Stacking unrelated views on one dashboard splits attention across questions nobody asked together

What it is

Every view is a request on attention: it asks to be scanned, decoded, and its conclusion held. Stacking unrelated views on a dashboard submits dozens of requests at once — the user's core task (the two or three questions this screen must answer) drowns in graphical noise unrelated to it. "Let's add a chart here" is the most common first step of dashboard decay.

Why it happens

The scattering follows from scarce attention and serial scanning: visual scanning proceeds by salience and position, so the more views, the less attention each receives; unrelated views also create processed-but-useless cost — every scanned graphic must be classified as relevant or not, and classification itself spends cognition. The subtler cost is misplaced salience: high-contrast graphics inside irrelevant views steal attention that belonged to key views, and the user's first glance lands on decorative numbers instead of abnormal metrics. Stacking compounds — every "keep it for now" view lowers the whole screen's signal-to-noise ratio until the core view needs scrolling to be found.

Where it stops holding

"Unrelated" is judged against the task, not taste: the finance-detail chart is irrelevant to the operations daily brief but relevant to the monthly review — the same views have different values in different scenarios, and the correct move is splitting dashboards by scenario rather than stacking every possibly-relevant chart onto one screen. Trimming also needs an outlet: removed views move to a detail page or a "more views" area; the problem being solved is the first screen's signal-to-noise ratio, not the deletion of information.

Applying it

• Run a quarterly view audit: for each view ask "which task does it serve, and did anyone cite it last month" — two noes means it moves to a secondary page.
• Every new view must declare the question it answers and its difference from existing views; no declaration, no first screen.
• Verification: measure the attention distribution on the dashboard (or scrolling and clicking proxies); when core views' attention share falls below long-tail views, the stack is overfull — start trimming.

Related

• Same group: U7.01.1 Multiple views divide distinct questions · U7.01.2 Views need an explicit data relationship to each other
• Nearby: U7.07.3 One screen's capacity is bounded by reading ability · U7.10.5 Dashboards nobody views should be retired, not kept
• Search terms: dashboard clutter · attention economy · signal to noise dashboard