OpenAI says GPT-6 Astra may mark the start of the AGI era. Your Tuesday post can still fail for the same reason: impressive numbers are not the same as useful evidence.

Gather the last 4 weeks of captions + insights screenshots. For each post, calculate saves ÷ reach. Check comments for actual questions.

Above roughly 5% + questions = repeat.
High likes + near-zero saves = drop.
Anything else stays unclassified.

Map each repeat format to an open plan slot. Use 4 consistent weeks; fewer is too noisy.

Hypothetical cafe audit: the recipe reel scores 200 ÷ 2,000 = 10%, but has no question comments recorded, so it is not yet confirmed. The team photo has 150 likes, 2 saves and 1,800 reach, so mark it drop, low commitment. The discount post has question comments but only 5 ÷ 900 = 0.56%, so it does not pass. The before-and-after cake post scores 300 ÷ 2,400 = 12.5%, but its comments are not recorded as questions, so it remains unconfirmed.

Success check: every post has a ratio, a question-comment check and a repeat, drop or unclassified decision.

If the shortlist feels too short, keep it short. Do not promote an unqualified post to reach three formats.

Save this and run it on your last 4 weeks of posts.

#ContentAudit #ContentStrategy #SmallBusinessMarketing #SocialMediaWorkflow #MarketingMetrics