Peak-End Rule: An S-Tier Behavioral Designer’s Guide to Kahneman

Most companies obsess over averages and medians. They measure satisfaction across the whole experience, average the scores, and ship the result to the board. Daniel Kahneman spent three decades proving this is not how human memory works, and the gap between what actually happened to a customer and what that customer remembers is the single largest unexploited margin in experience design.

The Peak-End Rule is one of the experience-design frameworks catalogued in our Behavioral Framework Library, Yu-kai Chou’s curated index of the behavioral science behind world-class motivation and engagement design.

The Peak-End Rule says the memory of any emotional episode is dominated by two moments: the emotional peak (whatever was most intense, good or bad) and the ending. The middle mostly vanishes. Duration — how long the thing lasted — is almost ignored. A two-hour film with one shattering moment and a strong final scene will be remembered more fondly than a four-hour film of consistently high quality with a flat finish. The same asymmetry shapes how patients rate colonoscopies, how customers rate airlines, how players rate games, and how voters rate presidencies.

This post is the S-Tier designer’s guide to the rule that quietly governs every loyalty program, onboarding flow, and customer journey you have ever built. By the end you will know what Kahneman and his collaborators actually found, the three ways the original evidence is fairly criticised, why the rule still holds up in 2026 despite those critiques, and exactly how I use it alongside the Octalysis Framework when advising MrBeast, LEGO, Microsoft, Porsche, Coca-Cola, Tesla, and governments that need to motivate millions of citizens at scale. If you only remember one idea from everything I teach about experience design, let it be this: you are not designing an experience, you are designing a memory — and memory is peaks and endings.

⚡ Speed Run Notes

• Memory is not a recording — it is an emotional summary dominated by two moments. Redelmeier and Kahneman’s 1996 colonoscopy study and Kahneman, Fredrickson, Schreiber and Redelmeier’s 1993 cold-pressor study both showed the same pattern: patient ratings of an aversive experience correlate almost perfectly with the average of the peak moment and the final moment, and correlate near-zero with total duration. The middle of the experience gets averaged out of memory.
• Duration neglect is the stranger-than-fiction half of the finding. In the cold-water study, participants preferred a longer version of a painful experience — 60 seconds of very cold water followed by 30 seconds of slightly less cold water — over a shorter version of the same painful start that simply ended. Adding more total pain made the memory better. This is not a quirk; it is how our remembering self is built.
• The remembering self and the experiencing self are two different clients with two different budgets. Kahneman’s later framing in Thinking, Fast and Slow draws this distinction sharply. The experiencing self lives the moments one by one; the remembering self edits them into a two-point story. Every design decision you make is really two decisions: one for the experiencing self, one for the remembering self. Most designers only budget for the first.
• Peaks can be engineered, endings can be choreographed. Once you accept that peak and end dominate memory, the design move becomes obvious. Find the moment of highest emotional intensity you can honestly create, and invest there — even at the cost of flattening other moments. Then own the last impression. The “last mile” is not just a logistics problem; it is the moment your customer’s memory is being written.
• The rule is asymmetric for negative versus positive experiences. For painful experiences, adding a milder ending genuinely softens the memory (Redelmeier, Katz & Kahneman 2003 showed this in a clinical replication). For pleasant experiences, the peak effect is strong but duration neglect is softer — very long pleasant episodes do accumulate a bit more remembered value. Designers working on painful processes (returns, cancellations, medical procedures) get the biggest lift; designers of pleasant experiences still win, just less dramatically.
• The rule predicts why loyalty programs feel flat. Most loyalty programs are designed as evenly-distributed “earn and burn” economies — a steady trickle of points, a steady stream of rewards. Evenly distributed is exactly the emotional profile the remembering self under-weights. Programs that win — Disney parks, Nike Run Club, Sephora Beauty Insider, Starbucks Rewards — all engineer specific peaks (surprise upgrades, member-only moments, birthday theatre) and specific endings (the last interaction on a visit, the cashier’s farewell, the app closing screen).
• The original studies were small, clinical, and aversive — generalisation is not guaranteed. The colonoscopy sample was 154 patients; the cold-pressor sample was college students. The fair academic criticism is that the clean peak-plus-end average is probably a first-order approximation that interacts with context, meaning, and narrative structure in ways the original studies did not probe. Use the rule as a design heuristic, not a physics law.
• Narrative structure sits on top of the peak-end arithmetic. Chip and Dan Heath’s The Power of Moments (2017) refined the design-facing version: defining moments cluster around four families — Elevation, Insight, Pride, and Connection — and the most memorable experiences are the ones that engineer peak moments within those families. The Heath brothers are doing applied Kahneman — peaks and endings plus a taxonomy of what counts as a “peak” for human beings.
• The Octalysis mapping is direct: peak = Core Drive spike, end = Endgame design. The Peak-End Rule is a reminder that the eight Octalysis Core Drives are not evenly weighted across time. A Core Drive 1 (Epic Meaning) spike in the middle of an experience may be forgotten; a Core Drive 5 (Social Influence) peak at the very end of an experience will be remembered forever. Treat your Core Drive deployment as a time-series, not a checklist.
• If you only remember one sentence: design the peak, own the ending, and stop wasting budget on the middle. Every loyalty program, onboarding flow, customer-service process, conference, course, and product launch lives or dies on these two moments. Spending engineering capacity on the forgettable middle is the single most common way engagement budgets get wasted in 2026, and Kahneman gave us the receipts for why.

Table of Contents

About Yu-kai Chou

Yu-kai Chou is the creator of the Octalysis Framework, the gamification and behavioral-design system used by companies, governments, and organizations reaching over 1.5 billion users.

He has advised leaders including Google, LEGO, Microsoft, Porsche, Tesla, and the government of Ukraine, and his work sits at the intersection of behavioral economics, product design, and motivation science.

Verify: Wikipedia · Google Scholar · Wikidata · LinkedIn

What Is the Peak-End Rule

The Peak-End Rule is a cognitive heuristic first formalised by Daniel Kahneman, Barbara Fredrickson, Charles Schreiber, and Donald Redelmeier in a 1993 paper titled When More Pain Is Preferred to Less: Adding a Better End, and later expanded in Kahneman and Redelmeier’s 1996 work on colonoscopy patients. The central claim is that people’s retrospective evaluation of a hedonic episode — any extended experience of pleasure or pain — is almost entirely predicted by two data points: the peak emotional intensity reached during the episode, and the emotional intensity in the final moments. The rest of the experience is averaged out of memory.

Before Kahneman, the dominant assumption in economics and experience design was that a rational agent remembers an experience as an integral — the sum of moment-by-moment utilities across the full duration. Under this model, a longer pleasant experience should be remembered as more valuable than a shorter pleasant experience of the same intensity, because there is simply more of it. Kahneman’s research showed this model is wrong for retrospective memory in a way that is not marginal — it is massive, predictable, and exploitable.

In one of Kahneman’s most-cited experiments, participants held their hand in 14°C (57°F) water for 60 seconds in one trial, and in another trial held their hand in 14°C water for 60 seconds followed by an additional 30 seconds as the water was warmed to 15°C — still painful, but milder. Asked later which trial they would prefer to repeat, a majority chose the longer trial. They preferred a strictly greater total amount of pain because the remembering self read the gentler ending as “this wasn’t so bad” while the flat ending left only the sharp finish in memory. The experiencing self suffered more; the remembering self judged the longer version as less aversive; the remembering self’s judgement predicted future choice.

This finding was replicated and extended in the colonoscopy study: patients rated 682 colonoscopies in real time on a handheld dial and again in a global post-procedure rating. The global rating was predicted almost entirely by the peak pain moment and the final minute’s pain. The total duration of the procedure — which varied from 4 to 69 minutes — predicted the global rating barely at all. A 60-minute procedure with a moderate last minute was remembered as less unpleasant than a 15-minute procedure that ended at peak pain. Redelmeier, Katz and Kahneman (2003) later ran a clinical trial that deliberately extended the end of colonoscopies with a less-painful tail segment. Patients who received the extended-end procedure were significantly more likely to return for follow-up screening. The design intervention saved lives.

The Peak-End Rule is the hinge on which Kahneman’s later distinction between the experiencing self and the remembering self turns. The experiencing self lives moment by moment, summing its utilities honestly. The remembering self constructs a compressed summary that is then used for every future decision — to come back, to recommend, to repeat. Design that optimises only for the experiencing self leaves enormous retention value on the table, because the remembering self is the one who will buy the second subscription.

The Core Findings

Four specific findings fall out of the Kahneman research program directly, and together they form the operational core of the Peak-End Rule.

First, the peak dominates retrospective memory far more than any other moment. Across the cold-water, colonoscopy, and numerous replication studies, the single highest-intensity moment of an experience — whether that intensity is pleasure or pain — predicts the global retrospective rating better than any other single data point. When researchers average the peak and the end, the correlation with global memory climbs into the 0.7–0.9 range in the original Kahneman studies. When they include the middle, average intensity, or duration, the correlation barely improves. Peaks are load-bearing; middles are not.

Second, the ending carries disproportionate weight even when it is not particularly intense. The last few seconds or minutes of an experience function as a summary frame for the remembering self — the moment where memory writes “and then the story ended like this.” A mediocre ending flattens an otherwise strong experience; a strong ending can rescue a mediocre one. This is why restaurant tips correlate more with the warmth of the farewell than with the entrée, and why hotel reviews mention checkout more often than check-in. The ending is the last signature on a contract with memory.

Third, duration is neglected in a way that borders on heroic. In the Kahneman studies, doubling the length of an unpleasant experience while holding peak and end constant did not double — or even meaningfully increase — the remembered aversiveness. This is the finding that most reliably breaks executive intuition: most managers assume “longer is worse” when customers are suffering, and “longer is better” when customers are enjoying themselves. The Peak-End Rule says neither is strongly true. What matters is the shape of the intensity curve, not the area under it. A short-but-sharp peak beats a long-but-flat satisfaction.

Fourth, the effect generalises beyond clinical contexts to the experiences designers actually care about. Follow-on studies have shown the peak-end pattern in holiday memories (Kemp, Burt and Furneaux 2008), vacation evaluations (Mitchell, Thompson, Peterson and Cronk 1997), ad exposure (Baumgartner, Sujan and Padgett 1997), video game sessions (Tan, Tang and Cheung 2015), and customer service encounters (Verhoef, Antonides and de Hoog 2004). The pattern is not perfect — contextual and individual differences modify its strength — but it shows up reliably across hedonic domains whenever the experience has an identifiable peak and an identifiable end.

Taken together, these findings give designers a striking lever. If you know where your peak will be and how your experience will end, you already know most of what your customer will remember. If you do not know — if you are letting peak and end emerge by accident — you are outsourcing your memory design to randomness, and randomness does not care whether the user comes back.

What Kahneman Got Right

The Peak-End Rule has been on behavioural-economics syllabi for three decades, and unlike many psychological findings of the same era it has largely survived the replication crisis. Three of Kahneman’s contributions are genuine insights that no serious experience-design framework has been able to ignore.

First, the separation of experiencing self and remembering self. This is the deepest contribution and the one that earns Kahneman’s place in the canon. Until the Peak-End Rule was articulated, the implicit model in economics and in most experience design was that we remember experiences roughly the way we lived them. Kahneman made it impossible to keep believing that. His 2010 TED talk, The Riddle of Experience vs Memory, reaches a general audience with a single example that generations of designers will never forget: the listener who attends a symphony where the final chord is shattered by a loud mistake, and who says “the whole experience was ruined,” when in fact the forty beautiful minutes were lived exactly as planned. Only the memory was ruined. The distinction between living an experience and remembering it is the hinge on which every sensible discussion of hedonic design now turns.

Second, the demolition of duration as a first-order variable. Before Kahneman, “how long did it last” was treated as the dominant variable in evaluating any hedonic episode — economic utility was quite literally modelled as an integral over time. After Kahneman, duration is understood to be a weak second-order effect in retrospective evaluation. This is not a small methodological note. It is a fundamental correction to how economists, healthcare policy makers, and experience designers allocate budget. If duration does not matter as much as we thought, then “shorter is kinder” and “longer is more valuable” are both wrong without specifying the shape of the intensity curve.

Third, the design-facing precision of the rule. Most cognitive biases, once named, are frustrating to design around — they describe a failure mode without giving you a lever. The Peak-End Rule is the rare case where the bias comes pre-packaged with actionable advice: identify the intended peak, own the ending, and worry less about the middle. The Heath brothers’ The Power of Moments is effectively a design manual bolted onto this rule. Disney’s “magical moments” programme, the Ritz-Carlton’s empowerment-to-delight policy, the Apple Store’s goodbye-handshake protocol, and Airbnb’s long-running focus on “host peaks” are all examples of organisations that took Kahneman’s finding and operationalised it as playbook-level design moves. Not many pieces of academic research survive that translation. The Peak-End Rule does.

Where the Peak-End Rule Falls Apart

A rule that has survived three decades of scrutiny is not a rule with no fair criticisms — it is a rule whose fair criticisms have been filed as refinements rather than refutations. Three critiques deserve to be taken seriously by any designer who uses the rule professionally.

1. The generalisability critique: short, clinical, aversive experiences are a narrow base

Kahneman’s core evidence comes from short experiences (seconds to an hour), clinical settings (medical procedures, lab-induced pain), and predominantly aversive stimuli. The argument that the same peak-end arithmetic governs a two-week holiday, a three-year subscription, or a seven-year romantic relationship is an extrapolation — a plausible one, and one supported by follow-up studies, but not a direct test. Miron-Shatz (2009) showed that when episodes get longer and richer, the peak-end model captures a smaller share of retrospective variance and other factors — narrative coherence, contextual meaning, the presence or absence of personally significant moments — rise in importance.

The practical implication is not that the rule becomes useless for longer experiences; it is that the peak and the end become necessary but not sufficient. A six-month online course that lands the final assessment gracefully and has a brilliant mid-course project is better than one that does not, but it is not automatically great. You still have to design for narrative, meaning, and contextual fit. The Peak-End Rule is the foundation; it is not the whole building.

2. The intensity-ceiling critique: what counts as a peak depends on the ceiling

Fredrickson (2000) and others have pointed out that “peak” is not a fixed quantity. It is relative to the ceiling the experiencing self can plausibly register within a given context. A gentle surprise on a boring flight can be a peak. The same surprise in the middle of a concert cannot be — the ceiling for peaks is higher in a context already designed for emotional intensity. This means that designing a peak is not just an exercise in “make something bigger”; it is an exercise in “make something unexpectedly big relative to the emotional baseline of the domain.” Designers who try to copy peak moves across domains without adjusting for the ceiling produce the corporate-retreat trust-fall phenomenon — a peak in one context that feels cringeworthy in another.

This is not a refutation; it is a craft qualification. The peak-end skeleton is correct, but the specific design moves that produce peaks are genre-bound. What works as a peak in a hotel may not work as a peak in a SaaS onboarding. What works as an ending in a game may not work as an ending in a medical procedure. The Octalysis Level 2 Experience Phases are partly a response to this problem — different Core Drives produce peaks at different ceiling levels in different phases.

3. The narrative-scaffolding critique: memory is a story, not an arithmetic average

The deepest critique comes from the autobiographical-memory literature, particularly the work of Dan McAdams on narrative identity and the life-story model of memory. Humans do not remember experiences as peak-and-end averages; we remember them as stories with characters, causality, turning points, and moral arcs. The peak and end of an experience are privileged partly because they tend to serve as natural story-pivots — the climax and the denouement — but what makes a moment memorable is not its raw intensity, it is its narrative weight within the story the remembering self is telling about itself.

This matters for design because a “peak” engineered without narrative scaffolding — say, a generic free-upgrade surprise with no story — performs worse than a smaller peak that lands at the right narrative moment. The best engagement design puts peaks where the story needs them: the first moment of real progress after early struggle, the first experience of belonging after initial isolation, the first real triumph after a sustained effort. The peak-end arithmetic is the physics of memory; narrative structure is the grammar. You need both.

The Brain on the Peak-End Rule

Kahneman published the original studies in the early 1990s, well before neuroscience had the imaging tools to test his claim at the level of circuitry. The intervening thirty years have been unusually generous to the peak-end architecture. The underlying neuroscience is not yet a fully closed story, but the direction of the evidence is clear and converges on the Kahneman finding from several independent angles.

The first strand comes from research on the hippocampus and the consolidation of episodic memory. Hippocampal encoding is not uniform across an episode — it is biased toward high-affect moments and toward boundary moments (beginnings, endings, scene changes). Studies using event-segmentation paradigms (Zacks, Speer and Reynolds 2009) show that the brain slices continuous experience into discrete events with identifiable boundaries, and that encoding spikes at those boundaries. The “peak” in the peak-end rule is exactly a high-affect boundary; the “end” is exactly a structural boundary. Memory encoding mechanisms privilege both.

The second strand comes from the affective neuroscience of salience. The amygdala tags emotionally salient moments for preferential storage. Cahill and McGaugh’s line of work on emotional memory consolidation shows that high-arousal moments — whether positive or negative — produce enhanced long-term retention compared to neutral moments of identical duration. The peak of an experience is by definition an emotionally salient moment. It earns more storage resources at the moment of encoding, and it preferentially survives retrieval later. Middle moments at lower arousal are compressed or forgotten entirely.

The third strand comes from the specific psychology of endings. There is strong evidence that the final moments of an experience are disproportionately re-activated during retrieval — the serial-position “recency effect” is one of the oldest findings in cognitive psychology (Ebbinghaus 1885; Murdock 1962). More recently, work on the “end-of-day effect” in affective forecasting (Ganzach and Yaor 2018) has shown that ratings of an entire day shift systematically with how the final hour went. The ending is not just encoded well — it is also retrieved first when we try to remember the experience, which means it anchors the global evaluation.

Put together, the neuroscience tells a coherent story: the brain encodes experience selectively, privileging high-affect moments and structural boundaries; it consolidates those moments preferentially through amygdala-modulated pathways; and it retrieves them first when constructing a global evaluation. The Peak-End Rule is not a quirk of introspection — it is the behavioural signature of how the brain stores and recalls hedonic episodes. The Octalysis language of Core Drive “spikes” and Experience Phase “transitions” maps onto the same architecture. When I talk about designing a spike in Core Drive 7 (Unpredictability & Curiosity) at the right moment of the onboarding journey, I am speaking Kahneman’s language in a game-designer’s accent.

The Peak-End Rule vs Other Theories

The Peak-End Rule shares a shelf with several adjacent frameworks that designers frequently confuse with it. Understanding the differences tells you which tool to reach for when.

Peak-End Rule vs Recency Effect

The recency effect is the much older finding, dating to Ebbinghaus and the early serial-position studies, that the final items in a list are recalled better than middle items. The Peak-End Rule subsumes and extends this. Recency is one component — the final moment matters — but Kahneman’s contribution was to show that the intensity peak matters independently and with equivalent weight, and that duration is neglected even for experiences of dramatically different length. If you are thinking about the ending in isolation, you are thinking about recency. If you are thinking about the peak-and-end average and the irrelevance of middle duration, you are thinking about the full Peak-End Rule.

Peak-End Rule vs Primacy Effect

Primacy — the better recall of early items — is the mirror of recency, and it shows up in memory for lists more than for hedonic episodes. The Peak-End Rule does not privilege beginnings the way it privileges peaks and endings, which is a genuine and sometimes counter-intuitive asymmetry. First impressions matter for their own reasons (initial framing, expectation setting, social signalling), but the retrospective evaluation of a hedonic experience is dominated by peak and end, not by the opening. This is why onboarding design that front-loads delight without also engineering a mid-journey peak and a strong endpoint often under-performs in retention metrics.

Peak-End Rule vs Prospect Theory

Kahneman’s Prospect Theory describes how people evaluate choices under uncertainty — loss aversion, reference-point dependence, and diminishing sensitivity. The Peak-End Rule describes how people remember completed hedonic episodes. Both come out of Kahneman’s broader program, but they apply at different moments: Prospect Theory tells you how a customer decides whether to start an experience; the Peak-End Rule tells you how they decide whether to repeat it. The full experience-design loop uses both — Prospect Theory to get them in the door, Peak-End to get them back through it.

Peak-End Rule vs the Power of Moments framework

Chip and Dan Heath’s The Power of Moments (2017) is the design-facing successor to the Peak-End Rule. It accepts the peak-and-end architecture as baseline and adds a taxonomy of peak types — Elevation (moments that rise above the everyday), Insight (moments of sudden understanding), Pride (moments of recognised achievement), and Connection (moments of shared significance). The Heath brothers are not refuting Kahneman; they are giving designers a vocabulary for which peaks to build. If Kahneman is the physics, the Power of Moments is the engineering handbook. Most of my applied work uses both in tandem.

The Peak-End Rule in the Real World

The rule’s reach is much broader than its 1993 clinical origins. Any system that depends on sustained human engagement — a product, a service, a course, a brand — runs into the peak-end structure whether the designers know the vocabulary or not. Four domains where the framework still earns its keep:

Customer service and support

The service recovery paradox — the well-replicated finding that customers who experience a problem that is resolved with exceptional service rate the company higher than customers who had no problem at all — is the Peak-End Rule wearing a suit. The problem is the peak (a high-intensity negative moment); the recovery is the end (a transformed final frame). Average the two and the remembered experience can net positive. Companies like Ritz-Carlton, Zappos, and JetBrains have built their service cultures around this arithmetic: empower front-line staff to engineer peak-recovery moments with real budget, and design the end of the interaction — the “is there anything else I can help with?” moment — as a deliberate mic-drop, not an afterthought.

Product design and onboarding

In digital products, the first session is typically too short to contain a real peak-end arithmetic; the rule applies instead at the multi-session level. What matters is that the user experiences a peak within the first few sessions — a moment of visible mastery, an aha-insight, a social recognition event — and that each session ends in a clean, non-frustrating state rather than dropping the user at the moment of peak confusion. Products that get this right (Duolingo’s streak celebrations, Notion’s first completed database, Figma’s first shared-prototype moment) build durable retention. Products that don’t (any onboarding that ends with the user staring at a blank dashboard wondering what to do next) bleed second-session activation regardless of how clever the middle of the onboarding was.

Loyalty programs and retention design

The loyalty-program industry has relearned the Peak-End Rule painfully over the last decade. Linear, flat earning programs — 1 point per dollar, 100 points for a reward — produce memories that are mathematically forgettable because there are no peaks and every ending is the same. The programs that have broken through in 2026 — Starbucks Rewards’ surprise double-star days, Sephora Beauty Insider’s tier-up celebrations, Nike Membership’s member-only drops, and the growing wave of “moment-driven” loyalty arcades I have written about in the Loyalty Program Tier List — all engineer deliberate peaks (surprise upgrades, tier-up theatre, birthday experiences) and deliberate endings (the post-redemption frame, the app closing screen, the next-visit invite). The Yu-kai Chou tier-list data consistently shows that Level-4 and Level-5 programs (community and meaning) outperform Level-1 and Level-2 (transactional) programs on every non-coupon retention metric, and the peak-end explanation is the cleanest one: higher tiers build peaks and endings that transactional tiers cannot.

Education and learning design

Education is an unusually good test bed for the Peak-End Rule because learners can compare experiences and remember them at a wide range of time scales. The learning products that scale — Khan Academy’s mastery structure, Duolingo’s league theatre, freeCodeCamp’s project portfolio completion ceremony, the graduation rituals at App Academy and other well-designed bootcamps — are all built on engineered peaks (first code that runs, first real project shipped, first public presentation) and engineered endings (the completion certificate, the alumni handshake, the final celebratory moment). Products that teach well but end poorly — a course that just stops when the syllabus is exhausted — are rated lower than objectively worse courses with better endings. The data on student ratings of semester-long courses shows the pattern clearly: the last week’s experience accounts for a disproportionate share of end-of-term satisfaction.

The Elephant in the Room

Here is the thing almost nobody writes about the Peak-End Rule: it is ethically complicated. Once you know that memory is dominated by peaks and endings, you have a lever that works whether or not the customer’s actual experiencing self was well served. This is the design move that keeps casino hosts, discount retailers, and certain reality-television producers up at night — they are very, very good at engineering peaks and endings that camouflage mediocre or actively harmful middles.

The casino case is the cleanest example. A slot-machine session can be designed to produce one or two memorable peaks (a mid-size jackpot, a near-miss with lights and sirens) and a graceful ending (the “bonus round” that pays out just before you leave), knitted around a long middle of net losses. The remembering self walks out with a “great night.” The experiencing self has been systematically financially harmed. The Peak-End Rule is the mechanic that makes this specific form of exploitation possible, and it is exactly why my White-Hat versus Black-Hat Octalysis distinction exists.

The ethical version of peak-end design is the one that keeps the experiencing self honest while also designing the remembering self well. Build peaks out of real value (a genuine moment of mastery, a genuine recognition, a genuine creative breakthrough) rather than out of manufactured intensity. Design endings that honestly summarise what the customer just received, not endings that camouflage what they did not. The test I give teams is simple: if your customer’s experiencing self kept a full diary of the experience, would the remembering self’s summary be a fair compression of that diary, or a flattering lie? If it is the lie, you are using the Peak-End Rule against your user. The short-term metrics will look good; the long-term brand will not.

The honest version, which is what I try to teach: the Peak-End Rule is neutral. It describes a fact about memory, not a prescription for what to do. Used with integrity, it gives you the license to invest experience budget where it actually matters for retention. Used without integrity, it gives you the license to extract value from customers who will only realise later that they were being managed rather than served. Choose White-Hat deliberately. The framework will work either way; only one of the two builds a career and a reputation.

How to Apply the Peak-End Rule with the Octalysis Framework

The cleanest way to operationalise the Peak-End Rule in 2026 is to read it through the Octalysis Framework. Kahneman gave you two privileged moments — peak and end. Octalysis gives you eight Core Drives, each one a specific kind of motivational energy, and four Experience Phases (Discovery, Onboarding, Scaffolding, Endgame) that tell you where in time to deploy them.

The mapping I use when I walk senior teams through this has four moves.

Move 1: Identify where your peak currently lives and where it should live

For any customer journey — a product onboarding, a loyalty program lifecycle, a course, a service visit — plot the emotional intensity curve as honestly as you can. Most teams find that their current peak is either accidental (a byproduct of an unplanned moment) or misplaced (a peak in the Discovery phase when it needed to be in Scaffolding). Identify which Core Drive is producing the peak. In most digital products, the peak is a Core Drive 2 (Development & Accomplishment) moment — a visible mastery milestone. In most loyalty programs, it is a Core Drive 7 (Unpredictability & Curiosity) or Core Drive 4 (Ownership & Possession) moment. In most service encounters, it is a Core Drive 5 (Social Influence & Relatedness) moment. Name it, then ask: is that the peak I would have chosen if I were designing from scratch?

Move 2: Own the ending by designing the Endgame phase deliberately

In Level 2 Octalysis, the fourth Experience Phase is the Endgame — the phase after the customer has reached the main mastery plateau of the product or program. Most teams treat this phase as an afterthought: “we got them to power user, now what?” The Peak-End Rule says this is exactly backwards. The Endgame is where the last impression is written, and the last impression is one of the two data points that the remembering self will use to decide whether to recommend, return, or renew. Design the Endgame as its own product. Invest Game Technique #85: Meaningful Choices and Game Technique #93: Meta-Game mechanics here. Build in recognition moments, alumni identity, and community-elder roles. A well-designed Endgame is the difference between “that was a good phase of my life” and “I’ve moved on.”

Move 3: Stop investing in the forgettable middle

The Peak-End Rule’s harshest operational lesson is that engineering effort spent polishing the middle of the experience generates the lowest return on memory investment of any design work you can do. If you have a fixed budget — and you always do — move it out of the middle and into the peak and the ending. This is not an argument for neglecting the middle; the middle has to clear a competence bar, or the peak and end won’t even get reached. But once the middle is at “solid competent” level, further polishing produces shrinking returns. The next dollar goes into a bigger peak or a sharper end.

Move 4: Use Game Techniques to build peaks that match the Core Drive ceiling

Once you know which Core Drive should produce the peak, the Game Techniques layer tells you how. Want a Core Drive 1 (Epic Meaning) peak? Deploy Game Technique #1: Narrative, Game Technique #44: Higher Meaning, and Game Technique #31: Elder. Want a Core Drive 2 (Development & Accomplishment) peak? Use Game Technique #4: Progress Bar landing on a dramatic milestone, Game Technique #14: Badges at the right tier of significance, and Game Technique #32: Crowning. Want a Core Drive 5 (Social Influence & Relatedness) peak at the ending? Use Game Technique #11: Social Treasure, Game Technique #51: Conformity Anchor, and Game Technique #118: Mentor. The Heath brothers’ Elevation, Insight, Pride, and Connection map almost perfectly onto Octalysis Core Drives 1, 3, 2, and 5 respectively, and the Game Techniques are the specific design moves that instantiate each of them. Kahneman and Tversky gave us the physics; Octalysis gives us the engineering diagrams.

Practical Steps to Apply the Peak-End Rule

Here is the step-by-step protocol I run when a team asks me to redesign an experience through the Peak-End lens. It is deliberately compact — you can work through it in an afternoon with a whiteboard, a customer-journey map, and access to real user data.

Step 1: Plot the honest intensity curve

Map the full customer journey on a timeline from first touch to current-state end (for subscription products, a “first 90 days” window is usually right). For each major moment, rate the emotional intensity on a -10 to +10 scale based on real user feedback, not executive opinion. NPS verbatims, exit interviews, app-store reviews, and session recordings are your raw material. The result is a jagged line. Your peak is the highest point; your end is the last point. Everything else is the middle.

Step 2: Identify your current peak and honestly grade it

Look at the highest point on the curve. What was the moment? Which Core Drive produced it? Was it designed, or did it happen by accident? Grade its intensity on an absolute scale — not “compared to other moments in this experience” but “compared to the ceiling of what is possible in this category.” Most teams find their peak is a B-minus at best, because they didn’t realise they were supposed to be designing one.

Step 3: Identify your current ending and grade it separately

Now do the same for the final moment. The ending is almost never the same as the peak — it is the last thing that happened, not the most intense. What is the literal final interaction your customer has with you in the current journey design? For most subscription products, the honest answer is “an email receipt they didn’t read.” For most loyalty programs, it is “the transaction completion screen.” For most courses, it is “the last page of the last module.” Most endings are accidental. That is a design opportunity.

Step 4: Pick the Core Drive you want the new peak to spike on

Choose deliberately. For most White-Hat engagement design, the highest-impact peaks are Core Drive 2 (visible mastery milestone), Core Drive 3 (a moment of real creative expression with feedback), or Core Drive 5 (a moment of being genuinely seen by a community). Avoid Black-Hat peaks (Core Drive 6, Core Drive 7, Core Drive 8) unless the category demands them and your integrity check passes. Pick the Core Drive, then list the three highest-impact Game Techniques for it, then design the specific moment the customer will experience.

Step 5: Design the ending as a deliberate last impression

The ending is the moment you rewrite the remembering self’s summary. Design it to do exactly two things: (1) deliver a genuine, non-manipulative emotional closure — a “you made it” moment that honestly reflects the journey, and (2) open the door to the next journey — the next subscription cycle, the next course, the next visit. Avoid the two failure modes: endings that feel abrupt (no closure) and endings that feel clingy (too much next-step salesmanship). The best endings have the structure of a good handwritten thank-you note: genuine, specific, brief, warm, and ending with an open door rather than a closed one.

Closing Thoughts

The Peak-End Rule is the finding that changed how I think about experience design more than any other piece of behavioural-economics research. It told me that my job was not to design an experience; my job was to design a memory. It told me that most engagement budgets are spent on the forgettable middle because that is where the engineering work is most visible. And it told me that the difference between a loyalty program people tell their friends about and a loyalty program people forget is usually two design moves: a deliberate peak and a deliberate ending.

The version I teach at conferences and build into Octalysis is the 2026 edition of Kahneman: the core arithmetic is right, the clinical origins are narrow but the rule generalises, the critiques are fair but not fatal, and the neuroscience has been unusually kind to the peak-and-end architecture. Use it as your first-pass lens on any engagement design. Use Octalysis as the turn-by-turn navigator. Ship a peak, own the ending, stop polishing the middle, and you will outperform 90% of the experiences you compete with — because most of them still believe the customer is rating the full duration, and the customer has never rated the full duration in their life.

If you want to go deeper, the Octalysis Framework is the most thorough operationalisation of the peak-and-end architecture I know of, and my book Actionable Gamification is the step-by-step version of how to design through the peak-end lens for real products. Both start with the same premise Kahneman proved in 1993: you are not in the business of delivering moments to your customer. You are in the business of delivering memories. And memories are peaks and endings, almost all the way down.

If your team is sitting on a product, loyalty program, or customer journey where the middle is competent but the peaks and endings feel accidental, that is exactly the problem my advisory work is built around. Reach out through the contact page — I take a small number of engagements each quarter, and a peak-end-first audit is usually the cleanest way to start.

Frequently Asked Questions

What is the Peak-End Rule in simple terms?

The Peak-End Rule is a cognitive heuristic identified by Daniel Kahneman, Barbara Fredrickson, Charles Schreiber, and Donald Redelmeier in 1993. It says that when people look back on an emotional experience, their overall memory of it is dominated by two moments — the emotional peak (whether pleasant or painful) and the final moments — while the total duration of the experience and the intensity of its middle are almost entirely forgotten. This means the memory that drives future choices is a compressed summary, not a faithful recording, and designers who optimise for memory rather than moment-by-moment experience produce better retention.

Who discovered the Peak-End Rule?

The rule was formalised by Daniel Kahneman and colleagues in two landmark papers — When More Pain Is Preferred to Less: Adding a Better End (1993) by Kahneman, Fredrickson, Schreiber and Redelmeier, and the colonoscopy study (1996) by Redelmeier and Kahneman. It was later expanded in Kahneman’s 2011 book Thinking, Fast and Slow and in his TED talk The Riddle of Experience vs Memory. Kahneman won the 2002 Nobel Memorial Prize in Economic Sciences for his broader work on judgement under uncertainty, of which the Peak-End Rule is one of the most design-relevant findings.

What is duration neglect and why does it matter for design?

Duration neglect is the finding that the total length of an emotional experience has almost no effect on how it is remembered. In the Kahneman cold-water study, doubling the length of a painful experience made it remembered as less aversive if the added time had a gentler intensity. For design, this means “shorter is kinder” is the wrong default for unpleasant processes (returns, cancellations, clinical procedures) and “longer is more valuable” is the wrong default for pleasant ones (vacations, subscriptions). What matters is the shape of the intensity curve, not the area under it.

What is the difference between the experiencing self and the remembering self?

Kahneman’s distinction separates the self who lives each moment of an experience (the experiencing self) from the self who constructs a compressed summary of that experience afterwards (the remembering self). The experiencing self integrates moments honestly over time; the remembering self edits them into a peak-and-end summary and uses that summary to decide whether to repeat the experience. They are two different clients with two different preferences, and design that only serves one of them leaves significant retention value unused.

Does the Peak-End Rule apply to positive experiences as well as negative ones?

Yes, with some asymmetry. The original studies were on aversive experiences, where the effect is strongest. Follow-on research on holidays, vacations, films, concerts, and consumer services has shown the same peak-and-end pattern for pleasant experiences, but the duration-neglect effect is slightly softer for very long pleasant episodes — a two-week holiday does accumulate some remembered value above a one-week holiday of equivalent peak and end. For design purposes, treat the rule as strong for most experiences and strongest for painful ones.

What are the main criticisms of the Peak-End Rule?

Three critiques are academically serious. First, the original evidence comes from short, clinical, aversive experiences — generalising to long, rich, meaningful episodes is plausible but not directly demonstrated. Second, “peak” is context-relative, so copying peak designs across domains without adjusting for the category ceiling produces miscalibrated designs. Third, memory is narratively structured, not arithmetically averaged — peaks and endings matter partly because they function as natural story pivots, which means a “peak” without narrative scaffolding may under-perform a smaller peak at the right story moment. None of these refute the rule; they refine how it is used.

How does the Peak-End Rule apply to product and UX design?

Map the full customer journey, identify the current emotional peak and current ending, and ask whether each was designed or accidental. In most digital products, the peak should be an early-win mastery moment (Core Drive 2) or an aha-insight (Core Drive 3); the ending should be a clean, honest session close that invites the next session. Stop spending engineering effort polishing the forgettable middle once it is at a competent baseline, and move that budget into engineering the peak and the ending. Products that get this right build compounding retention; products that don’t churn despite high in-session satisfaction scores.

How does the Peak-End Rule relate to the Octalysis Framework?

Octalysis gives you eight Core Drives and four Experience Phases. The Peak-End Rule tells you that Core Drives are not weighted evenly across time — a Core Drive spike at the peak moment and a Core Drive landing at the ending shape the remembered experience, while the same Core Drives in the middle are largely forgotten. The Octalysis Endgame phase (the fourth Experience Phase) is structurally the ending moment the Peak-End Rule privileges, and it is the phase most teams neglect. Using the Peak-End Rule as a lens on Octalysis deployment turns Core Drives from a static checklist into a time-aware design plan.

Can the Peak-End Rule be used unethically?

Yes — and it regularly is, most visibly by casinos, certain discount retailers, and some reality-television production. If memory is dominated by peak and ending, a designer can engineer those two moments to camouflage a middle that is not well serving the customer’s experiencing self. This is a Black-Hat Octalysis deployment and the short-term metrics will usually look good, but the long-term brand consequence is loss of trust when customers eventually compare notes. The ethical test I give teams: if the customer’s experiencing self kept an honest diary, would the remembering self’s summary be a fair compression of that diary, or a flattering lie? If it is the lie, you are using the rule against your user.

What are the most important practical design moves from the Peak-End Rule?

Four moves, in order of impact. First, plot the honest emotional-intensity curve of your current customer journey and identify the peak and the ending. Second, redesign the peak deliberately — pick the Core Drive you want it to spike on, pick the Game Techniques that will instantiate it, invest engineering budget there. Third, redesign the ending — treat the final interaction as a designed artefact, not an afterthought, and aim for a genuine closure that invites the next visit. Fourth, stop polishing the middle once it is at competent baseline, and move that budget into the peak and ending. The cumulative effect on retention and referral metrics is almost always larger than any single feature ship.

References

1. Kahneman, D., Fredrickson, B. L., Schreiber, C. A., & Redelmeier, D. A. (1993). When more pain is preferred to less: Adding a better end. Psychological Science, 4(6), 401–405.
2. Redelmeier, D. A., & Kahneman, D. (1996). Patients’ memories of painful medical treatments: Real-time and retrospective evaluations of two minimally invasive procedures. Pain, 66(1), 3–8.
3. Redelmeier, D. A., Katz, J., & Kahneman, D. (2003). Memories of colonoscopy: A randomized trial. Pain, 104(1-2), 187–194.
4. Kahneman, D. (2011). Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
5. Kahneman, D. (2010). The riddle of experience vs. memory. TED talk, February.
6. Fredrickson, B. L. (2000). Extracting meaning from past affective experiences: The importance of peaks, ends, and specific emotions. Cognition and Emotion, 14(4), 577–606.
7. Fredrickson, B. L., & Kahneman, D. (1993). Duration neglect in retrospective evaluations of affective episodes. Journal of Personality and Social Psychology, 65(1), 45–55.
8. Heath, C., & Heath, D. (2017). The Power of Moments: Why Certain Experiences Have Extraordinary Impact. New York: Simon & Schuster.
9. Mitchell, T. R., Thompson, L., Peterson, E., & Cronk, R. (1997). Temporal adjustments in the evaluation of events: The “rosy view.” Journal of Experimental Social Psychology, 33(4), 421–448.
10. Baumgartner, H., Sujan, M., & Padgett, D. (1997). Patterns of affective reactions to advertisements: The integration of moment-to-moment responses into overall judgments. Journal of Marketing Research, 34(2), 219–232.
11. Kemp, S., Burt, C. D. B., & Furneaux, L. (2008). A test of the peak-end rule with extended autobiographical events. Memory & Cognition, 36(1), 132–138.
12. Miron-Shatz, T. (2009). Evaluating multiepisode events: Boundary conditions for the peak-end rule. Emotion, 9(2), 206–213.
13. Zacks, J. M., Speer, N. K., & Reynolds, J. R. (2009). Segmentation in reading and film comprehension. Journal of Experimental Psychology: General, 138(2), 307–327.
14. Do, A. M., Rupert, A. V., & Wolford, G. (2008). Evaluations of pleasurable experiences: The peak-end rule. Psychonomic Bulletin & Review, 15(1), 96–98.
15. Verhoef, P. C., Antonides, G., & de Hoog, A. N. (2004). Service encounters as a sequence of events: The importance of peak experiences. Journal of Service Research, 7(1), 53–64.

Related Reading

• Prospect Theory: An S-Tier Behavioral Designer’s Guide to Loss Aversion
• Dual Process Theory (System 1 / System 2): Kahneman’s Two Minds
• Anchoring & Adjustment: An S-Tier Behavioral Designer’s Guide to Decision Design
• The Framing Effect: Gain vs Loss Framing in Behavioral Design
• The Octalysis Framework: Complete Gamification Framework

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?