Quick answer
Compare three to five rows from the same category and intended use. Eliminate any row with a destination mismatch or missing blocker. Then compare evidence coverage, measurement confidence, source consistency, like-for-like price context and likely parcel impact. Keep the winner only if you can explain the tradeoff in one sentence.
Use two stages: eligibility, then preference
Most weak comparisons skip directly to preference. A lower price, cleaner thumbnail or popular label wins before anyone checks whether the rows describe the same product type, option or evidence level.
Stage 1: eligibility
Does the row match the category and intended use? Can the required photo, measurement and source checks be performed? If not, pause it.
Stage 2: preference
Among eligible rows, which candidate offers the clearest evidence and most acceptable tradeoffs for your stated question?
Compare five evidence fields
| Field | Strong comparison question | Weak shortcut |
|---|---|---|
| Photo coverage | Which row shows the decisive category details most clearly? | Which row has the most images? |
| Measurement confidence | Which figures have a clear method, unit and selected option? | Which row shows a familiar size letter? |
| Source consistency | Which destination still matches the row description and option? | Which source name sounds more trustworthy? |
| Price context | Are the same type, option and included parts being compared? | Which displayed number is lowest? |
| Parcel context | Which candidate has clearer weight or packaging information? | Ignore shipping until later |
Unknown is a valid comparison value
Do not reward a row for looking complete when the decisive field is absent. Write “unknown” in the same place you would write a number. This makes missing evidence visible and prevents a blank cell from disappearing psychologically.
Unknown does not always mean reject. A missing packed weight may be acceptable early in research; a missing garment measurement may be a blocker when fit is the main question. Label each field as required, useful or optional before comparing.
Example: three hoodie rows
| Field | Row A | Row B | Row C |
|---|---|---|---|
| Category/option match | Yes | Yes | Color conflict |
| QC coverage | Front/back/fabric | Front/back/cuffs/hood/fabric | Front only |
| Measurements | Chart; method unclear | Chest and length shown for selected size | Unknown |
| Source match | Matches row | Matches row | Destination title differs |
| Weight | Estimated item weight | Unknown | Unknown |
| Result | Keep as backup | Best evidence; weight still needed | Pause before ranking |
Row B does not “win” universally. It leads this comparison because the stated question prioritizes fit and construction evidence. Row A retains an advantage in weight context, and that tradeoff remains visible.
Use scores to reveal gaps, not manufacture certainty
A checklist score can show that one candidate answers more required questions. It should not turn subjective evidence into a false mathematical guarantee. If measurement confidence is essential, a one-point checkbox should not allow five optional details to cancel a missing fit requirement.
A safer method is:
- Mark required fields as pass, fail or unknown.
- Remove candidates that fail a true blocker.
- Count useful evidence only among the survivors.
- Write the final tradeoff in plain language.
How to break a tie
When two rows remain close, do not add arbitrary criteria. Return to the intended use. Which uncertainty would be harder to resolve later? A missing weight estimate may be easier to obtain than an absent measurement method; a small price difference may matter less than incomplete interior photos for a structured bag.
Tie-break rule: prefer the candidate that reduces the most important uncertainty, not the one with the longest feature list.
Common comparison questions
Can I compare products from different categories?
Only for a broad budget or parcel-planning question. Do not use the same photo, fit or construction criteria across unrelated product types.
Should I assign percentages to each field?
Only if the weights reflect a real decision and you can explain them. For most users, blockers plus a plain-language tradeoff are easier to audit.
What if the lowest-price row has weaker evidence?
Record that as the tradeoff. Do not treat the price as comparable until the product type, option, included parts and missing details are understood.
How often should I reopen a saved row?
Recheck before relying on it because destinations, options, prices and images can change. Record the date of the comparison when recency matters.