Social Format Playbooks

1) One strong idea is not one universal export

Suppose a team creates a 70-second 16:9 educational video and plans to publish the same export, hook, caption, CTA and KPI on every social surface. One surface may crop critical text, another may classify the asset differently, and the dashboard may merge views that do not mean the same thing.

In this lesson we build Channel Format Playbook Matrix v1. One content_intent remains stable while every channel/surface gets an explicit contract for geometry, duration, hook, caption, CTA, rights, publish state and metric semantics.

Central rule: intent can be shared across channels; format and measurement must remain surface-specific and versioned.

2) COM-012 scope: channel playbooks, not Human Anchor or Winner Repurposing

COM-012 owns the channel/surface contract. COM-011 owns Human Anchor + AI Variants and COM-013 owns Winner Repurposing بدون Copy. Those outcomes remain separate.

No live publishing, rights adjudication or universal-best-practice claim occurs. The lab is synthetic and offline.

3) Mental model: Intent → Channel/Surface → Format Contract → Rights/Publish Gate → Metric Contract

For each playbook row ask five questions in order: what is the intent; which target surface; what asset geometry/duration is currently accepted; what are the rights and publish states; and exactly what does the success metric count under which semantic version?

A platform name plus a generic KPI is not an auditable playbook.

4) FACT: YouTube Shorts format and duration constraints

Current official YouTube guidance says square or vertical videos up to three minutes can qualify as Shorts, with date/channel details documented in the same guidance. This is a platform FACT, not a universal short-form standard.

In the Matrix, aspect_ratio_class and duration_class trace to current evidence for that surface. Reusing a landscape master without checking the target surface is not valid adaptation.

5) FACT: the views label survived while its semantics changed

From 24 August 2026, YouTube counts a public view when playback begins across formats. engaged views and qualified views remain distinct for analytics, earnings or eligibility uses.

Therefore metric_definition cannot just say “views.” It must state what is counted, while metric_version_date binds the experiment to the semantics in force. Results from before and after a definition change are not automatically comparable.

6) Playbook structure: 18 fields for a reproducible decision

Each row includes playbook_id, channel, surface, content_intent, audience_context, asset_type, aspect_ratio_class, duration_class, hook_pattern, caption_contract, cta_contract, rights_state, publish_state, metric_definition, metric_version_date, experiment_ref, owner and evidence_ref.

Twelve are critical. Missing decisive evidence does not become synthetic success; the state remains HOLD_MISSING_EVIDENCE or a narrower non-success state.

7) Geometry and duration: adapt before final export

asset_type, aspect_ratio_class and duration_class are decided before final production. If a surface needs vertical/square handling and the team blindly reuses a 16:9 master, the correct state is BLOCKED_FORMAT_MISMATCH.

When a platform changes an accepted duration rule, version the affected playbook rather than rewriting unrelated channels.

8) Hook, caption and CTA may change while intent stays stable

The same bounded content intent can use a faster hook on one surface, more context on another, and a different CTA elsewhere. Those changes are channel-native expression, not failure.

If the adaptation changes the underlying intent, it is no longer the same playbook and its experiment relationship must be rebuilt.

9) Do not collapse Rights State into Publish State

publish_state=READY does not mean rights_state=CLEARED. An asset can be format-ready while a music, testimonial or image ingredient still has unresolved rights.

MAKNOW routes unclear rights to BLOCKED_RIGHTS. Prior publication or layout approval is not rights evidence.

10) Metric contract: each channel needs an explicit measurement definition

A bare views field is insufficient for cross-channel comparison. The playbook must say what behavior/event the metric counts and which semantic version applies.

Replacing a channel metric definition with generic views yields REVISE_GENERIC_METRIC. Exposure, engaged behavior, qualified outcomes and conversion can only be compared under explicit contracts.

11) Experiment reference and comparability: playbook version is part of the evidence

experiment_ref must remain tied to the playbook version used in the test. If the metric definition or accepted duration changes, previous results may remain historical evidence but are no longer automatically direct-comparison evidence.

REBASELINE_COMPARABILITY tells the team to update only affected playbooks and create a compatible experiment/baseline while preserving the core intent and unaffected channel versions.

12) Major failure modes and the correct typed state

• landscape master without target-surface check → BLOCKED_FORMAT_MISMATCH
• duration outside the current rule → REVISE_DURATION_CONSTRAINT
• metric stored only as generic views → REVISE_GENERIC_METRIC
• unresolved rights → BLOCKED_RIGHTS
• stale platform rule → STALE_PLATFORM_ASSUMPTION
• changed metric/duration semantics invalidate direct comparison → REBASELINE_COMPARABILITY
• critical evidence genuinely missing → HOLD_MISSING_EVIDENCE

Typed states stop “almost ready” from becoming publish-ready by assumption.

13) Lab and transfer: update only the affected playbooks

COM-012-LAB1 contains eight offline fixtures: baseline, generic metric, landscape mismatch, rights failure, stale rule, missing evidence, duration violation and a changed-context transfer.

In transfer, one platform changes its view definition while another channel changes accepted duration. The learner versions only those affected playbooks, preserves core intent, and explains why the earlier experiments are no longer directly comparable.

14) Bounded decision, freshness and what this lesson does not prove

This lesson does not declare one format or KPI universally best and performs no live publication. Its output only determines whether current evidence supports an auditable Channel Format Playbook or must remain HOLD/BLOCK/REVISE/REBASELINE.

YouTube is vendor-current evidence and should be rechecked by 2026-10-18 or earlier after a material change. The playbook schema, typed states and release policy are MAKNOW DESIGN.