Spoke

manager-effectiveness

Nine-domain Manager Effectiveness Index — empirical pillar blends via performance-validity evidence.

Character

Leadership demands a headline “people manager score” while you insist on nine differentiated domains with defensible measurement references.

Problem

External. Narratives sway exec committees because weights were negotiated politically, not evidenced empirically.

Internal. Updating weights consumes weeks inside Sheets—with zero append-only ledger for auditors.

Philosophical. Composite leadership constructs must admit evidentiary revision without silent retrofits.

Guide

manager-effectiveness expands each MEI pillar to enumerated measureIds, persists tenant pillar weights (tenant_domain_weight_profiles), blends predictive r² imports from performance-validity into defaults via configurable alphaEmpirical, and emits append-only audits per refresh.

Abstract

Background. Multi-domain leader models abound academically but rarely expose auditable pillar weights tethered to outcome predictability.

Methodology. buildMeiCompositeScore narrates composites with measurement-aware phrasing; pooled outcome summaries arrive through performance-validity.mei-predictive-evidence UPSERT feeding weights/recalibrate convex blends vs FiveTran-aligned priors (~0.7 empirical default).

Scope. Depends on trustworthy rating payloads—not a standalone survey SaaS.

Contribution. Typed HTTP + MCP, immutable audit ledger, explanatory strings for ops teams.

Evidence / Provenance. PAT-159 scaffolding + README references to empirical blend policy.

Plan

  1. 01

    Configure weights

    Upsert pillar profiles per tenant—or accept curated defaults mirrored from consulting baselines.

  2. 02

    Publish predictive evidence

    Populate performance-validity bridge rows whenever new cycles justify empirical refresh.

  3. 03

    Recalibrate + narrate

    Invoke POST …/weights/recalibrate; capture audits + conversational justification strings.

  4. 04

    Emit composites

    Downstream dashboards pull MEI composites with pillar drill-through via stable measure identifiers.

Call to Action

Direct. Inspect contracts/types.ts for domain vocabulary.

Transitional. Pair README with PAT-159 session artefacts.

Spoke I/O (visual language v1)

Every toolbox spoke shares the same abstract choreography: typed inputs on the left, distilled verbs in the center, typed outputs on the right, and (when relevant) cross-spoke HTTP composition along the bottom rail. Source package: @people-analytics-toolbox/spoke-illustrations.

Manager effectivenessINPUTSMAIN ACTIONSOUTPUTSValidity telemetryConstructDiagnosticsCalibration windowsCalibratorTrajectoryPackSegment membershipsMembershipEdgeRef[]Fuse cross-spoke signalsCompose MEI + archetypesLeader effectiveness bundlesMeiCompositePack↳ HTTP composition only↳ coaching narrativesCOMPOSES WITHperformance-validityperformance-calibrationsegmentation-studioworker-resolution

manager-effectiveness · weights recalibrate

Default weights are a starting prior — evidence moves them.

The nine-domain MEI ships with a FiveTran-aligned prior (upward-feedback weighted heaviest). Recalibration reads predictive-validity R² per domain from performance-validity and blends it against that prior — so the index learns which domains actually predict outcomes for a given tenant.

POST /api/spokes/manager-effectiveness/weights/recalibrate Authorization: Bearer $TOOLBOX_SERVICE_KEY { "tenantId": "TENANT_ID", "validityScorecardCycleIds": ["CYCLE_IDS"], "alphaEmpirical": 0.7, "outcomeKinds": ["retention", "promotion", "next-cycle-perf"] }
DomainPriorRecalibratedΔ
upward-feedback0.400.27−0.13
decision-discipline0.200.18−0.02
operational-health0.200.23+0.03
people-development0.100.11+0.01
team-output0.040.07+0.03
talent-stewardship0.020.03+0.01
equity-fairness0.020.05+0.03
self-awareness-consistency0.020.03+0.01
span-load-fit0.020.04+0.02

Worked example: newWeight = normalize(α·(R²ᵢ / ΣR²) + (1−α)·prior), α = 0.7. Per-domain R² are illustrative inputs to show the math — not a production tenant's result. The response also returns perDomainRSquared, alphaUsed, and a plain-English narration.

Try it now

Copy this curl. Paste in any terminal. Public read — no auth needed.

manager-effectiveness.health

GET

Heartbeat verifying schema reachability until you POST composite payloads.

curl -sS "https://people-analytics-toolbox.vercel.app/api/spokes/manager-effectiveness/health"

manager-effectiveness.tenant.weights.recalibrate

POST

SERVICE KEY REQUIRED

Convex blend of pooled Q2 explanatory mass vs priors — returns old/new pillar weights plus narration audit string. Populate tenantId + at least one performance-validity scorecard cycle ids you ingested beforehand.

curl -sS -X POST "https://people-analytics-toolbox.vercel.app/api/spokes/manager-effectiveness/weights/recalibrate" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TOOLBOX_SERVICE_KEY" \
  -d '{
  "tenantId": "YOUR_TOOLBOX_TENANT_ID",
  "validityScorecardCycleIds": ["cycle-uuid-from-performance-validity-ingest"],
  "alphaEmpirical": 0.7,
  "outcomeKinds": ["retention", "promotion"]
}'

Vendor the contract

The Zod contract is the source of truth. Vendor a copy into your consumer app — you keep it; we don't break it underneath you. Re-vendor when the version bumps.

// Vendor canonical types:
// src/spokes/manager-effectiveness/contracts/types.ts

Source path: src/spokes/manager-effectiveness/contracts/types.ts · GitHub

Try in Dev ConsoleRead the explainerInsight cards when seededBack to all spokes

Failure

Exec spats about weights devolve into politics—no reproducible predictive evidence attaches to the headline index.

Success

Every refresh cites blended empirical mass + audited defaults; narratives explain pillar shifts with quantitative provenance.