Here’s a fact that should reshape how you think about every product, restaurant, vacation, and customer experience in your life:
The single most important paper in modern behavioral economics is about colonoscopies.
I’m not joking.
In 1996, Donald Redelmeier and Daniel Kahneman published a study in the journal Pain. They had patients undergoing real colonoscopies — which in the early 90s were uncomfortable in ways modern procedures aren’t — rate their pain on a handheld dial in real time, then rate the overall experience afterwards.
The results were so strange they reshaped a generation of behavioral science.
Patients who went through a shorter procedure with a sharply painful final stretch remembered the procedure as more painful than patients who went through a longer procedure with a milder ending. The longer procedure clearly produced more total pain over its duration — their dials said so in real time — but the longer procedure was remembered as less bad.
Total pain didn’t predict memory. Peak pain and final pain did. Duration barely mattered at all.
Kahneman later called this the Peak-End Rule. The surrounding finding — that we don’t really remember how long an experience lasted — he called duration neglect. He spent much of his subsequent career working out what these findings imply about us.
What they imply, in short, is that you are two people.
There’s an experiencing self — the one actually living through the moments. And there’s a remembering self — the one who decides whether it was worth doing again. And the remembering self mostly ignores the experiencing self’s reports. It samples the peak and the end and calls it a day.
This sounds abstract until you realize that almost every decision a business makes about customer experience is implicitly a decision about which of these two selves it’s designing for.
You Have Already Lived This
If you’ve ever taken a two-week vacation and noticed, six months later, that all you really remember is one good dinner, one fight with your partner, and the chaos at baggage claim — congratulations. You’ve met your remembering self.
If you’ve ever watched a great movie that fumbled the ending and noticed how much that ending dragged your whole memory of the film down — same thing.
If you’ve ever come out of an eight-hour flight and spent more mental energy complaining about the twenty minutes at the carousel than thinking about the flight itself — same thing again.
Kahneman makes the point at length in Thinking, Fast and Slow, and the implication he draws from it is uncomfortable. He says: when you choose your next vacation, you are not really choosing what to experience. You are choosing a story to remember. The experiencing self gets almost no vote.
That’s a heavy thought for a Tuesday. But it’s also one of the most useful insights a product person or marketer can carry around.
The Heath Brothers’ Translation
The single best practitioner-level book on Peak-End is The Power of Moments by Chip and Dan Heath. If you’ve read Made to Stick or Switch, you know the Heaths are very good at translating academic behavioral science into things you can actually do on Monday morning.
Their core argument is that we remember experiences as a sequence of defining moments, not as a continuous flow. The defining moments cluster around four categories: peaks, pits, transitions, and milestones. Everything else is mostly forgotten.
Their key insight is that most organizations underinvest in defining moments. They spend enormous effort on averages — keeping every minute of the experience “pretty good” — and almost no effort on engineering the specific moments the remembering self will actually keep.
This is, if you squint, the same argument Don Norman makes in The Design of Everyday Things from the design side: most design effort is wasted on parts of an experience nobody remembers using.
Three Real-World Operators Who Get This
I’m one of those people who reads hospitality memoirs the way other people read crime novels — there’s something about the obsessive level of detail required to run a great restaurant or hotel that scratches the same itch as good detective fiction. So let me walk through three places where I’ve seen Peak-End engineering operationalized to a level that borders on the surgical.
1. Disney
If you’ve ever wondered why Disney parks feel different from every other theme park you’ve been to, here’s an unsexy answer: they obsessively engineer the peaks and the endings.
Lee Cockerell, who ran operations for Walt Disney World for ten years, wrote a book called Creating Magic in which he describes the deliberate engineering of the “kiss goodnight” — the fireworks finale and choreographed exit from the Magic Kingdom. They know your remembering self is going to encode the last twenty minutes of your day with disproportionate weight, so they engineer those twenty minutes within an inch of their life.
The famous Disney parades aren’t just entertainment, either. They’re scheduled peaks designed to break up long stretches of waiting in line and produce the kind of moment the remembering self will file away.
If you’ve ever come home from Disney remembering “the parade” instead of “four hours of standing in 95-degree heat,” that’s not a coincidence. That’s the design working as intended.
2. Apple
Walk into any Apple Store and notice something most retail experiences don’t bother with: an actual goodbye.
The employee who helps you doesn’t just hand over your bag and look away. They thank you, often make eye contact, sometimes walk you a step or two toward the door. The receipt arrives in your inbox before you’ve made it to the parking lot. The transaction ends on a polite note, not a transactional one.
This is engineered. Apple’s retail training is built around the APPLE acronym — Approach, Probe, Present, Listen, End — and the E gets disproportionate attention because Apple’s retail team knows the end is what gets remembered.
You can feel the absence of this when you walk out of a store that doesn’t do it. The eye-contactless “have a good one” exit is the retail default. Apple chose not to do that. And the chosen alternative has been one of the quietly compounding reasons Apple Stores generate the highest revenue per square foot in retail.
3. Will Guidara at Eleven Madison Park
If you want to see Peak-End engineering executed at its most obsessive level, read Will Guidara’s Unreasonable Hospitality. Guidara was the GM of Eleven Madison Park in New York during its rise from “promising” to repeatedly being voted the best restaurant in the world.
His core insight is that the part of the meal most guests would remember had almost nothing to do with the food. It was the unexpected gesture at the end — a tableside personal touch, a small surprise tailored to the guest, something the kitchen had improvised that morning because a server overheard the table mention something offhand.
There’s a story in the book where Guidara overhears a table of European tourists say they regretted leaving New York without ever eating a real street hot dog. He runs out to the corner, buys a $2 dirty-water dog from a vendor, has the kitchen plate it as an “unannounced course,” and serves it to them as a final amuse-bouche before dessert.
That meal cost over $1,000 a head. The thing those tourists told everyone they knew about, for years, was the hot dog.
Guidara had figured out something most luxury operators miss: the food was the duration. The hot dog was the peak. And the peak was the part that traveled.
This is essentially what Danny Meyer is arguing in Setting the Table with his line about “the last impression being the lasting impression.” It’s the Peak-End Rule restated by an operator who internalized it through ten thousand dinner services.
The Part Almost Nobody Gets Right
This is where I want to push past most Peak-End writing you’ll find online, because I think this is the bit that actually matters.
Most brands have an outdated map of where their customer experience actually ends.
For a hotel, “the end” is not the moment you check out. It’s the email three weeks later asking whether you’d return. It’s the lost-and-found request. It’s the bill dispute that gets resolved or doesn’t.
For an e-commerce store, “the end” is not the purchase confirmation. It’s the box arriving battered. It’s the return process. It’s the customer-service interaction when something went wrong.
For a SaaS company, “the end” is not the cancellation click. It’s the cancellation flow itself — and the email a month later trying to win you back.
This is the most under-designed surface in modern business. The end of an experience usually happens at the moment of the customer’s greatest frustration — a return, a cancellation, a billing dispute, a complaint — and almost nobody engineers it deliberately.
If you’ve read Nir Eyal’s Hooked, you’ve seen this argued from the inverse direction. Eyal’s frameworks focus on the peaks — the variable rewards that pull users back. But the same logic applies to endings in reverse. A horrible cancellation flow doesn’t just lose a customer. It writes a memory that no future marketing campaign can outrun.
I’d argue that if you took every dollar a typical mid-sized SaaS company spends on top-of-funnel paid acquisition and instead spent it on rebuilding the cancellation experience and the customer support flow, the LTV math wouldn’t even be close. The remembering self of every churned customer is doing future-marketing damage to you that you can’t measure on a dashboard.
The Digital Twist (Where Peak-End Becomes a Conversion Lever)
Most of what I work on professionally is digital — A/B tests, conversion funnels, behavioral nudges in software products. Peak-End shows up everywhere in CRO and product design once you start looking.
A few examples worth flagging:
Onboarding. Duolingo’s celebration animations after a completed lesson — the streak count animation, the owl shaking pompoms, the satisfying “ding” — exist for one specific reason. They are engineered peak moments designed to compete with the rest of your day. The remembering self files Duolingo as “the app that made me feel competent,” not “the app where I drilled forty French verbs and got most of them wrong.”
Email campaigns. A subject line is the peak. A CTA at the very end is the end. The middle of the email is mostly duration the remembering self will throw away. This is partly why Ron Kohavi, Diane Tang, and Ya Xu in Trustworthy Online Controlled Experiments show so consistently that subject lines and end-of-email CTAs dominate experimental results. They’re the surfaces that actually get remembered. The middle paragraphs of your beautifully written newsletter contribute mostly to duration, not to memory.
Cancellation flows. This is where digital companies fail Peak-End most expensively. The user is already emotional. The decision is final. A respectful, friction-free, even gracious cancellation flow is one of the highest-leverage product investments most SaaS companies could make. Almost none do. The default is a dark-pattern obstacle course that maximizes short-term retention by ten percentage points and salts the field for every future remarketing campaign.
Receipts and confirmation screens. This is the unglamorous “end” of every e-commerce purchase, and it’s mostly designed by accountants. There’s a vast amount of compoundable customer-loyalty value sitting unclaimed in the standard transactional email. It costs almost nothing to do this well, and almost everybody does it badly.
If you run experiments for a living, the operational takeaway is straightforward and uncomfortable: you should be testing endings more than you currently are. End-of-experience surfaces get less experiment volume than top-of-funnel surfaces, even though they shape memory far more.
A Quick Note on Why Pits Matter Too
One thing the Heath brothers got right that the original Kahneman framing partly obscures: Peak-End doesn’t just mean “engineer good peaks.” It means “engineer the peak emotional moment, in either direction.” Negative peaks — what they call pits — get encoded just as durably as positive ones.
A single bad pit can ruin an otherwise great experience. The screaming-baby seatmate. The rude waiter. The crashed checkout page at minute four of a fifteen-minute purchase flow. These are pit moments, and your customer’s remembering self will use them to overwrite hours of competence.
This is also why service recovery is so disproportionately valuable. The Ritz-Carlton’s famous $2,000 standard — the company empowered every employee to spend up to $2,000 per guest to resolve a problem without seeking manager approval — exists because their internal data showed that a successfully recovered complaint produced higher loyalty scores than a flawless experience. The pit, fixed properly, becomes the peak. The remembering self files it as “the time the hotel saved me,” not “the time the hotel screwed up.”
This is what Robert Cialdini was hinting at in Influence with his work on Reciprocity. Recovery from a service failure is a tiny gift, and tiny gifts at moments of high emotional charge produce loyalty that no amount of normal-service competence can buy.
What I Take From All This
The most useful thing the Peak-End Rule taught me wasn’t tactical. It was philosophical.
You are not designing for the experiencing self. You will never have a real conversation with the experiencing self. The experiencing self leaves no reviews, fills out no surveys, doesn’t tweet, doesn’t refer friends. The experiencing self lives and dies in the moment and reports nothing back.
Every signal you ever get back from a customer comes from the remembering self.
So if you want to grow a business — or design a product, or run a restaurant, or build a hotel, or write a movie, or plan a wedding — the question isn’t “what should our customers experience?” The question is “what should they remember?” And the answer is almost always the same two things: the peak and the end.
Get those right and the rest of the experience can be merely competent.
Get those wrong and it doesn’t matter how good the middle was.
This is what Kahneman was really trying to tell us all those years ago, hidden inside a paper about colonoscopies that almost nobody outside academic journals read at the time.
You have two selves. One of them lives the life. The other one writes it down.
Design for the one with the pen.