The Drop Score · presentation study

Confidence & Coverage,
said plainly.

The current readout — “C− (5.7), full coverage, 34% confidence” — makes two honest signals fight each other. “Full” sounds great, “34%” sounds broken, and the reader can’t tell whether to trust the grade.

The core reframe: that 34% isn’t the tool doubting its own math. It’s six real reviewers who showed up and disagreed. That’s a consensus story, not a confidence story. Name it right and the number explains itself. (For a quiet set the same low number means “barely anyone weighed in” — so every option below surfaces the reason, not just the percent.)

Today

C− (5.7) · full coverage · 34% confidence

Two badges, two opposite vibes, zero guidance on what to do with it — and “data-light” is engine jargon nobody outside this repo will parse.

01

Twin meters, plain English

Split the two axes into two labelled bars, each with a verdict sentence underneath. Breadth and agreement stop competing because each one says what it means. Best for the full set page.

C−5.7 / 10UCS X-Wing
How much we graded9 of 9 points

Whole set graded. Nothing was skipped for lack of data.

How much the crowd agreesLow · 34%

Split room. Six reviewers, no consensus — they ranged from rough to decent.

02

One verdict sentence

Collapse both numbers into a single human read that tells you what to do with the grade. The math hides in a quiet data line for the nerds. Most self-explanatory; great as a card subtitle.

Split verdict

We graded the whole set — but six reviewers came in split. Take this as a read, not gospel.

Coverage 9/9 points  ·  Agreement 34%  ·  built from 6 mined reviews + 18 Brickset ratings

03

Named status (kills “data-light”)

Turn the two axes into a small vocabulary of named states, so every grade carries one honest label instead of a jargon word. Signal bars show agreement; pips show breadth. Best as a compact badge on timeline / list views.

Locked in
Solid read
Split verdict
Early read
Too quiet
Split verdictC− · 5.7 — whole set graded, crowd disagrees
9/9 graded
04

The receipts (show the spread)

Make 34% concrete by showing it: where each reviewer actually landed. A wide scatter is the disagreement, visible at a glance — no percentage to interpret. Most convincing; honesty you can see.

The receipts

Built from 6 video reviews + 18 Brickset ratings

5.7 avg
0 — rip-off510 — grail

Reviewers ranged from D to B−. That gap is the 34% — the grade is real, the room just isn’t unanimous.

every point graded
05

Minimal word-band swap

The smallest change: keep the one-line shape, swap the jargon for words and a reason. A drop-in rewrite of the §7 line. Ship-today option if you want one quick win.

Before

C− (5.7) · full coverage · 34% confidence

After

C− · 5.7 — whole set graded · consensus: shaky (6 reviewers, split)
Rock solid
Solid
Mixed
Shaky
Thin

My pick

Rename the axis, then layer 02 → 04. Stop calling depth “confidence” and call it consensus / agreement — that one word change makes the X-Wing’s 34% instantly legible. Lead the card with the one verdict sentence (02), expand into the receipts spread (04) for anyone who taps in, and carry the named status (03) as the compact badge everywhere else. “Data-light” retires into “Early read” / “Too quiet.”

All five use the live X-Wing numbers and only restyle the presentation — the engine’s coverageTier, confidence, and per-cluster scoredCount already provide everything shown here.