Spoke

jobframe-analytics

Job-data analytics that tell you how much to trust each number — and ask the fewest people the fewest questions to make the shaky ones solid.

Character

You're a people-analytics or comp practitioner who needs job data you can stand behind. Half of it was a model's guess, some is stale, some is missing — and dashboards that hide that produce confident-looking nonsense.

Problem

External. Job data is too messy to trust, and the usual fix is to dump a giant cleanup project on one overwhelmed HR admin.

Internal. You're asked which business units you can actually price, and you can't say — because the numbers don't carry where they came from.

Philosophical. A dashboard should show its and its uncertainty, and corrections should be governed evidence — not a silent overwrite.

Guide

jobframe-analytics is the analytics-with-confidence layer. Every mapped value carries a five-state (observed / predicted / respondent-confirmed / reviewer-approved / tenant-override); six dashboards each show a per-segment provenance breakdown, a , and a min-N privacy gate. When a segment looks weak, launch a — the fewest structured questions routed to the people most likely to know, capped so nobody is over-asked. Answers update provenance; disagreements over approved or sensitive values route to governance.

Abstract

Background. Analytics over job data are only as trustworthy as the data, yet most dashboards hide how much was a guess and "fix it" by overloading one admin.

Methodology. A 5-state value provenance enum tags every cell; six dashboards carry per-segment provenance + a trust index + a min-N gate (composed from data-anonymizer); charts are viz-catalog payloads, never bespoke; an objective → required-fields → gap-scan → tasks planner builds minimal, info-gain-ordered question sets routed via org-graph with per-respondent burden caps.

Scope. Analytics-over-canon + the campaign planner + the provenance enum — it composes the canon, the tenant workspace, the survey runtime, and the governance console; it does not own them.

Contribution. Provenance-and-trust-first analytics and a that turns cleanup into the fewest questions to the right people.

Evidence / Provenance. PAT-JFE-ENT-6 + ENT-7; responses become governed evidence, not uncontrolled truth.

Plan

  1. 01

    Run a dashboard

    POST /api/spokes/jobframe-analytics/analytics for a provenance-aware dashboard with a per-segment trust index.

  2. 02

    Scan the gaps

    POST a data objective to find which profiles fall short of what you need to trust them.

  3. 03

    Plan a campaign

    POST /campaigns (preview first) — minimal questions, info-gain-ordered, routed to the people most likely to know, burden-capped.

  4. 04

    Ingest responses

    POST responses to update provenance; conflicts over approved values route to governance.

Call to Action

Direct. curl the analytics endpoint and rank your business units by trust index.

Transitional. Read the PRD 20/21 notes on provenance + the clarification loop.

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.

JobFrame AnalyticsINPUTSMAIN ACTIONSOUTPUTSMapped job-data valuesMappedValue[]Data objectiveDataObjectiveOrg-graph routingKnowledgeNode[]Score trust + provenanceScan gapsPlan clarification campaignTrust dashboard + tasksAnalyticsResult↳ 5-state provenanceCOMPOSES WITHcalculusdata-anonymizerorg-graphsurvey-orchestrator

Try it now

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

jobframe-analytics.health

GET

Schema reachability for JobFrame Analytics + Clarification Campaigns.

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

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/jobframe-analytics/contracts/types.ts

Source path: src/spokes/jobframe-analytics/contracts/types.ts · GitHub

Try in Dev ConsoleData-trust dashboard + campaign previewBack to all spokes

Failure

Dashboards hide their uncertainty and produce confident-looking nonsense; the cleanup lands on one admin and never finishes.

Success

Every number shows how much you can trust it, and the shaky ones get solid via the fewest questions to the right people — corrections governed, individual pay never exposed.