What evidence would help you separate feedback severity from visible repetition?
A rare report can expose a serious blocked task, while frequent comments may concern a small preference. Decide how to prioritize investigation without equating volume with impact. Research concrete consequences and recovery options, keeping observed frequency within the reviewed set separate from severity and avoiding claims about total affected users or failure prevalence that public coverage cannot support.
Prioritize these feedback reports by concrete task consequences and recovery options, keeping severity separate from exact observed report counts and unknown prevalence.Define the observation and counting unit
Define the workflow and the consequence categories relevant to the decision. Include public accounts describing what could not be completed or recovered. Exclude emotional intensity as a severity measure. Record lost work, blocked access, extra steps and cosmetic concerns separately, using only explicitly described effects rather than inferring business loss, personal hardship or safety impact from an angry tone.
Trace the concrete task evidence
For each account, capture trigger, interruption, available workaround and whether the task eventually completed. A simple workaround may reduce urgency but still impose recurring burden. Keep irreversible loss claims attributed unless independently supported. Do not reproduce risky failures or ask for private records, and distinguish a reported consequence from a hypothetical worst case proposed by a commenter discussing what might have happened.
Challenge attribution and coverage
Compare severity under the same task requirements. A missing export matters differently when export is essential than when it is optional. Preserve low-frequency high-consequence cases for owner review. Count distinct inspected reports exactly, without scaling them to the whole user base or treating incomplete coverage as proof that a severe issue is isolated, rare or already resolved for everyone else using the product.
Hand off a bounded interpretation
Deliver an investigation queue with consequence, recovery and evidence confidence as separate fields. Keep final priority with the owner who knows operational context. If severity cannot be established, request the missing task consequence instead of guessing. Do not promise a fix date or claim that a ranking predicts customer retention, support cost or incident probability from a bounded public feedback sample alone.
Keep the boundaries clear.
- For feedback severity, reviewed public responses are not a representative customer sample.
- The feedback severity badges describe context, not guaranteed access or verified campaign reach.
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 2026-11-13. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.