Designing dashboards is mostly the discipline of deciding what to de-emphasize. A field guide to hierarchy under heavy data.
When I designed the analytics and insights screens for a hospitality BI platform, the brief I was handed was, in effect, "show more data." Revenue and operations teams lived in it all day — occupancy, ADR, RevPAR, pace, pickup, channel mix, all updating in real time — and they wanted everything available. The tool already did that. It showed everything, all at once, with equal weight. And it was useless for making decisions.
The teams had a tell: they were exporting to spreadsheets to find what mattered. When people leave your interface to do the actual thinking, the interface has failed at its only job.
The real constraint in a data-dense product is not screen space — it's the user's attention. Every element you show competes for it. When everything is emphasized, nothing is, and the screen becomes a wall the user has to manually parse. The brief wasn't "show the data." It was "support the decision."
When everything is emphasized, nothing is. Hierarchy is the design.
I studied how revenue and operations teams actually worked — what they checked first, what they ignored — and the pattern was clear: they scan for outliers, glance at a handful of KPIs, and only occasionally need the full detail. So I built the screen as three explicit tiers of attention, each one earning its visual weight:
The old dashboard used color everywhere, which meant it signalled nothing — every screen looked alarming, so nothing did. I pulled color back to where it carries meaning: red for true exceptions, amber for watch-items, neutral for everything else. When color is rare, it works. When it's everywhere, it's wallpaper.
The last failure was structural: getting from a summary to the underlying records meant navigating away, which broke the user's train of thought. So detail became something you expand in place — summary and detail on one continuous surface, no screen change required. The number of context-switches needed to go from "something's off" to "here's why" dropped to zero.
In data-dense design, the hardest decisions aren't about what to show — they're about what to push back. Good information architecture is mostly the discipline of saying "not here, not yet" to everything that isn't the decision at hand. Treat attention as the scarce resource, and a wall of data becomes a tool people trust to tell them what matters.