Skip to content

Build a comparison row evidence contract

Build a comparison row evidence contract: A row contract exposes mismatched conditions behind a broad mobility label.

Guides

What should a comparison row evidence contract contain?

Consider a cart comparison where one account describes carpet and another shows an empty cart on tile. A row contract exposes mismatched conditions behind a broad mobility label. It preserves both accounts' limited usefulness without forcing a winner or implying matched evidence from unlike public tasks, surfaces and unreported loading conditions.

Artifact schema and decision check
Output columns: comparison_row; common_requirement; option_a_basis; option_b_basis; condition_mismatch; conclusion_state; required_match.
Decision check: Write a bounded conclusion or mark the row unresolved when sources answer different questions. Avoid composite scores and popularity tie-breakers.
01 / Guides

Identify the concrete review task

Define the common rolling task, surface and loading requirement before comparing. Do not choose whichever conditions make each option strongest. A broad row heading must not conceal separate questions about carrying, steering and rolling that the reviewed public accounts answer in different ways or leave unaddressed.

02 / Guides

Inspect admissible supporting material

Include explicit surface and load details. Exclude guesses and promotional adjectives without task descriptions. Inspect Montage originals where available; inaccessible context cannot be filled with a typical floor material or assumed empty state merely because complete cells look better than honest unknowns in the comparison table.

03 / Guides

Assemble the decision artifact

Put each passage beside the requirement and mark missing conditions. Separate unknown from observed incompatibility. Material excluded from this row may remain useful for another question; exclusion from a matched comparison does not itself demonstrate poor product performance or make the source irrelevant to every other claim.

04 / Guides

Check the inference and action boundary

Write a bounded conclusion or mark the row unresolved when sources answer different questions. Avoid composite scores and popularity tie-breakers. Automated synthesis can organize passages but cannot manufacture matching conditions or independent confirmation missing from the original descriptions the article proposes to use for its comparative verdict.

05 / Guides

Deliver the bounded review state

Deliver wording, mismatches and required matching evidence. Keep the task visible in the table legend. The contract supports a fair reviewed comparison, not a market-wide ranking, universal mobility verdict or guarantee that readers' different surfaces, contents and loading patterns will reproduce the reported outcomes from those specific public examples.

Keep the boundaries clear.

  • The contract supports a fair reviewed comparison, not a market-wide ranking, universal mobility verdict or guarantee that readers' different surfaces, contents and loading patterns will reproduce the reported outcomes from those specific public examples.

Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.

Primary references

Documentation reviewed 2026-11-15. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.

Know what social media shows about you.
Start with a question.

Get started
montage