Cadence Envelope and Less-or-None Recommendation

1. The real problem: the calendar must not decide instead of evidence

A content team promises five outputs every week. This week only two ideas have clear audience need, ready evidence, and meaningful unique value; one idea substantially duplicates an existing article, and two slots exist only because the calendar is empty. Treating “there is a slot” as “we must publish” converts schedule pressure into low-value output. This lesson builds Cadence Envelope & Publish-Defer Decision Matrix v1 so frequency becomes an outcome of quality and need rather than a fixed target.

2. Lesson boundary and owned artifact

The owned artifact is an auditable matrix that records at least audience_need, evidence_readiness, unique_value, redundancy, editorial_capacity, freshness_need, risk, provenance_ref, and review_date for each slot or candidate, then issues PUBLISH, REDUCE, DEFER, or NONE. PLN-010 owns Content Gap Detection and PLN-012 owns Audience Drift; this lesson does not absorb either outcome.

3. What a Cadence Envelope is

A cadence envelope is not a magic number such as “three posts per week.” It is a bounded operating range in which publication is allowed only when evidence, user value, and capacity jointly meet minimum requirements. One week may support three outputs, another one, and another zero. Exact bands are portfolio-specific; this lesson does not invent a universal threshold.

4. Evidence readiness comes before calendar priority

evidence_readiness closes before calendar pressure. A high-priority slot cannot manufacture missing sources, examples, review, or data. In the deterministic fixture, removing evidence readiness while leaving calendar priority high must produce REVIEW_REQUIRED or DEFER, never an optimistic publish decision.

5. Google's freshness boundary

Current Google Search guidance says adding large amounts of content mainly because you believe it will make a site seem “fresh” is not a recommended people-first strategy [GOOGLE-HELPFUL]. That guidance does not prescribe a universal publishing cadence. It provides a narrower boundary: cosmetic freshness alone is not sufficient reason to produce more content.

6. Scaled content abuse is a quality boundary, not a calendar model

Current Google Search spam policies define scaled content abuse as generating many pages primarily to manipulate rankings rather than help users [GOOGLE-SPAM]. The policy is not a complete editorial planning model, but it reinforces a critical guardrail: more volume without added user value is not evidence of a healthier content system.

7. Separate audience need from unique value

audience_need asks whether a current problem deserves a response; unique_value asks whether the proposed item adds something materially useful beyond existing coverage. High need with low unique value still does not authorize a new piece. If an article merely retitles existing coverage, high redundancy and low unique_value should produce NONE.

8. The redundancy gate prevents slot filling

An empty slot is not evidence. The matrix checks existing canonical coverage, overlap, and whether a candidate adds a new decision-useful contribution. If it substantially repeats an existing answer, the result is NONE for a new piece; creating a new URL only to satisfy quota is rejected by the contract.

9. Capacity is a real guardrail

editorial_capacity covers research, authoring, review, QA, and update budget—not only writer count. If the quota is five but reviewers can reliably close two items at the required quality, cadence must REDUCE. Keeping the quota fixed while reviewer capacity falls is an explicit failure probe in this lesson.

10. Freshness need is different from artificial freshness

Some content is genuinely time-sensitive: release notes, policy changes, or data updates. freshness_need exists for those cases. Date changes, superficial rewrites, or near-duplicate pages created only to look current are not valid publish triggers. Freshness must trace to an actual change or user need.

11. Four primary states: PUBLISH / REDUCE / DEFER / NONE

PUBLISH means the candidate passes value, evidence, capacity, and risk gates. REDUCE means useful demand exists but current capacity cannot support the requested quota. DEFER means value may be real but evidence, review, freshness window, or risk is unresolved. NONE means a new output does not currently add enough value or is redundant. None is a valid planning outcome, not planner failure.

12. Preserve uncertainty and non-success states

The broader contract preserves UNKNOWN, INSUFFICIENT_DATA, REVIEW_REQUIRED, REJECTED, or DO_NOT_USE where needed. Missing provenance makes a decision non-auditable. Unknown evidence readiness cannot be coerced into PUBLISH. A clean interface must not hide real uncertainty just to keep the calendar full.

13. Worked case: five-slot quota, only two valid candidates

The team has five slots. Candidates A and B have strong audience need, ready evidence, high unique value, and low risk, so both are PUBLISH. Candidate C duplicates existing coverage and receives NONE. Candidate D lacks complete evidence and becomes DEFER. Candidate E exists only because a slot is empty and lacks sufficient need/value, so it receives NONE. This week's envelope is two outputs, not five.

14. Deterministic negative cases and failure probes

Four shortcuts are rejected: 1) force every empty slot to PUBLISH → fail value/redundancy gates; 2) duplicate an existing article under a new title → NONE; 3) reduce reviewer capacity while keeping the same quota → REDUCE; 4) remove evidence readiness while keeping calendar priority high → REVIEW_REQUIRED or DEFER. Missing provenance is REJECTED, and stale evidence must carry a finite review date.

15. Transfer: irregular release notes

Transfer the same contract to release notes, where important updates may arrive twice in one week and then not at all for several weeks. Recompute the envelope from current evidence. A week with no meaningful update can validly return NONE; do not invent filler notes to satisfy a weekly quota. Transfer succeeds when the contract stays stable while frequency is recomputed from the new context.