Designing for the second click
2 min read
Most design reviews judge an interface the way a stranger would: cold, curious, seeing it for the first time. That matters, but it is a tiny slice of how anyone actually spends time with software. The first click is a handshake. Everything after it is the relationship.
The second click is where an interface either gets out of the way or starts to grate. It is where a playful transition turns into a wait, where a clever empty state becomes clutter, and where a tooltip you appreciated once becomes something you swat away every morning.
Novelty has a half-life
Anything designed to delight decays. A bouncy modal is charming on day one and a tax by day thirty. That doesn't mean delight is wrong — it means it should be spent where it compounds, not where it repeats.
A useful test: imagine the interaction on its two-hundredth run. If it still feels good, keep it. If it only feels good because it is new, make it faster, quieter or optional.
Delight that repeats becomes friction. Delight that rewards becomes trust.
Three things I check on every screen
- What does this look like for someone who already knows it? Expert paths deserve as much design as onboarding.
- What is remembered between visits? Scroll position, the last tab, the filter someone set yesterday.
- What can be skipped? Every confirmation and intro should have a quiet way around it.
Remembering is a feature
The cheapest way to respect a returning visitor is to remember what they did. It doesn't need a database — often a few lines are enough:
function useRememberedTab(key: string, fallback: string) {
const [tab, setTab] = useState(() => {
if (typeof window === "undefined") return fallback;
return localStorage.getItem(key) ?? fallback;
});
useEffect(() => {
localStorage.setItem(key, tab);
}, [key, tab]);
return [tab, setTab] as const;
}
It is a small thing. But small things, repeated a few hundred times, are what a product actually feels like.
Designing the long tail
When I present work now, I try to show the second visit alongside the first. What does the dashboard look like with real data, three weeks in? What happens to the empty state once it is never empty again?
It makes for less impressive slides. It makes for much better software.