The user is separating perceived performance from backend latency. They likely measured API response times and found them acceptable, but users still complain about slowness.
I should avoid jumping straight to server-side optimizations and instead frame the answer around what people actually experience.
Perceived slowness usually lives in the gap between user action and visible feedback:
- input latency before the request fires
- loading state placement and timing
- layout shifts when data arrives
- whether optimistic UI gives immediate acknowledgment
For a dashboard, I should also consider chart re-renders, independent card fetches, and skeleton dimensions that do not match the final layout.
Ranking likely bottlenecks for this scenario:
1. No immediate feedback on filter or refresh actions
2. Skeleton cards that collapse or expand when data lands
3. Waterfall requests where each widget fetches on its own
4. Empty-state flashes because stale data is not shown while revalidating
The answer should stay practical and ordered by impact, not by how technical each fix sounds.
The API can be fast while the interface still feels slow.
Start by checking:
- whether the button gives immediate feedback
- whether skeletons match the final layout
- whether any reflow happens after data arrives
- whether the loading state appears near the content it replaces