Proof Studio / Specimen gallery
Microsites you can open, compare, and inspect .
These are the smaller side-by-side surfaces from the demo corpus: color, composition, voice, grid, perception, product decisions, and typography. Each one shows the cold output, the system-backed output, the measured value, and the boundary on the claim.
Interactive reviewer
Pick a specimen. The proof opens in the frame.
The selector changes both the live frames and the measured row below. This is a gallery of artifact-level checks, not a universal performance benchmark.
Campaign review page
Color hierarchy specimen
Buyer question
Can the same brief become a clearer visual system when the run has a retained palette contract?
The backed output uses fewer colors and far fewer near-value collisions while preserving the same campaign task.
441 cold
18 backed
Lower is better: fewer color pairs collapse into the same visual weight.
66 cold
21 backed
The backed surface is more disciplined without becoming monochrome.
7 cold
0 backed
The backed surface keeps value hierarchy inside the palette.
What to inspect
- The primary action still reads as the primary action.
- Status and support regions no longer compete with the hero.
- The page survives as a designed artifact, not a pile of unrelated accents.
Boundary
One fixture about palette discipline. It does not prove conversion lift or brand preference.
Buyer question
Does the page create one deliberate reading path instead of evenly weighted blocks?
The backed output creates a stronger dominant element and records the composition system it is using.
2.889 cold
6.905 backed
Higher means the first visual decision is more decisive.
No cold
Yes backed
The backed output names the layout mode instead of only arranging content.
4 cold
3 backed
The backed layout relies less on default centering.
What to inspect
- The hero, schedule, and registration areas have distinct jobs.
- The backed version creates a clearer first read.
- The surface is still usable as a small public page.
Boundary
Composition metrics are proxy checks. They need reader testing before they become a performance claim.
Buyer question
Does the backed run keep the same product but make the promise sharper?
The backed output trims the page and moves the copy toward concrete route outcomes.
318 cold
294 backed
The backed version says less while keeping the product surface intact.
44 cold
46 backed
Shorter sentences carry more of the load.
4.55% cold
4.35% backed
The copy stays active enough that this is not the main lift claim.
What to inspect
- The backed headline names the operational miss, not a vague capability.
- The calls to action fit a buyer review path.
- The page avoids inflated AI product language.
Boundary
This is a copy-quality specimen, not a statistically tested messaging study.
Buyer question
Can retained layout rules reduce ad hoc width math?
The backed output replaces hand-computed widths with shared grid rules while preserving responsive structure.
17 cold
3 backed
Lower means less custom layout math to maintain.
0 cold
34 backed
The backed surface connects layout decisions to a reusable system.
0 cold
6 backed
The width system is explicit enough to review.
What to inspect
- The article rhythm changes without losing content hierarchy.
- The backed surface uses named layout primitives.
- The result is easier to retheme or extend.
Boundary
The metric measures code structure in this output, not the long-term maintenance cost of a team.
Buyer question
Does the surface guide attention without drawing boxes around everything?
The backed output uses proximity and salience tokens, reducing drawn grouping while preserving the target signal.
0 cold
52 backed
The backed surface uses spacing as an explicit perception tool.
0 cold
22 backed
The attention system is named in the artifact.
46 cold
39 backed
The backed version needs fewer hard outlines to group information.
What to inspect
- The important operational state is easier to find first.
- The surface uses spacing and scale before extra borders.
- The shift board keeps its task density.
Boundary
This is a visual-perception fixture. It needs user observation before becoming a workflow-speed claim.
Buyer question
Does the page move from outputs to testable product outcomes?
The backed output names outcomes, assumptions, and the riskiest bet more explicitly.
0.08 cold
0.73 backed
Higher means the artifact talks more about buyer outcomes than produced outputs.
10% cold
21% backed
More assumptions are tied to a test path.
No cold
Yes backed
The backed page points attention at the decision that should gate the roadmap.
What to inspect
- The roadmap reads as a decision surface, not a delivery inventory.
- The backed version gives a buyer something to agree with or reject.
- The proof does not depend on hidden JSON.
Boundary
The measured terms are text/artifact checks. They do not prove roadmap success in market.
Buyer question
Does retained type guidance make the page easier to read and maintain?
The backed output removes ad hoc pixel font sizes and introduces a bounded text measure.
11 cold
0 backed
Lower means the type system is easier to scale and audit.
0 cold
23 backed
The backed output anchors type decisions in reusable roles.
No cold
Yes backed
The backed article constrains line length for reading.
What to inspect
- The backed article has a clearer reading measure.
- Headings and body copy use a more stable hierarchy.
- The artifact can move into a design system without rewriting every size.
Boundary
This is an artifact-quality check. It does not prove reader comprehension without a reader study.
Full gallery matrix
Every specimen has an output and a value.
This matrix is the public doorway to the demo corpus. The deeper evidence registry can promote individual specimens into full case studies as each claim is reviewed.
Specimen Measured value Cold Backed Open