Compare three to five rows from the same category using the same six fields: identity, photo evidence, sizing, price context, source relevance, and weight. Remove rows with unexplained gaps before deciding which one looks most appealing.
Build a fair comparison set
Start with one category and one intended use. A zip hoodie should not compete with a light T-shirt merely because both are clothing. Keep the variant range similar enough that differences in price, measurements, and weight mean something.
Three rows are usually enough to expose weak evidence. More than five often recreates the noise you were trying to remove. If several rows point to the same source, treat them as duplicates until the live pages show a meaningful difference.
Use the same six fields for every row
Identity
Category, style, variant, and intended use.
Photo evidence
Relevant angles and visible details, not image count.
Sizing
Measurements, method, and fit context where needed.
Price context
The live variant and what comparable rows include.
Source relevance
Whether the destination still matches the row.
Weight
Item or packed estimate, clearly labeled as an estimate.
Example: three fictional jacket rows
| Field | Row A | Row B | Row C |
|---|---|---|---|
| Identity | Light zip jacket | Vague “top jacket” | Lined work jacket |
| Photos | Front, back, lining, zip | Two promotional images | Front/back, cuffs, pockets |
| Sizing | Garment chart | Label sizes only | Chest and length shown |
| Price context | Comparable light layer | Low, but details missing | Higher with lining |
| Source | Live item matches | Redirects to shop home | Live item matches |
| Weight | Estimate shown | Not stated | Estimate shown |
The matrix does not declare a winner. It shows why Row B should be removed first and which tradeoff remains between A and C.
Use scores to expose gaps, not manufacture precision
A one-point-per-field checklist can reveal that a row is incomplete. It cannot prove that a row scoring six is objectively better than every row scoring five. Keep short notes beside the score so important differences do not disappear into one number.
Useful scoring
“5/6 — weight estimate missing; otherwise comparable.”
Misleading scoring
“92/100 — best choice,” with no explanation of how the number was produced.
Write the save reason and the removal reason
A saved row needs one concrete reason. A removed row should also have a short reason, such as “source mismatch” or “no garment measurements.” This small decision log prevents the same weak row from returning when you revisit the sheet later.
Keep: “Row A stays because its garment chart and lining photos answer my main questions.” Remove: “Row B goes because the link redirects and the size evidence is only a label.”