Psychology

All the models in this course (AIDA, ELM, STEPPS, archetypes) stand on one foundation: people are not calculators. We decide fast, with feeling and shortcuts; reason comes afterward and writes the justification. Psychology in marketing is not an add-on; it is the bedrock.

That this is science and not just an impression was confirmed in 2002 by the Nobel Prize in Economics for the psychologist Daniel Kahneman (Wikipedia). From the 1970s on, Kahneman and Amos Tversky showed that people break the rules of rational choice predictably, and what is predictable can be used. The second pillar of this lesson is Robert Cialdini: according to Wikipedia, his book Influence (1984) has sold over 5 million copies, and it now contains seven principles of persuasion.

The third pillar is attention: there is little of it, it is broken into pieces shorter than a minute, and every message competes for it. The numbers come in the third part of the lesson.

Daniel Kahneman summed up forty years of research in the book Thinking, Fast and Slow (2011). He describes the brain as two systems. System 1 is fast, automatic and emotional, and it runs all the time at no cost: it recognizes a face, dodges a car, decides it likes the packaging. System 2 is slow, effortful and logical: it multiplies 17 × 24, compares specs, reads a contract. And it is lazy: when System 1 convinces it that the answer is already there, it just nods it through. Most marketing speaks to System 1 and leaves it to System 2 to justify the decision.

The best-known evidence is anchoring. In 1974, Tversky and Kahneman had people spin a wheel of fortune that was secretly rigged to stop at 10 or 65, and then asked them for the share of African countries in the UN. Those who saw 10 estimated 25% on average; those who saw 65 estimated 45% (overview of the experiments). Dan Ariely repeated the same thing with the last two digits of an American Social Security number: people with higher numbers bid 60 to 120% more in an auction. But experiments where the anchor was obviously random (a die, a number generator) sometimes failed to reproduce the effect.

The second building block is loss aversion from prospect theory (1979): a loss hurts about twice as much as an equal gain pleases (Wikipedia). “You’re losing CZK 500 a month” moves people more than “you’ll save CZK 500 a month.” The third is WYSIATI (what you see is all there is): System 1 builds a story from what is in front of it and doesn’t miss what is absent. One photo of a happy customer beats a statistic nobody showed.

A final word in fairness: part of the book did not hold up. The studies on priming (subtle cues that are supposed to change behavior) in chapter four did not replicate, and in 2017 Kahneman admitted that he had “placed too much faith in underpowered studies.” Anchoring, framing and loss aversion survived replication.

Attention is the one input without which no model in this course works: AIDA starts with the letter A. Since 2004, Gloria Mark of UC Irvine has been measuring how long people stay on one screen before they switch; today’s average is 47 seconds, and it keeps getting shorter. A Reviews.org survey (Q4 2025, about 1,000 American adults) counted 186 phone checks a day, one for every five minutes awake, and 45.8% of people consider themselves addicted. In Czechia, according to the AMI Digital Index 2025 (Stem/Mark, 1,010 users over 15), we spend 143 minutes a day on social media, and Gen Z spends 220 minutes. Over a year, that is 870 hours, or 36 full days.

The other side of the equation is supply. In the Advertising lesson, we saw that the famous 5,000 ads a day is an illustration, not a measurement, and that according to Media Dynamics (2014), an American adult encounters about 360 ads a day and notices 150 to 155. Attention works not like a spotlight but like a System 1 filter with three questions: Is it new? Is it for me? Is it a threat? Whatever doesn’t get through, System 2 never sees.

For design, this leads to one rule: one thing, big, right away. A headline of six words or fewer, one visual, one button. On a website, the first screen has to say what this is and what I should do before anyone scrolls. In video, no logo in the first seconds, straight into the action. At an event, blocks of 15 to 20 minutes with a change of format between them, because a room’s attention fluctuates just as it does on a screen. With notifications, every unnecessary one lowers the chance that the recipient opens the next one.

Why do we pick up our phones 186 times a day? The feed pays out rewards irregularly, like a slot machine (B. F. Skinner’s variable ratio), so 186 checks a day are 186 pulls of the lever; we take apart the whole dopamine loop in the Social Media lesson. Marketers use it in streaks and badges; users defend themselves with notification badges turned off, a grayscale screen and the phone kept out of the bedroom. Whoever sells attention should also protect it.

Robert Cialdini spent three years undercover in sales teams and among fundraisers and recruiters, and wrote Influence (1984) with six principles; in 2016 (Pre-Suasion), he added a seventh, unity, and the 2021 edition has all seven.

Reciprocity: whoever receives something wants to give something back. A sample at the supermarket, a free warehouse analysis, a trial workout, free shipping on the first order. Commitment and consistency: Freedman and Fraser (1966) asked homeowners to put a small “drive safely” sign in their window. When they asked them two weeks later to put a big, ugly sign in their front yard, 76% agreed; without the small request first, only 17% did. Hence loyalty programs and the shopping cart that reminds you of an unfinished purchase.

Social proof: when uncertain, we do what others do. Solomon Asch (1951) seated a person among seven actors who agreed on an obviously wrong line length; 74% of participants went along at least once, and 35.7% of answers were wrong (without the pressure, under 0.7%). The chart shows what a single ally does. Cialdini tested it in hotels: a card saying “most guests reuse their towels” worked better than an appeal to protect the environment, and best of all was “most guests in this room” (Goldstein, Cialdini, Griskevicius 2008). A caveat: a German replication (Bohner and Schlüter 2014, 928 guests) did not confirm the advantage of the norm over the environmental appeal.

Scarcity: Worchel, Lee and Adewole (1975) had people taste cookies from a jar holding either ten or two; the ones from the nearly empty jar tasted better, and best of all were the ones that dwindled from ten to two before people’s eyes. Limited-edition jerseys, “only 3 left,” “sale ends at midnight.” Authority: in Milgram’s experiment (1961, 40 participants), 65% of people “gave” a stranger shocks of up to 450 V because a man in a gray lab coat calmly told them to. “Used by Olympians,” “recommended by 9 out of 10 dentists.” Liking: we say yes to people we like, to people similar to us, to good-looking people, to those who praise us. Unity: it is not that you like me, but that I am one of you. Family, club, “we Slavia Prague fans.” The strongest principle and the hardest to fake.

Each of the seven principles has a dark twin: a countdown that starts over once it runs out; “Petr from Plzeň just bought this” coming from a script; a subscription you turn on with a click and cancel by letter. These techniques are called dark patterns (a term coined by Harry Brignull in 2010). In 2022, the European Commission, together with authorities from 23 member states, Norway and Iceland, checked 399 online stores: 148 of them (37%) used at least one of the three patterns under review. Most often they hid information (the shipping cost, a cheaper option), then used design to push toward the more expensive choice (false hierarchy), and ran fake countdown timers.

The sweeps continue. Around Black Friday 2025, the Commission checked 314 retailers: at least 30% did not follow the rules for advertising discounts (a discount is calculated from the lowest price in the previous 30 days, not from a fictitious “original” price), 10% used drip pricing (the price grows with fees during checkout), and 18% used pressure techniques. In 2023, a sweep covered 576 influencers: 97% published commercial content, but only 20% consistently labeled it. In Czechia, the same rules are enforced by the ČOI (Czech Trade Inspection Authority), and since 2024 they have been joined by the Digital Services Act (DSA), which explicitly bans dark patterns in platform interfaces.

Where is the line? Two tests. Truthfulness: “only 3 left” is honest scarcity when three are left, and fraud when there are three hundred. Symmetry: saying no must be as easy as saying yes; when “Accept all” is a big green button and “Reject” is a gray link below the fold, that is false hierarchy, even in a cookie banner. A checklist before every campaign:

Cognitive biases are systematic errors of System 1, and four of them drive most pricing. We have already met the anchor: the first number becomes the yardstick, which is why a price list starts with the most expensive enterprise tier, and why a crossed-out price works even if nobody ever paid it (banned in the EU). Framing: Tversky and Kahneman (1981) described an epidemic expected to kill 600 people. A program that “will save 200 people” was chosen by 72% of people over the risky option; described as “400 people will die,” it flipped 78% to the risky option. People rate beef labeled “75% lean” better than the same beef labeled “25% fat” (Levin and Gaeth, 1988). And framing works on experts too: at a conference (Nottingham 2006), 93% of young economists registered on time when the organizers wrote about a $50 “penalty” for late registration, and only 67% when they called the same amount a discount. Senior economists did not respond to the frame (Gächter et al., 2009).

The decoy effect is the most elegant. Dan Ariely showed MIT students a price list from The Economist: web $59, print $125, print + web $125. Nobody chose the middle option, but it changed how they saw the others: 84% chose print + web, 16% web only. Without the decoy, the ratio flipped: 68% web, 32% print + web. A caveat about replications: Yang and Lynn (2014) tried the decoy in 91 variants, and it worked reliably in only 11; it holds mainly where numbers are being compared, which is exactly what happens in a price list. Loss aversion: a trial that expires hurts more than a discount you never got; “you’ll lose the points you’ve collected” works better than “earn points.”

Two biases outside the price list. Confirmation bias: people seek out what they already believe; a fan skips the negative review, and a buyer picks, out of three case studies, the one that confirms what they wanted to buy anyway. Don’t try to convince people they are wrong; give them evidence for what they already believe and attach your product to it. The availability heuristic: what we can easily recall, we consider common; one loud customer on social media looks like a thousand.

Emotion is a decision-making mechanism, not a malfunction of it: System 2 can compare options forever, but System 1 has to choose. Marketing works with this through emotional triggers; in the lecture we work with six: joy (Coca-Cola), fear and urgency (insurance, security), guilt (charity, “your kids deserve it”), pride (status, performance), belonging (club, team) and nostalgia (retro packaging, jersey comebacks).

Abraham Maslow gave marketing a framework of needs in his paper A Theory of Human Motivation (1943): physiological needs, safety, love and belonging, esteem, self-actualization; a higher need makes itself felt only once a lower one is satisfied. Two extra notes: there is no pyramid in the original paper (management textbooks drew the triangle only in the 1960s), and the order is not fixed (the starving artist exists). For marketing, the question is which need the product serves: Nike sells shoes as self-actualization, a logistics company sells logistics as safety (the goods will arrive) and esteem (the buyer chose well), a kids’ camp sells belonging.

A story carries emotion: the brain remembers plot and characters and forgets specs by lunchtime. Nike’s Dream Crazy (2018) is a story about overcoming obstacles, the Coca-Cola Christmas trucks are nostalgia, a B2B case study is the story of a customer with a problem. What spreads is high arousal (awe, humor, anger), not sadness (STEPPS). Using the IPA Effectiveness Awards database, Les Binet and Peter Field show that emotional campaigns are roughly twice as effective as rational ones and deliver twice the profit, and they recommend putting at least 60% of the budget into long-term brand building (the 60:40 ratio is covered in the Brand lesson).

You took apart Petty and Cacioppo’s Elaboration Likelihood Model (ELM) in the Advertising lesson from the creative point of view. From the psychological point of view, ELM and Kahneman’s systems both belong to the family of dual-process theories, but they are not the same thing. ELM describes how deeply a person processes one message, and the same element can be a shortcut one time and an argument another (an attractive face is an argument for a shampoo, but only a shortcut for insurance). Kahneman describes thinking in general. What links them is that the peripheral route relies mainly on fast System 1 shortcuts (a face, music, the number of likes; an attitude forms quickly and fades quickly), while the central route takes System 2 effort (the arguments get processed, and the attitude lasts). The central route switches on only when the recipient has the motivation and the time; otherwise the peripheral one decides, whether the advertiser wants it to or not.

Through this lens, AIDA works as a rule of thumb, not a research finding. The first two steps, Attention and Interest, rest mainly on the fast “new, for me, a threat?” filter and the first emotion. Desire and Action add comparisons of price and risk, which need System 2. Feeling does not switch off even there: the decision often comes first, and reason justifies it. That is why a good product page has an image and one sentence at the top, and specs and reviews at the bottom for whoever came looking for an alibi.

The difference between B2C and B2B is not that companies decide rationally. A consumer buys a yoga mat alone in two minutes. In a company, according to Gartner (2025), the buying group has 5 to 16 people from as many as four departments, and the cycle takes months; the central route is mandatory. But each of those people has a System 1: the fear of choosing wrong and taking the blame; the pride of choosing well; fatigue from the twentieth PDF. “Nobody ever got fired for buying IBM” is loss aversion in one sentence. B2B needs emotion just as much as B2C, only a different kind: safety and esteem instead of joy and status. And creative drives 49% of an ad’s sales effect, targeting 11% (NCSolutions, 2023; a 2017 study put it at 47% and 9%, covered in detail in the Advertising lesson): even in B2B, reaching the right person is not enough.

The oldest typology is the four temperaments from ancient medicine (Hippocrates, Galen): choleric, sanguine, phlegmatic and melancholic. As a customer typology: the choleric wants the result and the price right now, the melancholic wants a spreadsheet and a guarantee, the sanguine a story and people, the phlegmatic calm and time. It is not science, but it is a language for a sales conversation. The same goes for DISC (Marston, 1928: dominance, influence, steadiness, conscientiousness), a vocabulary for sales teams, not a measurement.

MBTI (Myers–Briggs Type Indicator) is the most widely used test in the world: four dichotomies (E/I, S/N, T/F, J/P), 16 types and, according to the publisher, up to 2 million tests a year. The problem is reliability: after about five weeks, 39 to 76% of people get a different type (Pittenger 1993, Schuwirth 2004, Grant 2013), because anyone close to the middle of a scale flips from one side to the other. McCrae and Costa (1989) showed that the MBTI scales correlate strongly with the Big Five (E–I with extraversion −0.74, S–N with openness 0.72): they measure the same thing, only worse and with a label. Great fun for team building, a poor basis for hiring or segmentation.

The Big Five (OCEAN: openness, conscientiousness, extraversion, agreeableness, neuroticism) is the standard in personality psychology: five continuous scales; the result holds up when the test is repeated, and it predicts behavior. The tests are free: bigfive-test.com (10 minutes, also in Czech) or the shorter IPIP. In marketing, it serves psychographic segmentation: you don’t split the audience into types; you ask what a point on a scale means for tone, channel and product. A persona then isn’t “Pavel the sanguine” but “Pavel, 34, high openness, low conscientiousness, shops at night on his phone.” In the lecture, we use four customer anchors for this (they have nothing to do with the anchor in a price list): what people pride themselves on and what the product should help them keep:

Psychology explains the individual, sociology what the individual does in a group, and buying almost always happens in a group, even when you stand at the checkout alone. There are three reference groups. Aspirational: where I want to belong (luxury, pro athletes). Membership: where I belong (a club, a company, a brand community). Dissociative: where I don’t want to belong, and its brand is dead to me (which is why teenagers leave a network once their parents show up). Every product has all three.

How many people are in those groups? In 1992, Robin Dunbar derived from the size of the primate neocortex that a human can manage about 150 stable relationships (the regression gave 148, with a 95% interval of 100 to 230), with inner layers of about 5 closest and 15 close people, who get two thirds of our social time. A 2021 reanalysis (Lindenfors, Wartel and Lind) gives 69 to 109 or 16 to 42, depending on the method: take 150 as an order of magnitude, not a constant. For marketing: the influence that counts is not reach but those 5 and 15.

The tribes from the Social Media lesson are membership groups with their own language, ritual and hero. A brand in them is either a guest (a sponsor that respects the rules) or an intruder (a sponsor that tells the tribe what to feel); a member can tell the difference in a second, with System 1. The safest approach is to let the tribe speak for itself and just hold the camera.

The group also decides what a trend is. The bandwagon effect is social proof in motion: the more people join, the more others follow. FOMO (fear of missing out) adds loss aversion to the mix. You know Rogers’ curve of adoption (from innovators to laggards) from the Trends lesson; psychologically, what drives it is precisely social proof, because each group waits for the one before it. For a brand: jump on a trend when it fits your tribe and you are among the first; when everyone is already doing it, you are a laggard.

System 1 forms an opinion about a person before System 2 can do anything. Janine Willis and Alexander Todorov (Princeton, 2006) showed faces for 100 milliseconds, and judgments of trustworthiness, attractiveness, competence or aggressiveness matched judgments made with no time limit; more time mainly added confidence, while agreement with the no-limit judgments did not improve. A tenth of a second, at a job interview and in the opening shot of a video.

The second mechanism is the halo effect: one prominent trait colors all the others. In 1920, Edward Thorndike noticed that officers gave physically fit soldiers higher marks for intelligence and character too. Landy and Sigall (1974) gave men the same essay with a photo of its author: the attractive author got an average of 6.7 points out of 9, the unattractive one 5.9. For marketing: packaging, typography, photos and website speed are the product’s halo. To System 1, a slow website with a blurry logo means “unreliable company,” and one bad detail (a typo in the name, an expired certificate) drags everything down.

And what about sex sells? Robert Lull and Brad Bushman (Ohio State, 2015) summarized 53 experiments with 8,500 people over 44 years in Psychological Bulletin: sexual content in ads did not improve memory for the brand, worsened attitudes toward the brand, and did not change purchase intent. Sex draws attention to itself, not to the brand. In sports, something else works: the body as proof of performance and as aspiration, not eroticism.

Reputation is social proof with a timeline: not what people say about you today, but what they have said over the last five years and whether it adds up. Online, it is carried by reviews. BrightLocal (2026, 1,002 adults in the US): 97% of people read reviews of local businesses, 68% want at least four stars, and the bar is rising (see the chart). In Czechia, the same role is played by Google reviews, Heureka (a Czech price-comparison site) with its “Verified by customers” badge, and Firmy.cz (a Czech business directory): the star rating is the anchor, the number of reviews is the social proof, and trust comes from the owner’s reply to a negative review.

Trust in institutions is low, and that changes whom people trust. In its Digital News Report 2026, the Reuters Institute measured trust in news in Czechia at 31% (33% the year before; the trend is in the Media lesson), and only 13% of people pay for online news. When we don’t trust institutions, we trust people: creators, reviewers, “someone like me.” So a brand’s reputation isn’t made by a press release, but by an employee on LinkedIn, a customer with a photo, and how the company responds when something goes wrong.

Psychologically, reputation is commitment and consistency stretched over time: a company that has done what it promised for five years has an anchor that one mistake won’t move; a company with no track record is judged by its latest review. So: collect reviews when satisfaction is high, reply to all of them, don’t delete the negative ones and don’t buy positive ones. A fake review is social proof turned inside out: once it is exposed, it casts doubt on the twenty genuine ones too.

A machine can already do everything in this lesson. In a randomized experiment, Francesco Salvi, Manoel Horta Ribeiro, Riccardo Gallotti and Robert West (EPFL, Princeton and FBK Trento) had 900 people debate either a human or GPT-4. When the model knew basic facts about its opponent (gender, age, ethnicity, education, employment, political orientation), the odds that a person would move toward the opponent’s view after the debate rose by 81.2% compared with a debate against a human. Without personalization, GPT-4 was on par with a human. The study appeared in Nature Human Behaviour (2025); the 2024 preprint reported 820 people and 81.7%. The point is not that the model is more eloquent: it can tailor Cialdini’s principles, frames and anchors to each person, at the scale of millions and for the price of tokens.

But AI can also protect attention. My dashboard from part three uses it the opposite way from social media: AI there sorts things so that as few alerts as possible come through, not as many as possible. The last sentence of the lesson: understand yourself first, then the customer. More about models in the Artificial Intelligence lesson.