Skip to content

Assessment Format

Schema: scale-assessment.schema.json

An assessment records observed metric values for a single period, plus the journey and outlook narratives for that period. The catalog holds the timeless structure; the assessment holds what changed this cycle.

{
  "$schema": "https://productbuildershq.com/schema/scale/v0/scale-assessment.schema.json",
  "frameworkId": "scale",
  "period": "2026-Q3",
  "asOf": "2026-09-30",
  "observations": [
    {
      "metricId": "obs.consumption.otel-adoption",
      "value": 82,
      "numerator": 107,
      "denominator": 130,
      "source": "telemetry-backend presence check",
      "observedAt": "2026-07-15"
    }
  ],
  "narratives": [
    {
      "id": "obs-journey",
      "kind": "journey",
      "title": "Observability Q3",
      "body": "OTel adoption expanded from 45% to 82%...",
      "domainId": "observability"
    }
  ]
}

Top-level fields

Field Required Purpose
frameworkId Yes Must match the framework id the assessment targets.
period Yes The reporting period (e.g., 2026-Q3).
asOf No The as-of date for the period.
observations Yes Observed metric values.
narratives No journey and outlook blocks for this period.

Observation fields

Field Required Purpose
metricId Yes References a metric in the catalog.
value Yes The observed value.
numerator / denominator No Show the fraction (e.g., "107/130").
source No Where the data came from.
observedAt No When it was measured (RFC 3339).

Denominators matter

numerator / denominator document what was eligible, not just what was counted — the difference between "82% of 130 services" and a bare "82%".

Assessment narratives

Only time-bound narratives belong in an assessment:

  • journey — what changed this period.
  • outlook — where we're going next (may reference prism-roadmap initiative IDs so the next cycle can compare expected vs. actual movement).

Attach a narrative to a domain with domainId.

Validation

scale validate -catalog catalog -assessment assessments/2026-q3.json
catalog OK: 4 domains, 2 external models
assessment OK: period 2026-Q3, 24 observations