Agile Learning Long Form Articles Segmentation vs. Stereotyping

Is this the worst article you’ve read today?

The answers to most questions teach us very little.

Let me illustrate using a question I found that was designed to help shy people network with their peers more effectively – presumably in some version of introvert hell, aka the “mixer.”

“What great movie have you seen recently?”

What’s wrong with that question? Well, it’s certainly better than many common alternatives: Where are you from? What do you do? What’s your sign? When did you first notice that growth on your cheek?

(Don’t laugh. I heard someone ask that last question at a networking event. And no, it wasn’t me. And no again, I wasn’t the unlucky soul with the growth.)

The point is, of course, to “break the ice” and “get people talking,” but more often than not, this style of question tends to deliver weak answers that don’t tell you anything meaningful about the other person.

Why not?

Try answering the question yourself. What comes to mind? If you’re like most people, you’ve seen several movies in the recent past. But beyond its lack of specificity, but what does “great” mean in the context of this professional setting?

Let’s say you enjoyed the recent movie “A Star is Born” starring Lady Gaga and Bradley Cooper. How will this new professional acquaintance interpret your answer? Do co-stars Andrew Dice Clay and Dave Chappelle trigger an emotional reaction in the other person (even if they haven’t seen the movie)? What does the person think about romance movies? Will that person view you less professionally?

Is there a correct answer to that ice-breaker question?

No, there isn’t. To say that the best strategy is to “be honest” and to “not care what others think of you” is silly and unrealistic. Humans are social creatures; we continually assess our status within groups. We can’t help it.

This positively-worded and seemingly innocent question triggers a quick and almost subconscious response: What is the social setting? What do I know of the other person? What can I guess from body language and non-verbal cues? What am I hoping to achieve at this meeting? What is my status in relation to the other person?

We want to be “liked” in a social setting. Our true answers (and our true preferences) are confounded with our social preferences. In other words, you aren’t likely to learn anything meaningful about that person. You learned their facade. You learned what they wanted you to learn. And you’re nowhere closer to that professional relationship you hoped to spark.

Want to change that?

Simply flip the question: What was the worst movie you’ve seen recently?

We may not always be sure of what we want or like (we can have lots of “likes” depending on the circumstances), but we’re all experts on what we don’t want and don’t like. Try it sometime. I’ll bet you’ll find the answer comes quicker. It’s less guarded. Before the other person can put up their defenses, you’ve actually learned something important about the other person.

Let’s find out why negative is so positive.


If positive questions are so poor at yielding useful information, why are they so common in our everyday interactions? Why are customer service surveys filled with them? Why do we hire people based on the answers? Why do we choose to go on a date with someone who matches our preferences?

Simple. Positive questions are easier to ask.

A positive question yields a positive response. It may not be useful, but it feels good. What’s more, because a positive question tends to spawn more positive questions (aka follow up questions to clarify the intent of the initial question), they create more data. In this case, like so many others, more data is not better data, it’s actually poorer data.

A negative question, by contrast, yields a negative response. The answer is more useful, but it takes courage to face negativity.

It took me the better part of 20 years of my professional career to learn that. I’ve been asking some version of positive questions my entire advertising and marketing career. I’ve been so consistently frustrated with the results that I harbored a deep skepticism for social science research of all kinds.

More often than not – in fact, much more often than not ­– the information was useless.

Where did I learn to ask better questions? From the most negative people on the planet: Trial lawyers.


“Most people completely misunderstand the purpose of voir dire. It isn’t to select jurors, it’s to eliminate them. The only way to know them is to get them to talk. Most judgements we make about people are just plain wrong.”

– Jeremy Rose, Ph.D., consultant for the National Jury Project

We don’t need a detailed explanation of the jury selection process in the United States. A brief summary will do.

The term voir dire (VWAH-deer) was originally a French phrase, “to see to speak.” The process is designed to examine potential juror circumstances and biases that might impact their objective review of the facts in the case. Of course, no one can put aside all their biases, but voir dire hopes to expose the most blatant ones that might impact a defendant’s right to a fair trial.

A brief example: A defendant is accused of robbing a liquor store at gunpoint. Voir dire questioning likely would focus on the following lines of questioning: Have you ever been the victim of a violent crime? Do you own a liquor store? What are your opinions about guns?

Lawyers from each side (as well as the judge) are looking for answers that might bias a potential juror for or against the defendant. That dichotomy is the critical point: Because the legal system in the United States is an adversarial system, the defense lawyer will have an incentive to eliminate a juror from the pool who has been the victim of a violent crime as possibly biased against the defendant. The prosecuting attorney has the opposite incentive. In reality, the both attorneys tend to “strike” (eliminate) potential jurors on either end of the spectrum, leaving a group of people in the middle who are more likely to see the facts impartially.

Read that again carefully: The goal is not to select the correct juror (your adversary will want to eliminate that same person), your goal is to eliminate the wrong ones.

It’s a negative system. To be successful in a negative system, trial lawyers are good at asking negative questions. In fact, they’re some of the best at it.

No one wants to come out and say they feel the “sin” of a liquor store attracts trouble, and that the owners “got what was coming to them.” Could you imagine a positively-worded question being effective? What’s your opinion of liquor stores? That question wastes time – time the trial lawyer may not have. The jury pool might have 15-20 (or more) potential jurors, all of whom need to be questioned.

Instead, good trial lawyers go straight at the crux of the issue. Instead of What’s your opinion of liquor stores? they are more likely to ask, Do you feel that the liquor stores shouldn’t exist? The first question will require multiple follow up questions. The second question will deliver a useful answer immediately.

Trial lawyers don’t have the luxury of wandering lines of questioning nor to over-sensitivity to your feelings. When their questioning fails to uncover hidden bias, their client goes to jail. Or worse.


If this all seems a little abstract or outside the realm of your everyday experience, let’s try a more concrete example. Anyone who’s ever taken a customer satisfaction survey will recognize the following problem.

Below is a data set of results common in a “Net Promoter Score” survey – the answer to the question, Would you recommend a product or service to a friend or colleague? on a 0 to 10 scale.


10, 9, 9, 9, 10, 2, 3, 8, 7, 7, 8, 8, 10, 9, 8, 2, 10, 9, 8, 7

The average of this small data set (n=20) is 7.7


What comes next is a strategic question: Is it better to focus on respondents scoring 7 and 8 (to increase their scores to 9), or is it better to remove the very low results from the dataset? The answer isn’t immediately obvious. If we count the 7s and 8s, we have eight opportunities for modest improvement. There are only three very-low results (2s and 3s). It seems like we should focus on improving the experience of the respondents who provided “almost there” ratings.


Simple math gives the counterintuitive answer:


Positive strategy: Work to improve all of the 7s and 8s to 9s, all else equal.

New average (assuming success): 8.2


Negative strategy: Remove the 2s and 3s, all else equal.

New average (almost certain success): 8.6


In other words, we would do better to focus on removing the negative responses. Why is that? In the first case, we had to improve eight average scores by hoping to understand the preferences of many more people. In the second case, we need only focus on three negative responses – why those ratings were low, and correct the problems; or failing that, choose not to continue serving those customers.

You may be tempted to think that focusing on middling results somehow “focuses on your strengths” and that more satisfied customers somehow are “more profitable” than less satisfied ones. In the first case, focusing on strengths only works effectively for respondents who score 9 or 10. In the second case, notions of satisfaction are often poorly correlated with profitability.

I think you now understand why.

The math shows us what trial lawyers already know: The middle is messy. Focus on the edges, especially the worst edge. Be negative.


Okay, enough with the storytelling and abstract reasoning. Let’s put negative questions to work. If all that seems a little harsh for everyday conversation, don’t worry, I’ll teach you how to tone it down a notch.

I’ve hunted down some of the more uninformative questions I’ve found in several professional and personal surveys.

Let’s use our newfound skills to fix them, shall we?


Customer Satisfaction:

Positive question: What did you like best about your most recent visit to our restaurant?

Why it won’t work: Even when you provide a list of options, or ask people to rank them, you’re likely to see a muddy middle of preferences without any clear sense of the emotional attachment to those preferences. If the purpose of the survey is to prompt changes, what action will you take on these results? Likely, nothing. You’re doing well, why change anything?

Negative question: What was the worst thing about your most recent visit to our restaurant?

Why it works: Even if people liked their experience overall, everyone has something that bothered them. Maybe the table was too close to the door, the bathroom was cold, or the water tasted funny. You wouldn’t learn that from a positive preference question. However, the answer to the negative question provides clear direction: Move the table. Install a heater in the bathroom. Filter the water.


Human Resources:

Positive question: What work environment do you do best in?

Why it won’t work: You just walked by a soul-crushing “open office” floor plan, what do you think people are going to say? Exactly.

Negative question: What work environment have you learned will absolutely not work for you?

Why it works: People are adaptable creatures. Depending on the work tasks and corporate culture, plenty of work environments could be manageable. What you really want to know are the few situations that the candidate will find unacceptable. The answer to that question will eliminate a candidate from consideration before either party invests too much in the process.



Positive question: What candidate do you prefer in the upcoming primary election?

Why it won’t work: Seeing a pattern yet? The obvious follow up questions that try to gauge strength of commitment and likelihood to vote inevitably begin to add layers of assumptions onto layers of assumptions. No amount of fancy statistics will get people excited about a dull candidate in a crowded field of similar contenders.

Negative question: What candidate will you absolutely not support in the upcoming primary election?

Why it works: In a general election, strong negative opinions could be used to your advantage, but in a primary election (designed to winnow down a large group to a smaller one) the middling candidate that emerges usually isn’t the strongest. You learn more about voter preferences by examining what they dislike. Negative emotions motivate action (aka voting) more strongly than positive ones.



Positive question: What do you want in a partner?

Why it won’t work: Relationships are more complicated and situation-dependent than we like to admit. I’ve seen dating apps that try to use “regression analysis” to determine the unique combination of preferences that best “match” you with an ideal partner. In reality, their algorithms often aren’t much better than random guessing, for reasons you now understand.

Negative question: What is unacceptable to you in a partner?

Why it works: Is smoking off the table? How about a person of a different faith? How about anyone under 6-feet tall? Instead of preferences, define a (narrow) set of knock-out criteria. The resulting matches are still random guesses, but now you can focus on how well you connect with that person without the expectation that they share your preferences.


Positive questions are wonderful social lubricant, but that’s often all they are. You keep the conversation going, you learn what you already knew, or worse, you think you’ve learned something useful.

Negative questions are incisive and instructive. Handled correctly and politely, they also can make fun dinner conversation. But more to the point, they sharpen conversations, they expose what people may have kept hidden, and they point to clear actions.

Using them will take some practice, but if you want better answers, there is no “worse” way to learn.


About Jason Voiovich

Jason’s arrival in marketing was doomed from birth. He was born into a family of artists, immigrants, and entrepreneurs. Frankly, it’s lucky he didn’t end up as a circus performer. He’s sure he would have fallen off the tightrope by now. His father was an advertising creative director. One grandfather manufactured the first disposable coffee filters in pre-Castro Cuba. Another grandfather invented the bazooka. Yet another invented Neapolitan ice cream (really!). He was destined to advertise the first disposable ice cream grenade launcher, but the ice cream just kept melting!

He took bizarre ideas like these into the University of Wisconsin, the University of Minnesota, and MIT’s Sloan School of Management. It should surprise no one that they are all embarrassed to have let him in.

These days, instead of trying to invent novelty snack dispensers, Jason has dedicated his career to finding marketing’s north star, refocusing it on building healthy relationships between consumers and businesses, between patients and clinicians, and between citizens and organizations. That’s a tall order in a data-driven world. But it’s crucial, and here’s why: As technology advances, it becomes ordinary and expected. As relationships and trust expand, they become stronger and more resilient. Our next great leaps forward are just as likely to come from advances in humanity as they are advances in technology.

Thank you! Gracias! 谢谢!

Your fellow human.