Virality And Retention
May 3, 2026 ยท View on GitHub
Content Machine treats virality as an editorial and learning system, not as a guarantee. The goal is to make better shorts by checking hooks, retention, platform fit, packaging, and real outcomes before repeating a format.
Skill Sequence
Use this order when a short needs serious social-native review:
short-form-archetype-researchchooses the right format and quality gates.hook-packaginggenerates hook, title, cover, opening-frame, and metadata variants.short-form-production-playbookkeeps the script, visuals, captions, and review loop editorially coherent.virality-reviewchecks hook, retention, clarity, payoff, risk, and platform fit.retention-passcatches dead air, weak first frames, repetitive visuals, caption overload, and payoff problems.platform-packagingprepares platform-specific titles, descriptions, hashtags, CTAs, disclosures, and upload checklists.publish-prep-reviewremains the final rendered-video readiness gate.metrics-feedback-loopturns publish receipts and performance metrics into style-profile or niche-profile updates.
Rules
- Do not promise guaranteed views, reach, conversions, or algorithmic performance.
- Treat reference winners as structure to learn from, not footage to copy into a new render.
- Keep packaging truthful to the script and source evidence.
- Optimize for retention, saves, shares, qualified comments, follows, clicks, or conversions instead of raw views alone.
- Preserve artifacts so another agent can see why a hook, platform package, or style-profile update was chosen.
Minimal Artifacts
For a serious production run, expect:
packaging-variants.v1.jsonvirality-review.v1.jsonor a compact equivalent review noteretention-pass.v1.jsonor scene/timestamp-specific retention notesplatform_packaging.jsonpublish-prep/metrics-feedback.v1.jsonafter the short has real performance data
Practical Prompt
Use Content Machine's virality and retention workflow on this short.
Choose the archetype, generate hook/title/cover variants, review the
script or render for hook and retention problems, prepare platform
packaging, and do not call it publish-ready until publish-prep passes.
After publishing metrics exist, turn the result into repeat/kill/test
learning for the relevant style or niche profile.