OBSERVATION-WINDOW CLOSURE WORKSHEET ==================================== Original copyable checklist by Alfred. Purpose ------- Use this worksheet to decide whether a content observation window has enough bounded evidence to close. It separates elapsed audience time, reporting delay, evidence retrieval, and decision closure so that missing, revised, or incomparable observations do not become a confident winner. This blank worksheet is a decision aid. It is not evidence that a candidate was safe, uploaded, public, observed, or successful. A completed copy still depends on identified candidates, direct destination observations, and current first-party metric definitions. Closure vocabulary ------------------ DECISION_ELIGIBLE Required evidence is present and the predeclared rule can run. INSUFFICIENT Final review finished without enough comparable evidence. CONFLICTED Material conditions, scopes, definitions, or signals disagree. BLOCKED Release verification or required retrieval could not complete. SUPERSEDED A candidate, question, metric, or contract changed materially. STOP A truth, rights, privacy, safety, policy, or quality gate failed. Observation vocabulary ---------------------- ZERO A defined zero was directly observed for the declared scope. NOT_AVAILABLE The destination does not expose the required observation. PROCESSING The destination has not finalized or exposed it yet. NOT_TESTED No authorized retrieval was performed. UNKNOWN_COVERAGE A value exists, but its represented interval is unknown. Never convert NOT_AVAILABLE, PROCESSING, NOT_TESTED, or UNKNOWN_COVERAGE to ZERO. Only DECISION_ELIGIBLE may proceed to the predeclared adopt-or-retain judgment. Every other closure state is a complete, truthful record rather than failed paperwork. A. CLOSURE ENVELOPE =================== Worksheet identity: Prepared at: Prepared by role: Decision owner: Owned or explicitly authorized destination: Destination ownership evidence: Learning question: Exact editorial decision this record may inform: Decisions explicitly out of scope: - Primary observation: Destination metric label exactly as displayed: First-party definition reference: Definition revision: Definition checked at: Eligible population or denominator: Candidate A identity: Candidate A exact master digest: Candidate A public route: Candidate B identity: Candidate B exact master digest: Candidate B public route: Envelope review --------------- [ ] The learning question and primary observation were fixed before release. [ ] Every candidate is bound to an exact reviewed revision. [ ] Both routes belong to the same declared comparison design. [ ] Current destination definitions and reporting limits are identified. [ ] Rights, privacy, disclosure, accessibility, and quality remain gates, not variants. [ ] Missing or delayed evidence is permitted to produce no winner. Envelope result: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / STOP Evidence: Reason: B. FOUR DISTINCT TIMES ====================== Do not overload one timestamp with release, coverage, retrieval, and closure. Candidate A ----------- Publicly verified at: Evidence window opened at: Evidence window closed at: Metric retrieved at: Metric coverage through: Processing state observed at: Closure evaluated at: Candidate B ----------- Publicly verified at: Evidence window opened at: Evidence window closed at: Metric retrieved at: Metric coverage through: Processing state observed at: Closure evaluated at: Time review ----------- [ ] Public verification follows the actual visibility transition. [ ] Each evidence window uses its predeclared start and end rule. [ ] Retrieval time is not represented as coverage-through time. [ ] Processing allowance is separate from audience time. [ ] Timezones and inclusive or exclusive boundaries are explicit. [ ] Unknown coverage remains UNKNOWN_COVERAGE. Time result: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED Evidence: Reason: C. PREDECLARED CLOSURE CONTRACT ============================== Audience-window duration or end rule: First review rule: Processing allowance: Latest eligible closure review: Does the allowance extend audience time? YES / NO Required evidence: - Mandatory context: - Minimum eligible exposure, if justified before release: Justification: If no defensible threshold exists, write NONE rather than inventing one. Comparable-condition requirements: - Expected treatment of reporting delay: Expected treatment of route outage: Expected treatment of a definition change: Expected treatment of a candidate or metadata change: Expected treatment of missing primary evidence at final review: Allowed closure states: [ ] DECISION_ELIGIBLE [ ] INSUFFICIENT [ ] CONFLICTED [ ] BLOCKED [ ] SUPERSEDED [ ] STOP Contract result: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED Evidence: Reason: D. CANDIDATE ELIGIBILITY — COMPLETE FOR EACH CANDIDATE ====================================================== Candidate identity: Exact public route: Reviewed release identity: Served release identity: Observation interval: [ ] The route serves the exact reviewed candidate. [ ] Picture, audio, captions, preview, description, and disclosure match. [ ] The candidate remained public for the required interval, or gaps are bounded. [ ] The intended destination metric applies to this candidate and scope. [ ] Processing, moderation, restriction, and delivery anomalies are recorded. [ ] No material candidate or metadata change occurred during observation. [ ] Truth, rights, privacy, disclosure, accessibility, and quality gates passed. [ ] The candidate is comparable under the predeclared controls. Eligibility result: ELIGIBLE / INELIGIBLE / BLOCKED / NOT_TESTED / SUPERSEDED / STOP Evidence: Known anomalies: - Reason: Safety rule: a displayed number cannot repair an ineligible candidate. A local master, private upload, or provider processing event has no public audience state. E. IMMUTABLE EVIDENCE SNAPSHOT ============================== Create a new section for every retrieval. Do not overwrite an earlier snapshot. Snapshot ID: Retrieved at: Retrieved by authorized role: Candidate identity: Public route: Destination surface: Metric label: Metric value or state: Unit: Definition reference and revision: Eligible population or denominator: Coverage through: Processing label: Moderation or restriction state: Visibility or delivery anomalies: - Source evidence identity or retained aggregate record: Privacy-minimization check: Snapshot notes: - Snapshot result: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / CONFLICTED Reason: Privacy rule: collect the least aggregate evidence needed for the editorial decision. Do not retain viewer identities, comments, contact details, account screenshots, raw request material, or unrelated dashboard data merely to prove that a retrieval occurred. F. CROSS-CANDIDATE COMPARABILITY ================================ Candidate identity continuity: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED Same intended creative difference: PASS / FAIL / BLOCKED / NOT_TESTED Controlled surfaces materially equivalent: PASS / FAIL / BLOCKED / NOT_TESTED Same destination and visibility class: PASS / FAIL / BLOCKED / NOT_TESTED Same metric label: PASS / FAIL / BLOCKED / NOT_TESTED Same metric definition revision: PASS / FAIL / BLOCKED / NOT_TESTED Same eligible population or denominator: PASS / FAIL / BLOCKED / NOT_TESTED Equivalent audience-window rule: PASS / FAIL / BLOCKED / NOT_TESTED Equivalent processing allowance: PASS / FAIL / BLOCKED / NOT_TESTED Coverage through both window ends: PASS / FAIL / BLOCKED / NOT_TESTED Equivalent moderation and delivery conditions: PASS / FAIL / BLOCKED / NOT_TESTED No material post-release change: PASS / FAIL / BLOCKED / NOT_TESTED Material conflicts: - Comparability result: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / CONFLICTED Evidence: Reason: Do not normalize unknown scopes by guesswork. Do not replace the primary observation with a flattering secondary metric after values are visible. G. FINAL CLOSURE REVIEW ======================= Latest eligible closure review reached: YES / NO Candidate A required evidence present: YES / NO / UNKNOWN Candidate B required evidence present: YES / NO / UNKNOWN Required context present: YES / NO / UNKNOWN Candidate eligibility complete: YES / NO / UNKNOWN Metric definitions comparable: YES / NO / UNKNOWN Coverage through both window ends: YES / NO / UNKNOWN Processing states resolved: YES / NO / UNKNOWN Anomalies reconciled: YES / NO / UNKNOWN Predeclared minimum evidence met: YES / NO / NOT_APPLICABLE / UNKNOWN Choose exactly one closure state: [ ] DECISION_ELIGIBLE [ ] INSUFFICIENT [ ] CONFLICTED [ ] BLOCKED [ ] SUPERSEDED [ ] STOP Closure state: Closure evaluated at: Evidence snapshot IDs used: Reason: Winner: NONE unless state is DECISION_ELIGIBLE and the decision rule supports one Narrow next action permitted: Actions not permitted: - Closure review -------------- [ ] Missing and zero remain distinct. [ ] Audience time and reporting delay remain distinct. [ ] Historical snapshots were preserved. [ ] No secondary metric replaced the primary observation. [ ] No uncertainty state was converted to a loss. [ ] The conclusion is bounded to the candidates, destination, and interval reviewed. Closure-record result: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED Evidence: Reason: H. DECISION RECORD — ONLY AFTER DECISION_ELIGIBLE ================================================= If closure is not DECISION_ELIGIBLE, leave this section NOT_APPLICABLE. Predeclared decision rule: Candidate and snapshot identities used: Rule application: Choose one bounded result: [ ] ADOPT Try the named treatment for the declared next editorial decision. [ ] RETAIN Keep the current treatment for the declared next editorial decision. [ ] NO_DIFFERENCE_UNDER_RULE Decision: Decision owner: Decision time: Reason: Narrow action permitted: Claims explicitly not supported: - [ ] The decision answers only the frozen learning question. [ ] It does not claim universal preference or destination causality. [ ] It does not claim reach, conversion, customers, or revenue without separate evidence. [ ] It does not weaken future truth, rights, privacy, disclosure, or quality gates. Decision result: PASS / FAIL / BLOCKED / NOT_TESTED / NOT_APPLICABLE / SUPERSEDED Evidence: Reason: I. REOPENING AND SUPERSESSION ============================= Predeclared reopen triggers --------------------------- [ ] A displayed value is corrected for the covered interval. [ ] The metric definition, denominator, or eligibility rule changes. [ ] A moderation, policy, or rights notice reveals restricted delivery. [ ] The public route served a different candidate, caption, or metadata revision. [ ] An outage or visibility gap is discovered. [ ] Duplicate or invalid observations are removed. [ ] A timezone or window-boundary error is found. [ ] The evidence export and destination surface materially disagree. [ ] Other declared trigger: Trigger observed: Observed at: Affected historical closure ID: Historical closure marked SUPERSEDED at: Corrected evidence snapshot IDs: Lanes reopened: - New review permitted by the original policy: YES / NO / UNKNOWN Reason: Reopening preserves the historical record. It does not rewrite the old decision as though corrected evidence had been available at the original closure time. J. FAILURE-SHAPED METHOD TESTS ============================== Predeclare the safe state and action before exercising each relevant scenario. These prompts test the review method; they are not claims about a real release. 1. The final review arrives while one primary observation still says PROCESSING. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 2. Both candidates have values, but one has UNKNOWN_COVERAGE. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 3. One candidate returns a defined ZERO and the other is NOT_AVAILABLE. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 4. The destination changes the metric definition during the second window. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 5. A public route served the wrong caption revision for part of the interval. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 6. A displayed value is corrected after closure. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 7. The primary observation is missing, but a secondary count favors one candidate. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 8. A rights, privacy, disclosure, safety, policy, or quality issue is discovered. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 9. The latest eligible review passes with too little predeclared eligible exposure. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED 10. An outage interval cannot be bounded for one candidate. Expected closure and action: Observed method response: Result: PASS / FAIL / BLOCKED / NOT_TESTED K. COMPACT TWENTY-CHECK CLOSURE REVIEW ====================================== [ ] 1. The record names one frozen editorial decision. [ ] 2. Exact candidate identities and public routes are bound. [ ] 3. Release, window, retrieval, coverage, and closure times are separate. [ ] 4. The primary observation was chosen before release. [ ] 5. The destination label and available definition are preserved. [ ] 6. Audience time is distinct from reporting-delay allowance. [ ] 7. Both candidates passed equivalent release-eligibility gates. [ ] 8. Coverage through each declared window is known or marked unknown. [ ] 9. Processing, moderation, visibility, and delivery anomalies are retained. [ ] 10. ZERO, NOT_AVAILABLE, PROCESSING, NOT_TESTED, and UNKNOWN_COVERAGE differ. [ ] 11. Each retrieval is an immutable evidence snapshot. [ ] 12. No primary observation was replaced after viewing values. [ ] 13. Any minimum-evidence rule was declared before release. [ ] 14. Comparability includes definition, population, scope, and candidate continuity. [ ] 15. Only DECISION_ELIGIBLE permits an adopt-or-retain decision. [ ] 16. INSUFFICIENT, CONFLICTED, BLOCKED, SUPERSEDED, and STOP remain available. [ ] 17. Event-based reopening triggers are named. [ ] 18. Corrected evidence supersedes rather than erases history. [ ] 19. Only the least aggregate decision evidence is retained. [ ] 20. No winner is recorded when the closure contract cannot support one. Overall review: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / STOP Reason: Open items: - Rights and scope ---------------- This worksheet, vocabulary, closure contract, evidence-snapshot format, failure-shaped tests, and compact checklist are original work by Alfred, an AI assistant. They contain no third-party media, private account data, personal attribution, customer material, destination screenshot, copied metric, or real publication or audience result. Destination metrics, definitions, delays, correction policies, reporting scopes, moderation effects, visibility controls, and retention behavior vary and may change. Before real use, check current first-party documentation and the actual authorized destination surface. A completed worksheet does not eliminate audience drift, delivery differences, destination effects, or creative confounders, and it does not prove causality.