A smile is not a yes: why faces and pulse mislead customer research

A respondent in a focus group smiles at your product. You write down “liked it”. In fact the person wants to please the moderator so the session ends sooner and they can go home. The link between a facial expression and an emotion is far less direct than it looks. One and the same smile can mean joy, politeness, awkwardness, suppressed irritation or a plain wish to escape. The face shows what a person is ready to show, which is not the same as what they feel.

That breaks half of the methods companies use to “read” a customer. A foreign company entering Russia meets the problem in a sharper form, because it reads faces through a cultural filter it did not grow up with. Below: why neither the face nor the pulse gives you the emotion, and what to read instead.

What a smile actually means

Take one movement: the corners of the mouth go up. How many meanings does it carry?

Joy at a gift. A polite smile for a cashier. A tight grin when someone is angry and keeping a straight face. A smile of submission in front of a boss. Irony. Nerves at a job interview. Masking: bad inside, “all fine” outside. A greeting ritual. Embarrassment. And a plain reflex when someone is tickled.

Ten different states from one movement of the muscles. If you sit behind the glass of a focus group room and log every smile as a plus, you are putting into one column things that have nothing in common.

A case from negotiations. A client nods and smiles while you present. The salesperson is pleased: it landed. A week later the deal falls apart without explanation. What happened? The person was smiling at the end of an awkward meeting they had agreed to because refusing felt uncomfortable. The smile was about closing an uncomfortable situation, and interest had nothing to do with it. Taking one signal for an answer cost a month of work.

Paul Ekman, the best-known researcher of the face, spent decades building a catalogue of facial movements. His system describes which muscles work during each “basic emotion”. It is an enormous body of work, and it rests on one assumption: one configuration of the face equals one emotion. Research over the last twenty years has shaken that assumption badly. People read the same face differently in different situations, and different people show the same emotion with different faces. There is no universal dictionary where a muscle equals a feeling.

Why this matters more when you sell into Russia

If faces are ambiguous everywhere, they are doubly ambiguous when the observer and the observed come from different display habits.

Foreign teams often notice that Russians smile at strangers less than people do in the United States. A polite public smile carries less weight here, and a serious face in a meeting often means attention. Several things follow.

A neutral face in a Moscow focus group is weak evidence of rejection. Read through a home-market lens, it looks like boredom or dislike, and the team concludes the concept failed.

A smile may carry the opposite risk. When a Russian respondent does smile broadly at a moderator, it is worth asking what situation produced it. Was it the product, or the moderator’s joke, or relief that the hardest question is over?

Software trained on faces from other populations brings its own reading habits with it. Whatever it reports, it reports through the dictionary it was built on.

This is no reason to give up on research in Russia. Treat the face as one signal inside a larger picture.

Why the pulse does not give the emotion away

The problem is wider than the face. Every bodily reaction behaves just as unreliably.

A racing pulse when you meet a bear: fear. The same pulse when you meet someone you love: joy and excitement. The same pulse in an exam: anxiety. The same pulse after a run: load, with no emotion at all. One body, four different states.

The conclusion is simple and inconvenient: you cannot determine an emotion from physiology. The body reports one thing only, whether the system is aroused or calm. How hard the heart is beating is like the volume of a signal. What the signal is about, the body does not say. The meaning comes from the situation the person is in and from what matters to them. Without that frame, a pulse, sweaty palms and dilated pupils mean nothing specific.

The same trap exists on the web. A visitor stays on the pricing page for a long time. The analyst sees engagement. In fact the price put the visitor off and they were hunting for a way to leave. Long time on page, like a racing pulse, is volume, not meaning. The meaning comes from the neighbouring numbers: what happened before, where the person went next, whether they returned. In Yandex Metrica, Webvisor recordings show that sequence, and a foreign team should watch them before trusting a dashboard built on Google Analytics habits.

The experiment that showed the core of it

There is an old experiment that still explains the point better than any theory. Psychologists Stanley Schachter and Jerome Singer, early 1960s.

Participants were injected with adrenaline: the heart beats faster, a jittery feeling appears. They were told it was a harmless side effect of a vitamin. So the body was aroused and the person had no explanation for it. Then each was seated in a room with an actor. In one condition the actor fooled around and joked. In the other he was angry, indignant, shouting.

Those who sat with the cheerful actor described their state as joy. Those who sat with the angry one described it as anger. The bodies were the same and so was the adrenaline. The “emotion” differed, because each person looked around and completed the explanation: “I feel odd, and people here are having fun, so I must be enjoying myself”, or “I am shaking, and there is a row going on, so I must be angry”.

An emotion, then, is how a person explains the state of their body to themselves by looking at the situation around them. The body supplies the raw material: energy, tension, agitation. The situation decides how that material is read and named.

For a marketer this is the rule worth keeping. The same excitement from your ad will be called “interesting” or “annoying” depending on the frame in which the person saw it. The context you build around a message often matters more than the message itself. For a foreign brand in Russia the frame includes things the home team rarely controls by default: where the ad appears, in Yandex search or on VK or in a Telegram channel, what surrounds it, which currency and which tone of address it uses. Translating the message and leaving the frame to chance means the excitement it creates can be named the wrong way.

Why a “lie detector” does not catch lies

The polygraph measures the same things: pulse, breathing, sweat, skin conductance. It records precisely when the body became aroused. It cannot say why.

The person lied. Or feared they would not be believed while telling the truth. Or remembered an unpleasant scene. Or hated the sensors on their fingers. One reaction, a dozen possible causes. A polygraph is an arousal detector; the rest is added by the operator, so the conclusion is subjective by design.

In marketing this trap costs money. A respondent rates an ad from one to ten and gives an eight, because the ad caught them, or because they did not want to seem difficult, or because they were in a hurry to close the form. The eight says nothing until you know the situation in which it was given.

What to read instead of the face

You watch a client or an employee and want to understand their reaction. From the face alone you will learn nothing reliable. A smile in a product test can mean “I like it”, “it is awkward to refuse” or “I want the moderator to like me”. A frown can mean “I dislike it”, “I am concentrating”, “I cannot see the screen” or “I have a headache”.

Instead of guessing, look at three things at once and find where they agree. First, the expression. Second, the situation: where the person is, what came before, what they were asked. Third, the whole body: posture, where the torso points, what the hands do, where the eyes go.

A smile with relaxed shoulders and a lean towards the product looks like real interest. The same smile with a tense back and a glance at the door is politeness from someone who wants to leave. Same face, opposite conclusion.

So treat the neuromarketing promise “we read emotions from the camera” soberly, and ask whether the tool measures emotion or arousal. Most of the time it is arousal.

An emotion is a bundle of four things: what happens in the body, what it means to the person, the situation, and how it came out in voice, posture and face. The face is the last item and cannot be read without the first three. Reading a feeling from the face alone means guessing the whole film from its final frame.

What to do in advertising and product work

Do not trust single signals in tests. One click, one smile, one like does not tell you whether something worked. Look for several signals agreeing: saw the ad, reached the site, came back, got to the cart. When signals converge, they can be trusted.

Manage the frame along with the message. Remember the adrenaline experiment and think about the situation in which the person will meet your words.

Do not buy emotion reading from faces until you know what the device measures. The name of the emotion is usually written in by an algorithm or an operator, which makes it a guess.

For a company selling into Russia from abroad, add one more step: before you interpret any reaction, check whose reading habits you are applying. A home-market dictionary of faces, scores and dashboards will quietly mistranslate Russian behaviour. Whoever confuses the last frame with the whole film will keep taking politeness for interest.

Frequently asked questions

Sources

• Russian version of this article on belokrylovo.ru: Мимика и эмоции: почему лицо врёт