A social review should ask whether the post accomplished its job. Reach is distribution; it is not comprehension, trust, or a qualified inquiry.
Read performance in layers
For each post, record:
- Distribution: reach and impressions, with paid/organic separated.
- Attention: watch time, completion, or meaningful dwell where available.
- Response: saves, shares, substantive comments, replies, or profile visits.
- Action: link visits, messages received, qualified inquiries, or another defined next step.
Use the metric that fits the format; a static carousel and a video do not have identical attention measures. Compare like with like: same market, objective, placement, and a reasonable time window. Note paid spend, influencer amplification, or seasonal context.
Diagnose before declaring a winner
High reach and weak action may mean the content attracts curiosity but not the intended audience, or the next step is unclear. Low reach and strong response may mean a distribution problem despite a useful asset. A high click rate with few completed inquiries may point to the destination or offer. These are hypotheses to inspect, not labels from a dashboard.
Review a sample of actual comments and the landing path. Record what to retain, change, or test next. Do not attribute sales to a post without a reasonable link between the exposure and confirmed outcome; report correlation and attribution limits clearly.
Optional AI-assisted work brief — Purpose-based social review
For each post, compare its intended job with the matching evidence: qualified attention, saves, useful replies, referrals, or completed action. Segment by format and audience rather than ranking raw counts. Note distribution changes and sample limits. Convert one finding into a next test and one thing to stop doing; do not turn correlation into a platform algorithm rule.
Worked example — Read results by the post’s job
A how-to carousel is meant to be saved and revisited; a profile introduction may drive qualified profile visits; a service explainer may lead to relevant clicks. Compare each with its intended behavior and downstream evidence. A low like count can coexist with useful saves, while high reach can produce no qualified action. Avoid ranking every format on one engagement rate.
AI disclosure: AI tools assisted with drafting. No performance benchmarks are claimed.