What the data actually shows
How we remember experiences is heavily shaped by two moments. Daniel Kahneman's work on the 'peak-end rule' found that retrospective judgments of an experience are dominated by its most intense point (the peak) and its ending, rather than by the average of every moment or how long it lasted. A worse experience that ends well is often remembered more fondly than a better one that ends badly — which is why endings carry outsized weight in memory.
People also tend to prefer improving sequences. Research by George Loewenstein and Drazen Prelec found that, given a choice, people often favour outcomes that get better over time over ones that decline — they'll even forgo a larger reward now to have things on an upward trajectory. The shape of the sequence, not just the total, affects how good it feels.
But the present self has its own claims. Research on delayed gratification and the trade-offs of waiting shows that hunger, impatience, and the diminishing certainty of the future all push against deferral. Saving the best for last only pays if you actually reach 'last' — and circumstances, appetite, and chance don't always cooperate.
Why this feels different from how it actually is
We feel the pull to save the best for last partly because anticipation itself is pleasurable — looking forward to the good part can extend the enjoyment. That makes deferral feel virtuous, like we're being disciplined and savouring. The catch is that we tend to forget how much can change before 'last' arrives.
It also feels different because we underrate how much endings dominate memory. In the moment, every part of an experience feels equally real, so it's not obvious that the final stretch will define how we recall the whole thing. The peak-end effect operates quietly, mostly visible only in hindsight.
And we swing between two errors. Some people defer almost everything, building toward a finale that risks never coming; others consume the best first and let experiences trail off. Neither is obviously right, because memory and momentary living pull in different directions — and we rarely notice which one we're optimising for.
Saving the best for last only pays if you actually reach 'last' — and appetite, circumstance, and chance don't always cooperate.
What the research says to do about it
When you can shape how something will be remembered, end on a high. Because the peak and the ending dominate recall, a strong finish — the best dish last, the best stop at the end of a trip — tends to leave a more positive memory than front-loading the good parts. Designing for a good ending is a cheap way to improve the story you keep.
Lean toward improving sequences for things that unfold over time. The Loewenstein and Prelec research suggests people generally feel better about trajectories that get better, so arranging an evening, a project, or a meal to build rather than fade tends to feel more satisfying than the reverse.
But don't defer enjoyment so long you miss it. The delayed-gratification trade-offs are real: the future is uncertain, appetite and attention fade, and the best part you keep postponing can simply never arrive. The balanced approach is to savour and live now while still ending on a high — not to sacrifice the present entirely for a memory that depends on getting to the end.
What the research says does not help
Always deferring the best part doesn't reliably make experiences better, and it carries a real cost: by the time 'last' arrives you may be too full, too tired, or out of time to enjoy it — or it may never arrive at all. Treating saving-the-best as an absolute rule ignores that the present self has legitimate claims.
Always consuming the best first isn't a clean solution either, because it tends to leave experiences declining toward a weak ending — and endings disproportionately shape the memory. A good start followed by a fade often reads, in hindsight, as a worse experience than it was.
Optimising purely for the memory at the expense of the moment doesn't help when you're actually living the experience, not just storing it. The peak-end rule describes how we remember, not how we feel minute to minute; over-engineering the memory can quietly drain the enjoyment of being there.
The peak-end rule describes how you'll remember something, not how it feels while you're in it.
What this looks like in real life
The meal you eat in the wrong order
Eat the favourite dish first and the meal fades toward a weaker finish — which, in hindsight, tends to read as a worse meal than it was, because endings carry outsized weight in memory. Save the favourite for last and the meal builds to its best moment. The catch is the opposite risk: defer it until you're too full to enjoy it, and you trade a sure pleasure now for a memory that never fully forms.
The trip whose best stop kept getting pushed back
Planning the standout destination for the end of a trip can leave a stronger overall memory, since the ending disproportionately shapes recall. But keep deferring the good part — waiting for the perfect day, the right mood — and circumstances, weather, or time can run out first. Saving the best for last only pays if you actually reach 'last.'
Real numbers in context
This is a 'which goal are you serving' question rather than a numbers one, and it would be dishonest to attach precise figures to it. The robust findings are qualitative: retrospective judgments are dominated by the peak and the ending (Kahneman's peak-end rule), and people generally prefer improving over declining sequences (Loewenstein and Prelec). Both are about shape and emphasis, not magnitude.
The practical balance is the takeaway. For memory, end on a high and let things build. For the lived moment, weigh that against hunger, impatience, and the plain risk that endlessly deferred enjoyment is enjoyment you never actually have. The two goals genuinely conflict, and the honest move is to hold both rather than pick a single rule.