The Peak-End Rule: How Memory Shapes Customer Experience

Most customer experience programmes measure the wrong thing. They average satisfaction scores across every touchpoint, weight each interaction equally, and optimise for consistency. The psychology of memory does not work that way. Customers do not remember averages. They remember peaks and endings — and almost nothing else.

That is the peak-end rule, and understanding it properly changes how you design journeys, allocate service investment, and interpret feedback data.

What the Peak-End Rule Actually Says

The peak-end rule is a cognitive heuristic — a mental shortcut — describing how people evaluate and recall past experiences. Rather than integrating every moment into a running average, the mind compresses an experience into two data points: the emotional peak (the moment of greatest intensity, positive or negative) and the ending. Those two moments dominate the remembered experience, often overwhelming everything that happened in between.

The core insight: Customers do not remember your average. They remember your peak and your ending. Design accordingly — or cede control of memory to chance.

The heuristic was identified and formalised by Daniel Kahneman and colleagues through a series of studies in the early 1990s. One of the most cited — Kahneman, Fredrickson, Schreiber, and Redelmeier's 1993 paper "When More Pain Is Preferred to Less: Adding a Better End", published in Psychological Science — demonstrated that participants who experienced a longer, more painful medical procedure rated it as less unpleasant than a shorter one, provided the longer version ended with a period of reduced (though still uncomfortable) sensation. The total duration of discomfort was greater; the remembered discomfort was lower. Duration barely registered. The ending mattered enormously.

Kahneman later named this phenomenon duration neglect — the tendency to ignore how long an experience lasted when forming a retrospective judgement. For anyone designing customer journeys, this is not an academic footnote. It is an operational instruction.

Why This Matters More Than Your Average CSAT Score

The standard CX measurement apparatus — CSAT surveys distributed across touchpoints, NPS collected at arbitrary intervals — implicitly assumes that all moments contribute roughly equally to the customer's overall impression. The peak-end rule tells us this assumption is wrong in a specific, predictable direction.

Consider a hotel stay. A guest checks in smoothly, sleeps well, enjoys breakfast, and then spends forty minutes at checkout resolving a billing error. The checkout experience is the ending. It is also likely the emotional peak — frustration, waiting, a sense of being wronged. Every pleasant hour before it is functionally irrelevant to how that guest will remember and describe the stay. The checkout is what they will tell people about.

This is why customer experience strategy that focuses only on eliminating friction across the middle of a journey — the long, uneventful stretches — often fails to move loyalty metrics. Friction reduction in low-salience moments has diminishing returns. The memory-forming moments are elsewhere.

How the Peak-End Rule Interacts with Other Biases

The peak-end rule does not operate in isolation. It compounds with several other well-documented cognitive mechanisms that CX designers need to account for.

• Loss aversion (Kahneman and Tversky): Negative peaks register more powerfully than positive ones of equivalent intensity. A moment of genuine delight needs to be substantially more positive than a painful moment is negative to produce the same memory weight. This asymmetry means that protecting against bad peaks is at least as important as engineering good ones.
• The affect heuristic: People's current emotional state colours their retrospective judgement. A warm, confident ending puts the customer in a positive affective state at the moment of recall — which then colours how they reconstruct the whole experience.
• Confirmation bias: Once a peak or ending has set the emotional tone of a memory, customers tend to interpret ambiguous earlier moments through that lens. A bad ending reframes neutral touchpoints as slightly negative; a strong ending does the reverse.

Together, these mechanisms mean that the emotional arc of a journey — not its average quality — is what determines remembered experience. This is the foundation of structured journey design: mapping not just what happens, but when the emotional intensity rises and falls, and whether the peaks and endings are being actively designed or left to accident.

What "Designing for Memory" Looks Like in Practice

Applying the peak-end rule is not about manufacturing fake moments of delight. It is about making deliberate choices about where you invest service design effort and what you protect at all costs.

There are four practical implications worth building into your design process.

1. Identify the Peaks — Do Not Assume Them

The emotional peak of a journey is rarely where organisations think it is. A bank may assume the peak moment is account opening; customers may experience the peak as the first time they need to dispute a charge. Voice of customer research — qualitative, narrative, and mapped against the journey — is the only reliable way to locate where intensity actually concentrates. A robust VoC strategy should be designed to surface these moments explicitly, not just average satisfaction across stages.

2. Engineer the Ending Deliberately

The ending of a customer interaction is almost always under-designed. Checkout flows, offboarding processes, service resolution sign-offs, post-purchase communications — these are the moments that seal the memory, and they are routinely treated as administrative rather than experiential. A strong ending does not need to be elaborate. It needs to be warm, clear, and to leave the customer with a sense of completion and competence. Even a well-crafted confirmation message or a brief, genuine acknowledgement from a service agent can shift the emotional valence of the entire preceding interaction.

3. Protect Against Negative Peaks More Than You Invest in Positive Ones

Given loss aversion's amplification of negative peaks, the highest-return intervention is often not adding a delight moment but eliminating a pain spike. Identify the touchpoints in your journey where negative emotional intensity is highest — waiting, escalation, billing, complaint resolution — and treat them as priority design problems. This is where service design investment pays its clearest dividend.

4. Use Rituals to Anchor Positive Peaks

Signature moments — deliberate, repeatable rituals that create a strong positive peak — are one of the most durable tools in experience design. They work precisely because the peak-end rule means a single well-designed moment can define the memory of an entire interaction. The ritual does not need to be expensive; it needs to be specific, unexpected enough to register, and consistent enough to become associated with the brand.

The Measurement Implication: Stop Averaging, Start Mapping Arcs

If the peak-end rule is correct — and the experimental evidence strongly supports that it is — then averaging satisfaction scores across touchpoints is not just imprecise. It is systematically misleading. It hides the moments that actually drive memory and loyalty behind the noise of low-salience interactions that customers will not remember in a week.

The more useful measurement approach maps emotional intensity across the journey arc, identifies where peaks occur and whether they are positive or negative, and tracks the emotional state at journey end. This is a different analytical frame from the standard touchpoint-by-touchpoint CSAT grid — and it produces different, more actionable insights. If you are unsure where your organisation sits on this capability, a structured CX maturity assessment is a useful diagnostic starting point.

A Note on Ethical Design

The peak-end rule is a description of how memory works, not a licence to manipulate. Designing a strong ending to mask a genuinely poor experience is a short-term tactic that erodes trust the moment customers compare notes or return for a second interaction. The ethical application is to use the heuristic to prioritise where genuine quality improvements will have the most durable effect on customer perception — not to paper over systemic failures with cosmetic moments of warmth.

The organisations that use behavioral economics in CX design most effectively are those that treat these principles as tools for honest prioritisation, not for engineering impressions that the underlying experience does not deserve.

Memory Is the Product

Customers do not buy experiences. They buy the memories those experiences will create — and the stories they will tell as a result. The peak-end rule is the clearest empirical account we have of how those memories form. It tells you that the middle of your journey, however carefully optimised, is largely invisible in retrospect. What endures is the moment of greatest intensity and the feeling at the end.

That is not a reason to neglect the middle. It is a reason to be far more deliberate about the peaks and endings you are currently leaving to chance — because right now, in most organisations, they are exactly that.