The dilemma
Illustrative scenario
Nadia runs customer experience for a car-rental company with airport locations. She has budget for one major project this year.
Option one is the pickup counter. Operations data shows it is the longest wait in the journey, and it draws the most complaints while customers are still on site. Option two is the return. Today, many rentals end at a self-service kiosk with an unexpected fuel or damage charge and an argument about it; the redesign would add a quick walk-around with photos and a final total sent before the customer reaches the terminal.
The post-rental survey, sent the day after return, tells a different story from the operations data: the lowest ratings cluster around disputed return charges, not long pickups. The board cares about one number above all: whether customers book again.
The longest pain or the last impression: where should Nadia spend the money?
The short answer
The peak–end rule describes how people evaluate experiences after the fact: the overall judgment is strongly shaped by the most intense moment (the peak, which can be good or bad) and by the final moments, while total duration has surprisingly little influence, a pattern called duration neglect. The idea comes from research by Daniel Kahneman and colleagues in the 1990s on how people remember pain and discomfort.
How strong is the evidence?
Documented behavioral effect
An effect found in psychology experiments. Its size depends on conditions, and some findings replicate better than others.
This is a documented behavioral effect with laboratory and clinical support.
In a 1993 study, Kahneman, Fredrickson, Schreiber and Redelmeier had participants hold a hand in painfully cold water in two trials. One lasted 60 seconds. The other had the same 60 seconds plus 30 more, during which the water was warmed slightly but stayed uncomfortable. Asked which trial to repeat, most chose the longer one: more total discomfort, but a gentler ending. In 1996, Redelmeier and Kahneman asked patients undergoing colonoscopy or lithotripsy to rate their pain at regular intervals during the procedure. Their later overall evaluations tracked the worst and final moments, and the length of the procedure had little relationship to them. A 2003 randomized trial tested the idea directly (see the example).
Conditions matter. The rule describes retrospective judgments, and fits best for experiences with a clear emotional arc. Duration neglect is not total, and other features, such as whether an experience is improving or worsening, also play a part. How and when you ask changes the answer: a moment-by-moment rating, a survey the next day and a later decision to return measure different things.
Investigating an example
Documented study
A gentler ending in a medical procedure
Redelmeier, Katz and Kahneman randomized 682 colonoscopy patients. For one group, the procedure was extended by a short interval at the end that was less uncomfortable than what came before; the control group had a standard procedure. Patients given the gentler ending remembered the procedure as less unpleasant, even though it lasted longer.
For service design, the takeaway is that the final part of an experience can shape memory enough to be worth deliberate attention, independent of the total time spent.
Limits: this is a medical procedure, run under clinical ethics oversight, with memory ratings as the outcome. It does not show that a pleasant ending makes up for a bad service, and the size of the effect in a rental desk, a checkout or a support call is unknown until you test it.
A published study or record, linked below. Its limits are stated.
Choose your move
You are in the dilemma above. Which move would you make? Open each one to see its trade-offs.
A Spend the budget on cutting the pickup wait.
Helps every customer from the first minute, and a long wait may itself be the negative peak. But if return disputes stay, many rentals still end badly, and the effect on repeat bookings is uncertain.
B Redesign the return with a photo walk-around and a clear final total before customers leave.
Targets the ending most linked to low ratings and removes a source of disputes. But the survey is sent right after the return, which may exaggerate the ending's weight; if pickup frustration drives defection, repeat bookings may barely move.
C Split the budget between smaller fixes to both.
Easy to agree on internally, but may fix neither well. Unless the rollout is staggered by location, the team will not learn which change mattered.
No move wins in every business. The law tells you what to check, not what to choose.
How to apply it
- Map the journey by stage. Combine time spent, complaints and in-the-moment feedback for each stage, so you see where the experience peaks and how it ends.
- Find the peaks, good and bad. Identify the single worst moment and any genuinely good one. Negative peaks usually deserve attention first.
- Design the ending deliberately. Give the final step closure: no surprises, a clear confirmation, and an obvious next step if one exists.
- Measure memory and behavior separately. Use a short in-the-moment rating, a later retrospective survey and actual repeat behavior; they can disagree.
- Test by location or cohort. Change the ending at some sites and compare retrospective ratings and repeat bookings with matched control sites.
What to measure
Retrospective satisfaction a day or more after the experience, in-the-moment ratings at key stages, the share of experiences that end in a dispute, and repeat purchase within 12 months. Guardrail: if remembered ratings rise while complaints or total waiting time worsen, you may be polishing the memory while degrading the experience.
Worked example
| Stage | Rental A | Rental B |
|---|---|---|
| Pickup counter | 25 minutes, frustrating | 10 minutes, fine |
| Driving the car | Pleasant | Pleasant |
| Return | 5 minutes, clear receipt | 20 minutes, fee dispute |
| Total waiting | 30 minutes | 30 minutes |
| Likely memory under peak–end | Annoying start, clean finish | Sour ending dominates |
Illustrative scenario, not data from a real company. The last row is a prediction from the peak–end pattern, not a measurement. Rental A's long pickup may itself be a negative peak, so neither version is guaranteed to be remembered better; a test would tell.
Where it breaks
- It describes memory, not the experience itself. People still suffer through a long, bad middle even if they later remember it less vividly.
- Duration neglect is partial. Long waits still cost goodwill and time, and some customers will count them.
- Measurement can manufacture the effect. A survey sent minutes after a pleasant ending will over-represent the ending.
- A nice ending does not erase a severe failure. Lost luggage or a billing error remains the peak.
- Ethics: do not use a pleasant ending to mask a bad deal, such as a small gift at the end of a high-fee sale, and do not stretch unpleasant experiences to shape memory outside contexts with proper oversight.
What if you apply it incorrectly?
A telecom team reads "endings matter most" and adds a cheerful thank-you offer at the end of its cancellation flow, leaving the long hold queue before it untouched. Exit-survey scores tick up slightly. Complaints about the queue keep growing online, and the team has polished the one moment that no longer affects retention: the customer has already left.
Myth versus evidence
The social-video version
Only the peak and the ending matter, so the rest of the experience can be ignored.
What the evidence supports
Peaks and endings weigh heavily in remembered evaluations, but duration and the middle still affect people, and results depend on how and when you measure.
Five-minute field assignment
Pick one customer journey: an order, onboarding or a support ticket. List its stages, then use complaints or reviews to mark the single worst moment and the final moment. Check when your satisfaction survey is sent relative to that ending. Write down one change to the ending you could test at a single location or cohort.
Checkpoint
Two questions: one about the idea, one about a decision. Answer both correctly to master this law. Questions are drawn from a larger bank and graded on our server; a retry brings different questions.
Evidence ledger
Every factual claim in this lesson, where it comes from, and how far it has been checked.
-
In the 1993 cold-water study, most participants chose to repeat a longer trial that ended with slightly less painful water over a shorter trial.
Limit: Laboratory pain; verify trial durations (60 s and 60 + 30 s) in the paper.
-
Patients' retrospective evaluations of colonoscopy and lithotripsy tracked peak and final pain, with procedure length having little relationship.
Limit: Medical procedures; correlational design.
-
In a randomized trial of 682 colonoscopy patients, a gentler extended ending led patients to remember the procedure as less unpleasant.
Limit: Clinical setting with memory ratings as outcome; effect sizes not generalizable to services.
-
The car-rental scenario and the two-rental table.
Limit: Teaching scenario; no real company.
Further reading
- 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.
- 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.
- Redelmeier, D. A., Katz, J. & Kahneman, D. (2003). Memories of colonoscopy: A randomized trial. Pain, 104(1–2), 187–194.
Which other law could explain this decision?
Editorial review: Ali Dezhami, November 8, 2026. Last substantive update: November 8, 2026. Our methodology