x-analytics-loop

Review X performance and choose next experiments. Use when analyzing weekly post metrics, diagnosing winners and losers, or planning what to repeat.

Run this weekly.

Collect

For each post, ask for or infer:

• date/time
• topic
• format
• hook
• impressions
• likes
• replies
• reposts/quotes
• bookmarks if visible
• profile visits
• follows
• link clicks if relevant

Calculate

Use:

• engagement rate = engagements / impressions
• reply rate = replies / impressions
• repost rate = reposts / impressions
• profile-click rate = profile visits / impressions
• follow conversion = follows / profile visits

If data is missing, continue with qualitative diagnosis and mark assumptions.

Diagnose

• High impressions, low engagement: hook/retrieval worked, post underdelivered.
• Low impressions, high engagement: content resonated, distribution needs work.
• High replies, low follows: conversation works, profile may be weak.
• High profile clicks, low follows: bio/pinned post problem.
• High likes, low reposts: agreeable, not share-worthy.
• High reposts/bookmarks: make more templates/frameworks.

Decide

Pick:

• 3 things to repeat
• 3 things to stop
• 3 experiments for next week

Output

Return:

• Top winners
• Hidden winners
• Losers
• Signal diagnosis
• Next week’s experiments
• Posting plan

Operating Principle

Treat the public X algorithm as a signal map, not a cheat code. Optimize for posts and replies that earn useful positive actions (dwell, replies, reposts, shares, profile clicks, follows) while avoiding negative actions (not interested, mute, block, report, fast skips).