The report as an instrument dial: current score, trajectory, movement. The verdict band's first two organs, miniaturized. Answers "how are we doing" — but not "what's wrong."

Notes: calm, scannable, honest about trajectory. But an 89% and a 83% read as neighbors when one hides a failing control — urgency lives in a tone color alone, which is exactly one device but a quiet one.

The card as a briefing: lead with the single most urgent item, count the rest. c68's worst-thing-first verdict treatment, applied per report. Answers "what's wrong" — but a clean quarter renders four polite sentences and no numbers.

Notes: the GRC persona's Monday-morning question, answered. But trajectory vanishes — you can't see the needle move — and the board card's "worst item" is its whole headline, which overflows the register.

The card as the newest snapshot on the shelf: run id, timestamp, score, movement since the previous retained run. Speaks the evidence-trail language — schedulable, re-runnable, retained. Answers "is the paperwork current" more than "is the fleet healthy."

Notes: perfectly matched to Mark M's snapshot mechanics and the only candidate that surfaces when the report last ran — but it buries today's state behind last run's, and a monthly cadence means the card can be three weeks stale by design.

The card wears the artifact's identity: a tiny white page. Leans on the document face — reports are files you hand people, so the landing shows four files. Identity over density.

Notes: the strongest identity statement — "your landing page is four reports" is legible at a glance — and the weakest instrument: state hides in a caption, trend is gone entirely. A thumbnail answers "what is this," not "do I need to open it." Expanded: seven hierarchy-driven versions in report-card-minidocs.