Quick answer
Compare three to five rows from the same category. Give every row the same columns: intended use, QC evidence, measurements, source-link match, price context, shipping weight and the biggest unanswered question. Remove rows that fail a required condition, then order the remaining options by preference.
Define the job before collecting options
Write one sentence that describes what the item needs to do. “A light jacket with usable chest and length measurements” creates a comparison. “A good jacket” does not. The sentence should name the category and the one or two conditions that would make a row unusable.
This step prevents the sheet from changing your goal every time a new image looks interesting. It also separates a real requirement from a preference. A required measurement can remove a row. A preferred color should matter only after the row passes the requirements.
Build a small comparison set
Three to five candidates are usually enough to expose meaningful differences without turning the review into another spreadsheet. Choose rows that serve the same job. If a hoodie and a jacket solve different needs, compare them in separate sets even if both appear under clothing.
Do not open ten links simply because they are available. First remove obvious mismatches from the preview: wrong category, missing option, irrelevant photos or a source destination that no longer supports the row.
Use the same columns for every row
| Column | Question | Useful entry | Weak entry |
|---|---|---|---|
| Intended use | Does this row solve the job? | “Light outer layer” | “Looks good” |
| QC evidence | Which useful views are present? | Front, back, cuffs, lining | “Many photos” |
| Measurements | What can be compared directly? | Chest 58 cm, length 70 cm | “Size L” |
| Source match | Does the destination support the row? | Exact product and option visible | “Weidian link” |
| Price context | How does it compare with similar rows? | Mid-range; clearer sizing than lower option | “Cheap” |
| Weight context | What is known about mass or volume? | Known weight or labeled estimate | Blank |
| Open question | What could still change the decision? | Lining material not shown | “Need more research” |
Separate evidence from preference
Evidence
Measured dimensions, visible photo angles, option labels, source-page match, current listed price and a known or clearly estimated weight.
Preference
Color choice, styling, tolerance for bulk, acceptable price range and which tradeoff feels worthwhile after required checks pass.
Preferences are valid, but they should not masquerade as proof. “I prefer the second shape” is an honest way to choose between two well-supported rows. “The second is better quality” is not supported unless the visible evidence actually demonstrates a relevant difference.
Use removal rules before score rules
A score can hide a critical failure. A row may collect points for price, photos and source relevance while still lacking the one measurement required for fit. Set removal rules first: wrong category, no usable size evidence, incompatible specification, mismatched destination or another condition that makes the row unsuitable for your stated job.
Then use the seven-point checklist to compare only the survivors. This keeps a high total from compensating for a non-negotiable problem.
Compare price as a tradeoff, not a winner
Write what the higher or lower price changes. A lower-priced row with unclear sizing may require more research. A higher-priced row with better measurements may reduce uncertainty but still be poor value for your priorities. The important comparison is not the number alone; it is the package of evidence, fit for purpose and expected parcel consequence around that number.
A current external price may also change. Date the observation and avoid treating an old screenshot or spreadsheet cell as a live quote.
Make shipping weight visible early
Weight and volume belong in the first comparison pass for shoes, structured bags, heavy outerwear and electronics. If one row includes a box or rigid packaging and another does not, note the difference rather than comparing item prices as if the parcel consequence were equal.
Use known QC weight and size when available. Otherwise label the entry “estimate” and record the assumption. A rough estimate is useful when its uncertainty remains visible; it becomes misleading when formatted like a guaranteed quote.
Let the biggest unknown decide the next click
Each row should have one open question. If the same question appears in every row, you have discovered a category-level research need. If one row has a unique, answerable gap, look for that evidence. If the gap cannot be resolved and it affects a requirement, remove the row.
- “Need more photos” becomes “Need an interior view of the base seam.”
- “Unsure about size” becomes “Need flat chest width for size L.”
- “Link looks wrong” becomes “Current destination is a store page, not the exact item.”
- “Shipping unclear” becomes “Need item weight without retail box.”
A simple decision sequence
- Remove candidates that fail the intended use or a required condition.
- Mark every remaining fact as observed, stated by the page or estimated by you.
- Find the largest unresolved question for each row.
- Check only the gaps that could change the decision.
- Choose the survivor whose evidence and tradeoffs best fit your priorities.
- Keep one backup only if it solves a meaningfully different risk.
Copy a compact comparison row
Use: [what job this solves]
Evidence: [best useful photo or measurement]
Source: [exact product / store / unclear]
Price + weight context: [observed or estimated]
Largest unknown: [one specific question]
Decision: [keep / research / remove] because [reason]
Know when the comparison is finished
Stop when one or two rows meet the intended use, pass the required checks and have no unresolved question large enough to change the decision. Continuing to browse after that point often adds novelty rather than information.
A shortlist is complete when you can explain why each remaining row is there and why the removed rows left. The explanation is the useful output—not the number of links collected.
