Did the next step get easier?

Count: Commitments accepted by original deadline ÷ all authorized commitments originally due; retain late, blocked and canceled cases with reasons.

When: Daily 5 p.m.; weekly trend

If it needs attention: Recover overdue work and correct weak prioritization; distinguish revised-plan performance.

Proposed target: Establish a complete first-week baseline before setting a numeric reliability target.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: New authorized assignments with owner, actual executor, acceptance criteria, due time and return location ÷ all new authorized assignments.

When: Before dispatch

If it needs attention: Return incomplete briefs for correction.

Proposed target: 100% of new assignments meet the brief standard.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Learning questions due that receive a source-backed continue/change/stop decision ÷ all learning questions due.

When: Weekly

If it needs attention: Require the evidence and next experiment; do not count a discussion as a decision.

Proposed target: 100% of due learning questions have a documented disposition; effect targets follow baseline.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Reserved decision requests with clear recommendation, alternatives, cost/impact, evidence and deadline ÷ all reserved decision requests presented.

When: At escalation; weekly

If it needs attention: Resolve missing analysis before spending founder attention.

Proposed target: 100% of reserved decisions presented are decision-ready.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Due scheduled runs with a completed execution or explicit failure receipt by the next checkpoint ÷ all due scheduled runs.

When: Each checkpoint; daily

If it needs attention: Investigate missing execution and record recovery; configuration does not count.

Proposed target: 100% of due runs receive a success or failure receipt.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Overdue commitments with reason, recovery owner, next action and next checkpoint ÷ all overdue commitments.

When: Daily close

If it needs attention: Assign recovery or obtain an explicit cancellation decision.

Proposed target: 100% of overdue commitments have a current recovery record.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Age of each open authorized task; median and oldest by work type, with blocked/waiting/executing states.

When: Daily

If it needs attention: Unblock, reduce scope, or stop work that cannot progress.

Proposed target: Set aging thresholds after seven days of observed work; no invented historical target.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed time from detected execution failure to verified restoration of the promised workflow, by incident type.

When: At incident close; weekly

If it needs attention: Repair the recurrent cause and test the recovery path.

Proposed target: Baseline first; no unobserved recovery-time claim.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Due scheduled obligations whose required result was accepted by the original deadline / obligations due; no-work checks count only when explicitly intended.

When: After each due run; daily close.

If it needs attention: Classify misses as not-started, blocked, failed, returned-unaccepted or late; repair the cause.

Proposed target: Every due obligation has a disposition; establish first-week reliability before setting a rate target.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Handoffs due for response with an actual response and documented acceptance/disposition / handoffs due for response.

When: At response checkpoint; daily reconciliation.

If it needs attention: Read the destination before retry and surface unacknowledged or unanswered work.

Proposed target: No due handoff disappears between apps; absence remains explicit.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Unintended repeated state-changing actions for the same request/version/recipient/action key; list uncertain attempts separately.

When: At dispatch reconciliation.

If it needs attention: Stop repeat dispatch, inspect receipts and repair idempotency or lease recovery.

Proposed target: Zero unintended duplicates; uncertain delivery is checked before retry.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Attributable provider usage and actual incremental charges across attempts / accepted comparable automation outcomes, with receipt coverage.

When: At run close; weekly by workflow version.

If it needs attention: Reduce repeated context, empty model runs and unstable steps; retain necessary exception checks.

Proposed target: Baseline each workflow version; account capacity percentages never become task dollars.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Observed intended-user attempts that reach the specified supported next step / all observed attempts; report assisted and unassisted counts separately.

When: After each consented observation; weekly learning review.

If it needs attention: Fix the most consequential observed barrier and reobserve the same task class.

Proposed target: Establish a labeled sample first; no simulated sessions counted as human observations.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Observed participants accurately explaining next action, eligibility uncertainty and service limits / participants asked the same comprehension prompts.

When: At design validation for material journey changes.

If it needs attention: Revise the specific misleading label or layout, then repeat the prompt.

Proposed target: No unresolved critical misunderstanding in the release evidence; sample limits remain visible.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Open reproducible barriers preventing a required journey, grouped by severity, language and device; age since confirmation.

When: At review; check due repairs daily.

If it needs attention: Assign the earliest useful repair and show unresolved scope before release.

Proposed target: Every critical barrier has an owner and checkpoint; release treatment explicitly reviewed.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed time from bounded design brief to accepted design, with iterations and waiting recorded; compare similar changes only.

When: At design acceptance; weekly.

If it needs attention: Stop low-value polishing once criteria pass; investigate repeated review cycles.

Proposed target: Baseline first; reduce avoidable iteration while preserving observed usefulness.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Required journey/criterion cases with a dated Pass/Partial/Fail/Not Tested result and evidence / all required cases; show each status count.

When: Before release; after affected changes.

If it needs attention: Fill missing evidence and stop broad claims that exceed reviewed scope.

Proposed target: Every required case is explicitly classified; coverage is not pass rate or blanket conformance.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Confirmed critical usability/accessibility barriers first found after release, grouped by journey and affected input/device mode.

When: At confirmation; weekly learning review.

If it needs attention: Repair the barrier and add the smallest meaningful regression scenario.

Proposed target: No known unresolved critical barrier silently accepted for release.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed time from repair ready-for-review to dated verification result; include reopening and access waits.

When: At each repair verification.

If it needs attention: Remove review bottlenecks and return precise reproduction when a repair fails.

Proposed target: Establish the baseline; meet the stated review checkpoint or log the miss.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Runs justified by changed scope, a prior failure or a required release gate / total test runs; report runtime and attributable usage alongside scope.

When: At test run; daily close.

If it needs attention: Remove redundant reruns and retain the evidence that makes future reruns unnecessary.

Proposed target: Every repeated run has a reason; do not reward test volume.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Required English/Spanish task pairs passing action, safety meaning, date, form and recovery checks / required pairs on the same revision; state reviewer type.

When: After material copy or journey changes.

If it needs attention: Fix the specific parity gap and rerun the paired task.

Proposed target: Every changed critical pair is dispositioned; AI-only review remains labeled.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Observed Spanish-speaking intended users reaching the supported next step / observed Spanish task attempts; assisted/unassisted and sample size separate.

When: After consented observations; weekly learning review.

If it needs attention: Trace language or navigation barriers and compare only equivalent tasks.

Proposed target: Establish real observations first; no human success baseline inferred from translation checks.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed time from confirmed consequential language error to reviewed correction, with reviewer qualification and blocked time.

When: At correction acceptance.

If it needs attention: Prioritize errors affecting eligibility, safety, deadlines or action; retain remaining review limits.

Proposed target: Meet the stated repair checkpoint or record a miss; no unsupported blanket parity claim.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Changed high-risk strings/task steps with a current review result / changed high-risk scope; report attributable review usage and actual cost separately.

When: At localization review.

If it needs attention: Review the affected journey and reuse valid unchanged terminology evidence.

Proposed target: All changed high-risk scope classified; compare effort only across equivalent review scope.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Counts and known cents of in-scope transactions at order, invoice, payment, consumption and allocation stages; stages overlap and are not additive.

When: Daily exceptions; weekly

If it needs attention: Obtain the specific missing evidence and reconcile transaction identity.

Proposed target: 100% of inventory lines have a stated evidence stage; completeness percentage waits for a scoped inventory.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Report separately: available operating cash ÷ defensible daily cash operating cost; next uncovered dated payment; and actual cash receipts/payments minus the prior same-week forecast. A normalized tool-cost scenario is separately labeled.

When: Cash exceptions daily; roll the next 13 weeks forward weekly; refresh a 12-month view monthly.

If it needs attention: Resolve the earliest material payment gap, missing source or forecast error before committing to additional scope.

Proposed target: First reconcile available cash, payment dates and company share. Keep personal funding scenarios in the private Founder review; reserve thresholds are proposed until adopted.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Documentation coverage = reconciled in-scope cost cents with an explicit company-attributed, personal or shared-unallocated state / all reconciled in-scope cost cents in the same period and currency. Report allocation resolved separately = cost cents with a completed, supported company/personal assignment / that same denominator. Shared-unallocated cents count only toward documentation coverage, not resolved allocation; show their amount.

When: Weekly; proposed October 23 checkpoint

If it needs attention: Keep documentation coverage and resolved allocation visibly separate. Resolve shared-unallocated amounts only with a supported method, and preserve the remainder as unallocated. A 100% documented state is not 100% company-cost attribution.

Proposed target: 100% of reconciled costs have an explicit state. No target company-share percentage; resolved allocation remains unknown until evidenced. Empty denominators mean N/A.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: All attributable consumed cents for successful, retried, failed and abandoned effort in a closed comparable cohort ÷ accepted distinct units in that cohort.

When: Weekly when cost coverage exists

If it needs attention: Compare workflows alongside quality, speed and missing coverage.

Proposed target: Baseline first; no cents target while actual consumed costs or allocation are unestablished.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Displayed metrics with versioned definition, scope, denominator, source, observed time, cadence and decision rule ÷ all displayed metrics.

When: Before release; weekly

If it needs attention: Return unsupported rates to counts/unknown and repair metadata.

Proposed target: 100% of displayed metrics have a complete contract.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Sampled published readings reproduced from retained sources within stated rounding ÷ all sampled readings; unresolved cases visible.

When: At release; weekly sample

If it needs attention: Correct arithmetic or lineage and append a correction receipt.

Proposed target: 100% reproducibility in each declared verification sample.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Completed research briefs with an explicit recommendation, uncertainty and next test or documented no-action decision ÷ completed research briefs.

When: At acceptance; weekly

If it needs attention: Stop research that cannot change a decision or identify a needed evidence gap.

Proposed target: 100% of accepted research briefs state their decision relevance.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Externally benchmarked claims supported by primary evidence and an explicit comparability limit ÷ benchmarked claims used in decisions.

When: At review

If it needs attention: Remove unsupported causal or peer-success claims.

Proposed target: 100% of benchmark claims used have source and transfer limits.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Active revenue hypotheses with payer, problem, value, price logic, test, stop rule and trust review ÷ active hypotheses.

When: Before external testing; weekly

If it needs attention: Clarify or stop hypotheses that cannot be tested responsibly.

Proposed target: 100% of new hypotheses have a complete test contract.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Authorized real discovery attempts with a dated outcome and documented implication ÷ discovery attempts due; no-response remains a result.

When: After interaction; weekly

If it needs attention: Change the hypothesis or evidence plan based on actual responses.

Proposed target: Baseline first; no invented customer or meeting target.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Open opportunities with verified stage, source, owner and dated next action ÷ all open opportunities.

When: Weekly

If it needs attention: Remove unsupported stages and close abandoned opportunities.

Proposed target: 100% of recorded opportunities have evidence and next action; no opportunities means N/A.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Actual attributable test cost in cents for all successful, failed, retried and abandoned attempts in a closed, comparable experiment cohort / distinct predeclared experiment IDs that reached their agreed evidence threshold and a documented continue/change/stop decision in that cohort. Include all cohort attempt costs even when an experiment produces no accepted decision. Report raw cost, decision count and missing cost coverage.

When: At experiment close

If it needs attention: Define the hypothesis, decision, evidence threshold, cohort and experiment ID before execution. Do not split one experiment into extra learning units after seeing results. Compare only like experiment types and inspect decision quality alongside cost.

Proposed target: Baseline before setting a cost target. An unsuccessful hypothesis can yield an accepted decision; an uninformative attempt cannot. Zero accepted decision experiments means report cost and zero results, with no ratio. Forecasts and revenue remain separate.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Distinct in-scope transactions matched to available order/invoice/payment evidence with duplicates resolved ÷ known in-scope transactions; inventory status shown.

When: Daily exceptions; weekly close

If it needs attention: Retrieve missing evidence and retain unresolved states.

Proposed target: 100% of new records have identity, period and evidence status; full reconciliation baseline open.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Age and known cents of unresolved reconciliation exceptions, grouped by missing invoice, settlement, refund, consumption or allocation.

When: Weekly

If it needs attention: Escalate old or material gaps with a named owner and due date.

Proposed target: Agree materiality and aging thresholds after inventory review.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Renewals due within the declared lookahead with owner, expected amount, date, service need and source verified ÷ renewals due in that lookahead.

When: Weekly; proposed 30-day lookahead

If it needs attention: Obtain current terms and a continue/cancel decision before renewal.

Proposed target: 100% of known renewals in lookahead have an explicit disposition before their decision deadline.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Opening credits + top-ups − consumption ± adjustments − closing credits, in the provider's native unit and matching period.

When: On statement; monthly

If it needs attention: Investigate a nonzero difference; do not mix credit units and dollars without a supported conversion.

Proposed target: Zero unexplained difference where complete evidence exists; otherwise unresolved, not zero.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Deployments requiring immediate rollback or hotfix / production deployments in the same service and period.

When: At each release; weekly service review.

If it needs attention: Inspect the failure cause and narrow the next change; preserve affected releases in the denominator.

Proposed target: Establish a service baseline; any critical user-path failure triggers immediate review.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Median and longest elapsed time from first included code commit to verified production deployment, with sample count.

When: At verified deployment; weekly trend.

If it needs attention: Separate review, access and release waiting; remove the largest repeated delay.

Proposed target: Baseline comparable changes first; improve delay without weakening acceptance.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Predeclared critical journeys passing behavioral and recovery criteria on the named revision / all critical journeys required for that release.

When: Before release and after material journey change.

If it needs attention: Block release for an unresolved critical failure; list partial and untested scope.

Proposed target: Every required critical journey has a documented disposition before release; passing scope is explicit.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Full, partial and missing execution receipts / engineering executions for accepted and unsuccessful work, reported separately.

When: At task close; daily reconciliation.

If it needs attention: Fix missing capture before comparing compute per accepted change; include rework and failed attempts.

Proposed target: No silent attribution gaps; only compare cost or tokens within like work types.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Predeclared critical task/device cases passing state, interaction and recovery checks / cases required for the change; list untested cases.

When: On changed UI before acceptance.

If it needs attention: Repair the blocking behavior and retain a reproducible browser receipt.

Proposed target: Every required critical case is dispositioned; no screenshot-only acceptance.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Confirmed frontend regressions first discovered after release, grouped by severity and released change.

When: At issue confirmation; weekly.

If it needs attention: Add a targeted regression check for the underlying behavior.

Proposed target: Baseline first; any critical escaped regression receives immediate repair triage.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed time from confirmed reproduction to accepted repair, showing blocked time and production verification separately.

When: At repair acceptance.

If it needs attention: Remove recurring reproduction or review delays; do not hide reopened defects.

Proposed target: Improve the comparable-case baseline while preserving verification.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Provider input/cached/output receipts and actual incremental charges for bounded UI tasks, including retries; compare only like tasks.

When: At task close.

If it needs attention: Reduce repeated whole-page context and avoid unrelated refactoring when the task grows.

Proposed target: Every task declares full, partial or missing usage; no persona token allocation.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Required transition cases passing ownership, revision, idempotency and recovery assertions / predefined transition cases.

When: On queue/integration changes before acceptance.

If it needs attention: Repair the broken contract and retest the whole affected transition.

Proposed target: All critical transitions are evidenced; no accepted duplicate mutation.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Unexpected server or contract failures / valid requests in the measured API scope; separately report expected validation/auth refusals.

When: At observed failure; daily scoped review.

If it needs attention: Trace the request and repair the failure without weakening correct refusal behavior.

Proposed target: Baseline measured traffic; a safe refusal is not counted as a defect.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Uncertain state-changing attempts reconciled to a verified destination result / uncertain attempts due for reconciliation.

When: Immediately after uncertain writes; next documented checkpoint.

If it needs attention: Read receipts, preserve request identity and escalate unresolved ambiguity before retry.

Proposed target: Every uncertain write is resolved or explicitly blocked; no silent duplicate dispatch.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Attributable executor usage and actual incremental service charges per accepted comparable backend change, with failed attempts included.

When: At accepted change and reconciliation.

If it needs attention: Prefer the existing contract when it meets the requirement; explain new infrastructure cost.

Proposed target: Known coverage before efficiency claims; a new cost requires an explicit decision.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Reviewed query cases with at least one supported relevant result among the first three / reviewed cases with a known appropriate option; report no-known-option cases separately.

When: At ranking change; refresh benchmark with observed failures.

If it needs attention: Inspect missed cases by need and revise retrieval or data only with supporting evidence.

Proposed target: Baseline the approved test set; no aggregate claim without case coverage.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Confirmed unsupported eligibility, availability or contact claims in reviewed results, grouped by severity and source.

When: At review or source correction.

If it needs attention: Remove or qualify the claim and preserve the source conflict until resolved.

Proposed target: Zero known critical unsupported assertions accepted into changed routing.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Records due for review with a current source-backed disposition / records due under the defined freshness policy.

When: At source update; daily due queue.

If it needs attention: Prioritize high-impact stale records; unknown or failed checks remain dated.

Proposed target: Every due record has a disposition; source-read is not phone-confirmed.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Attributable usage and actual cost per evaluated case on a named benchmark, with cached input and reruns separated.

When: At benchmark run.

If it needs attention: Reuse stable fixtures and run only justified affected scope while preserving regression coverage.

Proposed target: Baseline like benchmarks; no repeated run solely to create fresh numbers.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Deployments with successful provider status, expected served artifact and critical-route readback / deployments attempted.

When: Every production deployment.

If it needs attention: Investigate divergence before marking published; keep failed attempts visible.

Proposed target: Every attempted deployment has a dated verified or failed disposition.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Due health-check windows with a timestamped result or access failure / windows due; separately report successful samples / samples attempted.

When: At scheduled check and missed-window catch-up.

If it needs attention: Repair missing checks or route failures; label sampled availability and coverage.

Proposed target: All due windows accounted for; no unsupported continuous-uptime claim.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed time from deployment-caused service impairment to verified restored service, with incident count and cause.

When: At deployment incident closure.

If it needs attention: Improve the demonstrated recovery step and capture the successful rollback or repair.

Proposed target: Baseline actual incidents; no incidents means not observed, not zero-minute recovery.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Vendor charge lines with service, period, receipt and actual/estimated allocation label / charge lines in declared infrastructure scope.

When: At invoice availability; monthly close and material service change.

If it needs attention: Resolve unexplained charges and estimate the effect of proposed infrastructure before buying.

Proposed target: Every known charge is reconciled or listed as open; shared provision is not actual company spend.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: In-scope systems due for review with recorded access/recovery disposition / in-scope systems due.

When: At scheduled review or access change.

If it needs attention: Assign missing review evidence and distinguish inaccessible systems from reviewed ones.

Proposed target: Every due system has a documented disposition; no inferred MFA or recovery status.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Open confirmed high-impact access/recovery exceptions, grouped by severity, age and next checkpoint.

When: At discovery; daily due exception review.

If it needs attention: Escalate the concrete decision and least-disruptive repair; do not silently broaden permissions.

Proposed target: Each material gap has an owner and review time; urgent gaps are acted on under existing authority.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Critical systems with an approved, dated recovery procedure and bounded verification evidence / critical systems in scope.

When: After recovery change; scheduled review.

If it needs attention: Prepare missing recovery documentation and identify the actual authorized holder.

Proposed target: No claim of recoverability without evidence; avoid disruptive account tests.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Elapsed blocked time caused by access configuration for already-authorized tasks, classified as necessary control or avoidable friction.

When: At blocker resolution; weekly review.

If it needs attention: Simplify the narrow access path when it preserves the required control.

Proposed target: Baseline first; reduce avoidable delay without weakening private-data boundaries.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material public claims with a matching current source and stated scope / all material claims reviewed in the same release

When: Each release; exceptions daily

If it needs attention: Correct, qualify, or remove unsupported material claims.

Proposed target: 100% before release; show the reviewed count and exclusions.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material corrections closed with a release receipt by their assigned deadline / all material corrections due in the period

When: Daily close

If it needs attention: Escalate late items and retain the original error and recovery time.

Proposed target: 100% of due corrections; severity-based deadlines must be assigned before work starts.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Audited evidence items retaining source type, date, direct/reported distinction and limitations / all audited evidence items

When: Weekly; after evidence-language changes

If it needs attention: Restore omitted human feedback or narrow an overstated claim.

Proposed target: 100% of audited items; report the sample, including disagreements.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Priority claims reviewed with a disposition by their deadline / all priority claims due in the period

When: Daily

If it needs attention: Reorder the queue and escalate the highest-consequence overdue claim.

Proposed target: 100% of due priority reviews; explain unresolved items instead of marking them verified.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Sampled research conclusions with an accessible source, source date, check method and bounded claim / all research conclusions sampled

When: At handoff; weekly sample

If it needs attention: Repair missing provenance before the conclusion is used publicly.

Proposed target: 100% of sampled conclusions; show sample size.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Closed research tasks resulting in an answer, bounded unknown, or next-owner handoff accepted by the reviewer / all research tasks submitted for closure

When: Weekly

If it needs attention: Stop circular searching and clarify the unanswered question.

Proposed target: 100% of closure requests have an accepted disposition; time targets follow a baseline by research type.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Documentation coverage = required field slots with an explicit evidence state, source or explicit unknown, check method/state and date the evidence state was recorded / all required field slots in the declared record cohort and field list. Report current support separately = field slots meeting the required evidence-method and freshness rules / that same denominator. Unknown, stale, contradictory and unchecked slots remain outside the current-support numerator. Elena's declared scope is the access fields of provider records reviewed in this cycle.

When: At each review; weekly

If it needs attention: Fill missing evidence-state documentation, then prioritize unknown or stale fields that affect the next step. Show the separate current-support count and preserve whether evidence is published-source review or direct provider confirmation. Keep the source observation date separate; recording an unknown state must not invent or refresh a source check.

Proposed target: 100% of required access-field slots have a documented evidence state. Current-support coverage is reported separately; no claim that explicit unknowns are verified. Empty cohorts mean N/A.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material provider contradictions assigned a precise question and resolution owner by the triage deadline / all material contradictions due for triage

When: Daily

If it needs attention: Prioritize an official-source recheck or an authorized contact request.

Proposed target: 100% by assigned triage deadline; external response time is reported separately.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Provider question briefs accepted as actionable without missing necessary context / all provider briefs submitted for review

When: At handoff; weekly

If it needs attention: Clarify the question and source conflict before requesting contact.

Proposed target: Baseline first five briefs; every rejected brief receives a specific repair.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Documentation coverage = required field slots with an explicit evidence state, source or explicit unknown, check method/state and date the evidence state was recorded / all required field slots in the declared record cohort and field list. Report current support separately = field slots meeting the required evidence-method and freshness rules / that same denominator. Unknown, stale, contradictory and unchecked slots remain outside the current-support numerator. Malik's declared scope is the required fields of active records in the named data audit.

When: At import; weekly audit

If it needs attention: Repair missing state/source/date fields, reconcile duplicate field identities and separately report unsupported or overdue evidence. Documentation completeness cannot establish the accuracy or availability of a service. Keep the source observation date separate; recording an unknown state must not invent or refresh a source check.

Proposed target: 100% evidence-state documentation in the declared audit. Show current-support coverage and unresolved field counts separately; publish the field list and cohort. Empty cohorts mean N/A.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Displayed counts exactly reproducible from the named canonical cohort and rules / all displayed counts audited

When: Every KPI release

If it needs attention: Fix the definition, duplicate or filter causing a disagreement.

Proposed target: 100% of audited counts; no mixing test and production cohorts.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Data-quality exceptions resolved with a change receipt by assigned due time / all data-quality exceptions due

When: Daily queue; weekly trend

If it needs attention: Escalate critical-field errors and document unresolved duplicates.

Proposed target: 100% of due exceptions; keep original and corrected values in history.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Consenting participants correctly distinguishing actual payment, estimated shared cost and available AI capacity / all participants completing the same comprehension check

When: Before finance-page release; after material wording changes

If it needs attention: Rewrite the label that produced a material wrong inference and recheck it.

Proposed target: Run an initial five-person exploratory check when participants are available; resolve every observed material misconception before claiming clarity.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Pages reviewed or corrected by their stated editorial deadline / all pages due for editorial review in the period

When: Daily Pulse check; weekly public-copy review

If it needs attention: Correct old publication states, dates and unsupported promises.

Proposed target: 100% of pages due; source unavailable remains visibly unresolved.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Reviewed financial and evidence labels that pass both editorial and factual review / all revised financial and evidence labels submitted

When: Every related release

If it needs attention: Restore necessary nuance or replace jargon before publication.

Proposed target: 100% of reviewed labels; reviewer acceptance must be recorded.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Feedback commitments fulfilled by their agreed deadline / all practitioner feedback commitments due in the period

When: Daily queue; weekly report

If it needs attention: Recover the missed commitment and tell the relationship owner what remains.

Proposed target: 100% of due commitments; contact requires the applicable authorization.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Completed feedback interactions producing a dated, specific workflow insight and proposed disposition / all completed feedback interactions reviewed

When: After each interaction; weekly synthesis

If it needs attention: Improve questions when the interaction yields only general praise or an unclear problem.

Proposed target: Establish a baseline from the first five available interactions; no universal rate target yet.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Consented observed practitioner attempts that identify a suitable supported next step / all observed attempts in the same study

When: Each study; weekly with sample size

If it needs attention: Fix the most common obstruction and repeat the relevant task.

Proposed target: Baseline first; report assistance, task, device and language before setting an improvement target.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Consented first-use attempts reaching the agreed next step / all first-use attempts observed in the same cohort

When: After each observation; weekly

If it needs attention: Remove the top onboarding barrier and report how much assistance was needed.

Proposed target: Establish a baseline; do not present assisted attempts as independent success.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Feedback items due for a response with a documented acknowledgment and disposition / all feedback items due for response

When: Daily queue; weekly

If it needs attention: Close the missing response through the authorized relationship owner.

Proposed target: 100% of due items; distinguish prepared responses from delivered responses.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Participating people whose stated language, accessibility and follow-up preferences were honored / participants who stated such preferences

When: Each session; weekly review

If it needs attention: Adapt the format or route unresolved access needs before another session.

Proposed target: 100% of stated preferences honored or an explicit agreed alternative; do not infer unstated needs.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material decisions with dated authority, scope, owner, next action and source in the canonical index ÷ all material decisions in the declared review scope.

When: After decision; daily

If it needs attention: Resolve conflicting instructions and preserve superseded versions.

Proposed target: 100% of new material decisions indexed with authority and action.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Active role assignments with current charter, actual runner, access scope, acceptance rules, handoff contract and named coverage ÷ all active role assignments.

When: Before activation; weekly

If it needs attention: Close capability or access gaps before assigning unattended responsibility.

Proposed target: 100% of newly activated assignments meet readiness rules; historical baseline open.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Retired assignments/integrations with open work reassigned, execution disabled where required, access disposition verified and evidence retained ÷ retirements due.

When: At retirement; weekly

If it needs attention: Remove stale automation/access and recover orphaned work.

Proposed target: 100% of retirements due have a verified disposition; no cases means N/A.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Active role/workflow types with a current representative capability check and documented limits ÷ role/workflow types in the declared review scope.

When: On material model/tool change; monthly

If it needs attention: Retrain the brief, narrow authority, or move work to a capable executor.

Proposed target: Define scope now; propose complete baseline coverage by October 23 without pretending it already exists.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material issues due with a dated decision/escalation, owner, source and next action ÷ material issues due.

When: On discovery; weekly

If it needs attention: Escalate unresolved reserved/legal questions to the founder or qualified adviser.

Proposed target: 100% of due material issues have an explicit disposition; this is not legal-compliance certification.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Controls in the declared review scope with evidence that actual behavior matches the current policy ÷ controls reviewed.

When: Before material release; periodic review

If it needs attention: Correct the implementation or revise the policy through the authorized owner.

Proposed target: Baseline by control family; all discrepancies receive an owner and due date.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Permission/retirement actions due with verified authorized disposition ÷ access/retirement actions due.

When: On change; weekly with Evelyn

If it needs attention: Remove unnecessary access and resolve ownership gaps after authorization.

Proposed target: 100% of due access actions have a verified disposition.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Known obligations due in the declared review period with source, responsible owner, original deadline and documented completion or decision evidence / all known obligations due in that same period. Obligations outside the due cohort are excluded from both terms; incomplete inventory is disclosed.

When: Before commitment; weekly

If it needs attention: Escalate unclear due obligations and distinguish completed, deferred, disputed and unresolved states. Do not count a documented decision as fulfillment of the obligation.

Proposed target: 100% of known due obligations have an owner and documented state; fulfillment is reported separately. No obligations due means N/A, not 100%.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Active product commitments with user problem, owner, original due time and acceptance criteria / active product commitments.

When: At assignment and scope change; daily.

If it needs attention: Clarify missing conditions before starting dependent implementation.

Proposed target: All active commitments are decision-ready; unknowns remain named rather than inferred.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Distinct product deliverables accepted by their original due time / distinct product deliverables due; preserve changed deadlines.

When: At acceptance; daily close.

If it needs attention: Identify whether scope, dependency or execution caused the miss and change the next plan.

Proposed target: Establish the first-week baseline; no backdated deadline repairs.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Experiments due for review with observed evidence and a keep/revise/stop decision / experiments due for review.

When: At experiment checkpoint; weekly.

If it needs attention: Close or reframe an experiment whose result cannot change a decision.

Proposed target: Every due experiment receives a documented decision or explicit evidence blocker.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Attributable actual costs and provider tokens across all attempts / accepted comparable experiments; show allocation and receipt coverage.

When: At experiment close; weekly by work type.

If it needs attention: Choose a smaller experiment when expected learning is equivalent; do not substitute estimates for charges.

Proposed target: Baseline with complete coverage before claiming an efficiency improvement.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Distinct planned publications verified live by original deadline ÷ all authorized publications due.

When: Per publication; daily

If it needs attention: Recover missed slots and fix readiness or publisher ownership.

Proposed target: Baseline first; every missed slot receives a recovery decision.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Distinct content items accepted at first substantive review against accuracy, accessibility and audience criteria ÷ items receiving first review.

When: At review; weekly

If it needs attention: Fix recurring source, alt-text or clarity issues in the content brief.

Proposed target: Baseline first; no publication acceptance without required quality checks.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Report two separate series for each named campaign and fixed window. Relevant-response yield = deduplicated relevant response events attributable to the campaign / verified campaign impressions for that same window. Qualified task-start rate = deduplicated eligible visit sessions with a predefined task-start event / all eligible attributed visit sessions for the same campaign and window. Never combine the numerators or substitute visits for impressions. Publish raw counts when matching attribution or denominator evidence is missing.

When: At campaign checkpoint

If it needs attention: Compare each series only with the same definition and attribution method. A response per impression is an event yield, not a person conversion rate; task starts do not establish completion or usefulness. Keep privacy-safe aggregate counts.

Proposed target: Establish a baseline separately for each series; predefine relevance, task-start, session and deduplication rules. Empty denominators mean N/A; no universal conversion target.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Poll/feedback follow-through items delivered by original promised date ÷ public follow-through items due.

When: At close date; daily queue

If it needs attention: Publish the decision, reason, owner and next action, including zero-response outcomes.

Proposed target: 100% of due public follow-through items have a disposition; delivery rate baseline open.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Recommendations with decision owner, alternatives, evidence, assumptions and revisit condition / all recommendations submitted for the same review

When: Each advisory packet

If it needs attention: Return incomplete advice for clarification before presenting it as decision-ready.

Proposed target: 100% of submitted recommendations meet the checklist.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material advisory conclusions linked to an exact source or explicitly labeled judgment / all material conclusions audited

When: Each packet; after source changes

If it needs attention: Repair citations and distinguish interpretation from observed fact.

Proposed target: 100% of audited conclusions; retain unresolved disagreement.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Recommendations with completed conflict/interest and authority-boundary review / all recommendations requiring review

When: Each packet

If it needs attention: Expose omitted influences and route reserved decisions to the founder.

Proposed target: 100% of relevant recommendations, with unknowns visible.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Field recommendations naming the task, evidence type, setting and limitation / all field recommendations reviewed

When: Each advisory packet

If it needs attention: Request context or narrow a claim drawn from a single experience.

Proposed target: 100% of reviewed recommendations; report evidence gaps explicitly.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Field recommendations with a feasible response and a defined observation that could test it / all field recommendations submitted

When: Each packet

If it needs attention: Replace general concern with an answerable question or bounded change.

Proposed target: 100% of submitted recommendations have a testable next step.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Field recommendations reviewing provider relationships, accessibility and competing user interests / all relevant field recommendations

When: Each packet

If it needs attention: Expose a conflict or overlooked access burden before the founder decides.

Proposed target: 100% of relevant recommendations; no assumed provider endorsement.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Technical recommendations with a versioned source, reproduction or acceptance test, and stated scope / all technical recommendations submitted

When: Each packet; after release changes

If it needs attention: Return vague error or reliability claims for a concrete scenario.

Proposed target: 100% of recommendations affecting release or instrumentation decisions.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Technology recommendations naming evidence that would reverse or narrow them / all technology recommendations reviewed

When: Each packet

If it needs attention: Identify the smallest useful experiment and stop open-ended architecture debate.

Proposed target: 100% of reviewed recommendations have a clear revisit condition.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Technology recommendations stating vendor/tool interest, integration burden, privacy and cost uncertainty / all recommendations proposing a tool or architecture change

When: Before a tool decision

If it needs attention: Expose maintenance cost and conflicts before treating a tool as an improvement.

Proposed target: 100% of relevant recommendations; prices must have a current source if quoted.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Numeric financial recommendations with a source, period, payer and actual/estimate/unknown label / all numeric recommendations audited

When: Every finance review

If it needs attention: Remove unsupported totals or narrow them to a planning estimate.

Proposed target: 100% of audited recommendations; no conversion of plan percentages into dollars.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Spending or revenue recommendations with a realistic alternative and sensitivity to stated unknowns / all spending or revenue recommendations submitted

When: Each financial decision packet

If it needs attention: Identify what changes under lower uptake, higher maintenance or missing allocation evidence.

Proposed target: 100% of submitted recommendations; no invented revenue or runway baseline.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Finance recommendations reviewing vendor interest, payer influence and mission/ranking conflicts / all relevant finance recommendations

When: Each packet; before purchase recommendation

If it needs attention: Expose the conflict and recommend a boundary the founder can decide.

Proposed target: 100% of relevant recommendations have an explicit disclosure state.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material privacy or public-interest recommendations naming the affected person, plausible harm, evidence and usable remedy / all such recommendations submitted

When: Each packet; after material product changes

If it needs attention: Replace broad concern with a specific risk and practical response.

Proposed target: 100% of material recommendations contain all four elements.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Privacy and disclosure conclusions linked to product evidence or a current primary source, with interpretation labeled / all conclusions audited

When: Each advisory review

If it needs attention: Correct stale or unsupported claims; route legal questions to qualified counsel when necessary.

Proposed target: 100% of audited conclusions; no legal assurance inferred from AI review.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Relevant recommendations with explicit payer/vendor conflicts, data exposure and AI-role disclosure analysis / all relevant recommendations

When: Before public transparency or monetization decisions

If it needs attention: Surface omitted conflicts or unnecessary public detail.

Proposed target: 100% of relevant recommendations have a recorded analysis and unresolved questions.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Growth recommendations naming an intended audience, meaningful action and evidence level / all growth recommendations submitted

When: Each campaign or launch review

If it needs attention: Replace attention-only goals with an action the product can support.

Proposed target: 100% of submitted recommendations have an explicit action and evidence basis.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Proposed experiments with a hypothesis, cohort, measurement window, decision rule and stop condition / all experiments recommended

When: Before experiment selection

If it needs attention: Narrow an ambiguous experiment before committing execution capacity.

Proposed target: 100% of recommended experiments; numerical targets follow baseline evidence.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Relevant recommendations with sponsorship/ranking interests, consent and claim limits addressed / all relevant growth recommendations

When: Each packet; before external distribution proposal

If it needs attention: Expose influence or unsupported promises and propose an alternative.

Proposed target: 100% of relevant recommendations; no implied partner approval.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Dispatched requests acknowledged by the real recipient/runner by ackDueAt ÷ dispatched requests due for acknowledgment.

When: At handoff; daily

If it needs attention: Check runner availability, update fallback and prevent duplicate dispatch.

Proposed target: Baseline first; every expired acknowledgment checkpoint gets a recovery action.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Due founder briefs delivered with results, due/late states, evidence, decisions and capacity timestamps ÷ founder briefs due.

When: Daily close

If it needs attention: Deliver a clear partial-state brief rather than withholding the report until everything succeeds.

Proposed target: 100% of delivered briefs include required fields; delivery reliability baseline open.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Founder requests requiring repetition or reconstruction because current state/source was missing; count with cause, not inferred sentiment.

When: Daily note; weekly

If it needs attention: Fix the recurring context gap or duplicate channel.

Proposed target: Establish observed baseline and reduce recurring causes; no fabricated time saved.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Scheduled provider observations with actual timestamp, account/bucket, window/reset and failure state ÷ observations due.

When: 9 a.m., afternoon and close if cadence adopted

If it needs attention: Refresh stale readings; preserve provider-specific gaps and reset events.

Proposed target: 100% of due observations produce a reading or explicit capture failure.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Requested advisory reviews returned by their agreed deadline with dated response ÷ reviews due.

When: At review; weekly

If it needs attention: Clarify scope or timing and record missing responses.

Proposed target: Baseline first; no recurring commitment inferred from the persona.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Reviewed advice packets containing a specific choice, alternatives, tradeoff, evidence/uncertainty and next observation ÷ advice packets reviewed.

When: At response

If it needs attention: Return vague encouragement or unsupported certainty for revision.

Proposed target: 100% of completed advisory packets meet the decision-support contract.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Material factual corrections to prior advice addressed by an agreed deadline ÷ such corrections due.

When: On new evidence

If it needs attention: Update the advice and name what changed; preserve original recommendation.

Proposed target: 100% of due material corrections receive a documented update.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.

Count: Advisory records explicitly distinguishing recommendation, founder decision and management assignment ÷ advisory records reviewed.

When: At response

If it needs attention: Correct any implication that advice itself authorized execution.

Proposed target: 100% of reviewed advisory records preserve the authority boundary.

Attribution: Assignment-level evidence with a declared cohort; shared department outcomes count once globally. This is a role view, not an independent result.