You Can't Make Every Moment Good. Engineer Two.

You have a finite budget of polish. Spread it evenly across every screen and you’ll ship a product that is uniformly fine and completely forgettable. That’s not a motivational line — it’s a consequence of how memory actually files an experience, and once you see it, you can’t design the same way again.

Here is the uncomfortable study. Donald Redelmeier and Daniel Kahneman tracked patients through medical procedures that were, at the time, genuinely painful, asking them to rate their discomfort moment to moment. Then they asked, afterward, how bad the whole thing had been. The remembered rating had almost no relationship to the total amount of discomfort — or to how long it lasted. It was predicted, with eerie accuracy, by just two numbers: the worst moment (the peak) and how it felt at the end. In one arm of the work, deliberately extending a procedure with extra time at a milder level of discomfort — objectively adding more total pain — left patients remembering the whole ordeal as less bad, because it now ended gently. More suffering, better memory.

They called the two culprits the peak-end rule and duration neglect. Your experiencing self lives through every minute. Your remembering self — the one that decides whether to come back — keeps a highlight and an ending and discards the runtime.

Now put that against how most products get built. Effort gets smeared evenly — every flow tuned to be pretty good, nothing allowed to be either transcendent or rough. It feels responsible. It’s actually the worst allocation available, because a flat experience has no peak to remember and no ending to speak of. You’ve spent your whole polish budget manufacturing a thing the remembering self will file under fine and never mention to anyone.

The move is almost aggressive by comparison. Pick the one moment your product can make genuinely great — the moment it does the thing nobody else does, the flash of “wait, that’s delightful” — and pour disproportionate effort there. Let some ordinary screens stay ordinary to fund it. Then, separately, design the end of every meaningful session with real intention, because the last taste is half of what gets remembered. Not the empty state you never revisited. Not the spinner that fades to a dashboard. The actual final beat — the confirmation, the little “done,” the hand-off — is load-bearing, and it’s usually the most neglected pixel in the whole product.

Two caveats, because the wedge only works if the science stays honest. This is about memory and the decision to return, not about the lived minute — you still can’t let the middle be genuinely painful, or the peak becomes a peak of frustration. And a great ending doesn’t rescue a broken core; it decides how a good experience gets remembered, not whether a bad one gets forgiven. Within those bounds, though, the instruction is clear and a little freeing: you were never going to make every moment good. Stop trying. Engineer two.