The Science of Pricing: What Your Brain Actually Does When It Sees a Price Tag

In the early 1990s, Williams-Sonoma introduced one of the first home bread makers to the American market. It cost $275. Customers browsed. They picked it up, turned it over, read the box. They put it back. Sales were dismal. The product worked fine. The price wasn't outrageous. But people had never seen a home bread maker before, and when you have no reference point for what something should cost, the brain does something that looks a lot like nothing. It stalls. The number on the tag doesn't connect to any prior expectation, so no computation resolves, and the customer walks away not because the price is wrong but because their brain can't tell if it's right.

Williams-Sonoma's fix wasn't a sale. It wasn't a coupon. They introduced a second bread maker — larger, fancier, priced at $429.

Sales of the original $275 machine nearly doubled. The $429 model barely sold. It was never supposed to. Its job was to make $275 feel like a deal.

The bread maker didn't change. The customer didn't change. The only thing that changed was the number next to it on the shelf, and that number rewired the entire computation. If that bothers you — if it seems like it shouldn't work, that a rational person should evaluate $275 on its own merits — then you're thinking about pricing the way most entrepreneurs think about pricing. And you're wrong in exactly the way that costs the most money.

The Computation You Don't Know You're Running

Your brain doesn't evaluate a price. It evaluates a price relative to something.

This sounds obvious until you realize most pricing strategies are built on the opposite assumption — that customers weigh absolute value, that a fair price should sell on its own merits, that transparency and simplicity are always better. They aren't. Not because customers are irrational, but because the neural hardware that processes price tags doesn't work the way pricing teams think it does.

Neuroscientists call the underlying mechanism reward prediction error. The brain maintains predictions about what things should cost, and dopamine neurons respond not to the price itself but to the gap between the predicted price and the actual one. A price that matches expectations generates no signal at all. A price that's better than expected generates a burst. A price that's worse than expected generates a dip. The feeling of a "good deal" isn't an opinion. It's a neurochemical event — the same signal that fires when any prediction is violated in your favor. And that dip is exactly why customers say they love a product and then never buy it: the pain-of-paying circuit didn't fire during the hypothetical evaluation, so the prediction error at the moment of real purchase comes as a shock.

This has four implications for how you price anything, and each one is counterintuitive enough to be worth its own principle.

The Anchor Is the Number

In 1987, Gregory Northcraft and Marjorie Neale ran an experiment that should make anyone who has ever priced a product uncomfortable. They took real estate agents — professionals who appraise homes for a living — and amateurs to tour the same house in Tucson, Arizona. Each person received an identical information packet with one exception: the listing price. Some packets said $65,900. Others said $83,900. Same house. Same neighborhood. Same square footage.

The amateurs who saw the $65,900 listing appraised the home at an average of $63,571. Those who saw $83,900 appraised it at $72,196. An $8,625 swing caused entirely by a number on a piece of paper they'd read five minutes earlier.

The real estate agents were anchored too. But here's what made the study famous: only nineteen percent of the agents mentioned the listing price as a factor in their appraisal, compared to thirty-seven percent of the amateurs. The professionals were more influenced by the anchor and less aware of it.

Anchoring isn't a bias you can train away. It's how the prediction engine sets its starting point. The first number installs a prior. The brain evaluates everything that follows as a deviation from that prior. Which means on your pricing page, the first number the customer encounters isn't just information. It's the reference point their entire evaluation will be computed against. If that number is your cheapest tier, every other tier looks expensive. If that number is your most premium offering, everything below it looks reasonable.

Williams-Sonoma didn't convince anyone to buy the $275 bread maker. They installed a $429 prior that made $275 computable.

The Deal Is the Product

In 1981, Amos Tversky and Daniel Kahneman published an experiment in Science that captured this with surgical precision. Participants imagined buying two items: a $125 jacket and a $15 calculator. A salesperson mentions the same calculator is available for $10 at a store twenty minutes away. That's a five-dollar savings. Sixty-eight percent of people said they'd make the drive.

In the second condition, the prices were swapped. A $15 jacket and a $125 calculator, and the calculator is $120 at the other store. Still five dollars saved. Still twenty minutes of driving. Twenty-nine percent would go.

The savings are identical. The effort is identical. But five dollars off a fifteen-dollar item feels like a thirty-three percent deal, and five dollars off a hundred-and-twenty-five-dollar item feels like four percent of nothing. The brain doesn't compute absolute savings. It computes the ratio between what it expected to pay and what it actually pays. The gap is the signal. The signal is the motivation. Remove the gap and the purchase might still happen, but the neurochemical push behind it disappears.

This is why Costco has kept its rotisserie chicken at $4.99 since 2009, even though each bird costs roughly six to seven dollars to produce. They lose an estimated thirty to forty million dollars a year on chicken alone. They sold 137 million of them in 2023. The chickens sit at the back of the store, pulling customers past thousands of other products, and the $4.99 price tag creates a prediction error so consistent it becomes the store's identity. Costco isn't in the chicken business. They're in the prediction-error business. The chicken is just the delivery vehicle. And the $1.50 hot dog at the food court exit ensures the last moment the customer remembers is another deal.

Your product's value isn't what you charge. It's the distance between what the customer expected to pay and what you actually asked for — a form of perceived value that context can multiply or destroy. If that distance is zero — if you've priced "fairly," with no anchoring, no contrast, no gap — you've engineered a product that generates no reward signal at the moment of purchase. The price might be right. The feeling will be wrong.

The Decoy Seat

Dan Ariely tested this with The Economist's subscription options and the results still stop people mid-conversation. The magazine offered three tiers: web-only for $59, print-only for $125, and print-plus-web for $125. When Ariely gave this to MIT students, sixteen percent chose web-only, zero percent chose print-only, and eighty-four percent chose print-plus-web.

Zero people chose the print-only option. It was objectively dominated — same price as print-plus-web but less product. Useless, right?

Ariely removed it. Offered only web-only ($59) and print-plus-web ($125). The results inverted. Sixty-eight percent chose web-only. Thirty-two percent chose print-plus-web.

An option that zero people selected was responsible for shifting fifty-two percentage points of demand toward the most expensive tier. It worked not because anyone wanted it, but because its existence changed what the other options looked like. Print-plus-web at $125 looks unremarkable next to web-only at $59. Print-plus-web at $125 looks like a steal next to print-only at $125. The comparison reshapes the computation, and the computation reshapes the choice.

The decoy doesn't need to be a good option. It needs to be a strategically bad one — bad in a way that makes your preferred option look better by contrast. If every tier on your pricing page is genuinely competitive, you don't have a pricing strategy. You have three products competing with each other, and the brain's computation has no clear resolution.

The Denominator Effect

In the mid-1990s, Procter & Gamble cut their Head & Shoulders line from twenty-six varieties to fifteen. Not a rebrand. Not a reformulation. They simply removed the least popular options. Sales rose ten percent.

The mechanism is the same one that makes you scroll Netflix for forty-five minutes and choose nothing. The brain evaluates each option's value divided by the total value of everything else available. Two options, the math resolves cleanly — one is better, signal is strong, you pick it. Twenty-six shampoos, most of them roughly similar, and every individual signal gets crushed toward zero. Not because the shampoos are bad. Because the hardware has a ceiling, and twenty-six similar options blow through it.

Sheena Iyengar demonstrated this at a grocery store in Menlo Park. A display of twenty-four jams attracted more browsers than a display of six — people are drawn to abundance. But only three percent bought from the twenty-four-jam display. Thirty percent bought from the six-jam display. Ten times the conversion rate with a quarter of the options. The large display was better at generating interest. The small display was better at generating a decision. Interest and decision run on different computations, and the first one is not a reliable predictor of the second.

If your pricing page has more than three core tiers, you're not giving customers more choice. You're giving their neural hardware a math problem it can't cleanly solve.