Eight signals that the comparison needs a reset
1. The goal keeps changing
You began with daily shoes and are now comparing boots, runners, and statement sneakers. Rewrite the job and keep only one lane.
2. Different items share one score
A rigid bag and a soft tote cannot be ranked on the same construction or parcel assumptions. Split the search.
3. The cheapest row controls attention
Displayed price may belong to another variant or exclude practical costs. Confirm comparable options before comparing price.
4. Polished photos replace evidence
Styled images can identify an intended look without showing measurement method, current item details, or variant continuity.
5. Measurements have no basis
A number without the measured object, method, and unit cannot support a fit decision.
6. Several rows lead to the same place
Duplicate destinations create the illusion of choice. Keep the row with the clearest description and evidence.
7. Every external page starts a new search
A source page should answer a named question. Leave when it becomes unrelated discovery.
8. No tab has an exit condition
If you cannot say what would close a tab, it will remain open regardless of evidence.
Match the symptom to the repair
| Symptom | Likely cause | Repair |
|---|---|---|
| More than six active product tabs | No candidate limit | Keep three; a new one must replace the weakest. |
| Price changes the winner repeatedly | Options or included parts are not comparable | Record the exact variant before price. |
| Photos look strong but notes stay empty | Image purpose is not classified | Label reference, seller, variant, or QC-style evidence. |
| Two candidates cannot answer the same question | Mixed use cases | Rewrite the purpose and separate the second search. |
| Every row is on hold | The required evidence is unrealistic or unavailable | Revise the requirement or restart in a better category. |
Use a five-step recovery
Recover the original job
Write what the item should do, not which label first attracted attention.
Close other categories
Save unrelated ideas elsewhere. Do not keep a second search inside the current comparison.
Deduplicate destinations
Compare URLs, product identity, option sets, and images. Keep one representation of the same source.
Protect the strongest two
Choose candidates that answer the same need with the clearest photos, measurements, and source continuity.
Name one decisive question
Resolve it, hold one row with a next action, or restart without forcing a winner.
Measure progress by subtraction
A healthy session has fewer candidates, fewer unknowns, and more consistent evidence fields. Dropping every option can be the right result when the available pages do not support the requirement.
Do not preserve sunk browsing time
Time spent opening a row is not a reason to keep it. Preserve only evidence that still helps the decision.
