The cold water experiment

In 1993, Kahneman and Fredrickson asked participants to submerge their hand in 14-degree water for 60 seconds. In a second condition, those same 60 seconds were followed by a further 30 seconds as the temperature rose slightly to 15 degrees — barely perceptible, but marginally less painful. When offered a choice of which version to repeat, a significant majority chose the longer one. More total pain, preferred.

The reason is the peak-end rule. People do not evaluate experiences by averaging across their duration. They evaluate them by two specific moments: the most intense moment — the peak — and the final moment. The middle is largely discarded. Duration barely registers. This is not a quirk of artificial laboratory conditions — Kahneman and Redelmeier replicated the finding in colonoscopy patients, where ratings of the procedure correlated almost perfectly with the average of the peak and final moments, and near-zero with how long it lasted. Deliberately adding a less painful minute at the end improved patients’ memories of the whole procedure and their willingness to return for future screenings.

The experiencing self and the remembering self

Kahneman’s framework for understanding why this happens distinguishes between two aspects of the self. The experiencing self lives through each moment as it happens. The remembering self constructs a narrative summary afterward — the story it tells about what occurred. These are not the same thing, and they are not equally weighted in determining future behaviour.

Loyalty, repurchase, and word-of-mouth are all functions of the remembering self’s story. The experiencing self’s moment-to-moment satisfaction determines what it feels like to be a customer. The remembering self’s narrative determines whether they come back and whether they tell others. A business that optimises entirely for the experiencing self — smoothing every friction, improving every moment of the journey — may be investing in exactly the moments the remembering self discards.

Kahneman shares the example of a man who listened to a symphony recording that ended with a terrible screech. Before the screech, he described the music as glorious. After it, he said it ruined the whole thing. The screech did not change the experience he had during the symphony. It only changed the story his remembering self now tells about it. This is the most commercially dangerous application of the rule for entrepreneurs: a negative ending does not produce a negative memory of the ending. It contaminates the entire preceding experience retroactively.

The beginning as the missing piece

The peak-end rule is technically about two moments, but research extends it to three. The beginning matters because it sets the reference point against which the peak is experienced. A strong beginning calibrates expectations appropriately so that a moderate peak registers as a genuine high. A weak or confusing beginning suppresses the emotional impact of even a strong peak because it distorts the baseline from which the peak departs.

This means the entrepreneur’s job is to design all three: a beginning that sets appropriate expectations without over-promising, a peak that meaningfully exceeds them, and an end that lands well in the remembering self’s narrative.

What the evidence shows across domains

A meta-analysis of over 30 studies confirmed that the peak and final moments dominate memory across healthcare, travel, customer service, and digital products. The effect is not a laboratory artefact — it is a consistent feature of how human memory evaluates bounded experiences regardless of context.

The commercial applications make the mechanism concrete. Mailchimp identified the campaign send moment as the most emotionally intense point in its user experience — high stakes, mild anxiety, no return — and invested in turning it from a stressful peak into a deliberately delightful one through animation and humour. The peak-end rule predicts the downstream memory: users who remember the send moment as fun and reassuring have a better memory of Mailchimp than users who remember it as stressful, even if the rest of the experience was identical.

IKEA’s ice cream stand at the exit is the most cited retail application. The deliberately positive end moment is cheap relative to the total investment in the store experience, yet it disproportionately shapes how customers remember and describe the visit. The labyrinthine showroom, the flat-pack anxiety — much of this is discarded by the remembering self because a 50p ice cream ended the experience on a positive note. The ending is cheap to design well and expensive to leave to chance.

The operational implications

Three specific prescriptions the research supports. Map the emotional intensity of your customer journey explicitly — not as a satisfaction score across every touchpoint but as a curve showing where the highest positive and negative intensities occur. The peak-end rule means your investment should be concentrated at these moments rather than distributed uniformly. If your current peak is a negative one — a stressful moment, a confusing process, a moment where something typically goes wrong — that is the first priority, regardless of what else is working well.

Design the end deliberately. Most businesses end the customer relationship by stopping communication — the moment the sale closes, the active engagement ends. The peak-end rule treats this silence as a final moment, and silence is a poor ending. A designed final touchpoint — even a simple one — produces a better remembering-self story than leaving the ending to chance.

And protect against negative endings specifically. A difficult cancellation process, poor offboarding, or an unresolved complaint as the last touchpoint does to your customers what the screech did to the symphony.

A book worth reading alongside this

The Power of Moments by Chip and Dan Heath is the most applied available treatment of how to deliberately engineer peak moments in customer and employee experiences. Their four elements of defining moments — elevation, insight, pride, and connection — provide an operationalisation of what the peak should be designed to achieve. For any entrepreneur who understands the peak-end rule theoretically and wants a practical framework for acting on it, this book is the most direct available bridge between the research and the design decisions.

Have questions about this article?

If any part of this article raised questions you want to explore further, courbot.co is built for exactly that. It is courben.co’s AI assistant, designed around the psychology of entrepreneurship. Ask it anything from this article.

This article is for educational and informational purposes only. Sources: Kahneman, D., Fredrickson, B.L., Schreiber, C.A. & Redelmeier, D.A. (1993), Psychological Science, 4(6). Redelmeier, D.A. & Kahneman, D. (1996), Pain, 66(1). Fredrickson, B.L. & Kahneman, D. (1993), Journal of Personality and Social Psychology, 65(1).