior caused by some third variable?
• Is causality reversed with the behavior causing the perception?
• Are you looking at a halo effect?
If the survey sample was small and included many relatives of employees, those relatives might say both that the employees are friendly and they themselves are likely to recommend Deli Depot to a friend. The correlation you see could be caused by this third variable: respondent relationship. It should be easy for Deli Depot management to confirm that their respondents were indeed a random sample of patrons by making sure that employees didn’t know when the survey would be conducted and didn’t have any influence over respondent selection.
If causality is reversed, likelihood of recommending Deli Depot causes people to perceive employees as friendly rather than the other way around. It’s not likely, but worth considering.
The “halo” effect would suggest that if we like something, we believe a wide range of positive things about it even if we don’t have evidence for those positive things. But if one or two positive perceptions are related to likelihood of recommending Deli Depot and the others are not, then the halo effect does not explain what we see.
Customers who perceive Deli Depot employees as friendly and competent are more likely to recommend Deli Depot to a friend. Despite what people say about the importance of “excellent food quality” and “competitive prices,” customers’ perceptions of Deli Depot on those attributes are less related to likelihood of recommending Deli Depot. Under current conditions, a small gain in perceived friendly and competent employees is likely to make a bigger difference in recommendation than a small gain in perceived food quality and competitive prices. If we are trying to increase recommendations, it is clear where we should put the emphasis.
The idea that the best way to increase recommendations of Deli Depot is to increase the perception that Deli Depot employees are friendly and competent is unexpected, provocative, and true.
A couple of additional cases should clarify how it is possible to unearth the real reason “Why?”
Discover Card
This example of the unearthing process comes from when Discover Card was a relatively new credit card.
At that time, many people had a Discover Card in their wallet, but use of the card was still low. Management wanted people who already had a Discover Card in their wallet to pull the Discover Card out first, that is, make it their preferred credit card.
We began by making a list of the most likely explanations for preference for Discover Card among people who already had a Discover Card. We came up with several possibilities. Here are some:
• Low annual percentage rate. People who prefer Discover Card associate it with a low APR.
• Low annual fee. People who prefer Discover Card associate it with a low annual fee.
• Wide acceptance. People who prefer Discover Card believe it is widely accepted by merchants.
• Classy image. People who prefer Discover Card associate it with a classy image or, at least, not a blue-collar image. (Discover Card was originated by Sears and it was feared the working class Sears image was holding the card back.)
• Savings plan. People who prefer Discover Card realize Discover Card offers a savings account.
• Cash back. People who prefer Discover Card associate it with cash back on purchases. This was in the days before “cash back” became a popular credit card attribute.
We surveyed people who already had a Discover Card and asked their perceptions of the card and their credit card preference. We looked to see which perception of Discover Card was related to preference for Discover Card.
The proportion of people who made Discover Card their preferred card was the same among those who thought Discover Card had a low APR as it was among people who did not and the same among people who thought Discover Card had a low annual fee as it was among people who did not. The proportion of people who made Discover Card their preferred card was actually lower among people who thought the card was widely accepted. Apparently, people who preferred the card had tried to use it more often and had more experience with rejection.
The proportion of people who made Discover Card their preferred card was about the same regardless of their perceptions of the image of card users. The proportion of people who preferred Discover Card was about the same among those who thought Discover Card offers a savings account as among people who did not.
The only perception that was positively associated with preference was “cash back.” People who associated Discover Card with “cash back” were much more likely than others to make Discover Card their preferred card. After talking to people who didn’t yet prefer Discover Card about the idea of getting cash back, we decided that “cash back” was an association that would also motivate them. In retrospect, this may seem obvious. At the time, it came as a revelation. Previous advertising for Discover Card talked about a laundry list of rewards, unable to decide what was most important.
We recommended a program of communication that dropped everything else and focused exclusively on “cash back.” We wanted to build an automatic association between Discover Card and “cash back.” It worked. Discover Card grew six-fold in seven years.
Cheese
Promoting cheese is another example of unearthing “Why?” rather than asking “Why?” Why do some people serve cheese more often than others? You may think you know the answer, but you probably don’t.
We were trying to win the job of advertising cheese on behalf of the National Dairy Board (NDB). We assumed that getting people who reject cheese to start serving cheese would be tough. We decided instead to try to get women who only occasionally serve cheese to their family to serve it more often. Among people who serve cheese to their family, we wondered what explained how often they served it. Along with the National Dairy Board, we had a number of hypotheses:
1. Perceived nutritional benefits of cheese. Women who serve cheese only occasionally have a lower perception of the nutritional benefits of cheese (calcium, protein) than women who serve cheese frequently; that is, they don’t automatically associate cheese with its nutritional benefits.
2. Perceived nutritional drawbacks of cheese. Women who serve cheese only occasionally have a higher concern about the nutritional drawbacks (calories, fat) of cheese than women who serve cheese frequently.
3. Perceived cost of cheese. Women who serve cheese only occasionally see cheese as more expensive than women who serve cheese frequently.
4. Perceived taste of cheese. Women who serve cheese only occasionally have a less positive perception of the taste of cheese than women who serve cheese frequently.
5. Awareness of recipes using cheese. Women who serve cheese only occasionally can’t think of as many simple recipes using cheese as women who serve cheese frequently.
We interviewed a lot of women who serve cheese to their families. We asked them how often they served cheese and a number of questions about their related perceptions and knowledge.
We found that women who occasionally served cheese and women who frequently served it had the same perception of its nutritional benefits and its nutritional drawbacks. Both types of women were enthusiastic about the benefits of cheese and not too concerned with the possible drawbacks.
We found that women who occasionally served cheese and women who frequently served it had the same perception of the cost and the taste of cheese. Both types of women felt that cheese was worth the price and both liked the taste.
What distinguished occasional cheese servers from frequent ones was awareness of easy to prepare recipes that use cheese. The less-frequent cheese servers couldn’t think of many simple ways to serve it. This fit with another fact we had gathered: women who only occasionally served cheese were less confident and less experienced cooks than women who often served it. The real reason some women were serving cheese less often than others was not nutrition, taste, or cost, but the availability in their mind of simple serving suggestions.
We recommended that the advertising be used to give women those very simple serving suggestions. The NDB took our recommendation. The advertising, for example, said that adding a little melted cheese could make broccoli “disappear” and turn peas into “whoopeas.” It worked. The “Don’t forget the cheese” campaign got occasional cheese servers to serve it more often and, to our pleasant surprise, also got frequent cheese servers to serve it more often. Increasing the availability of easy cheese recipes encouraged all cheese servers to serve it more often.
If we had directly asked occasional and frequent cheese servers why they served it as often as they did, we would have gotten answers that made complete sense to the respondents and to us. But the answers may have had little to do with the real basis of behavior and we likely would have set off to solve the wrong problem.
The cheese experience also highlights the importance of aiming at the act rather than the attitude. Attitude wasn’t holding people back. The target had a positive attitude toward cheese—its nutrition, price, and taste. What was holding people back was the availability of easy serving suggestions. When we made those easy serving suggestions available, the lizard’s behavior changed.
Don’t ask people why they do what they do, or how they choose, or what’s most important in their decision. People don’t know the answer, but they think they do. Bad information is worse than no information.
Unearth the answer with some basic research, whether informal or formal. See what people who already act as you would like associate with the behavior you seek. Decide whether the same association would also motivate your target. If so, begin to build the association. If not, find something else in the overlap of what your target wants and the possible outcomes of the action you propose. Build that association.
People don’t intentionally mislead us. They just are convinced they know why they do what they do when they really have no idea. As Rogers and Hammerstein said, “Who can tell you why? Fools give you reasons. Wise men never try.” 11