Accreditation Directory spec

Context and Problem Statement

ADR-0001 Wave 2 commits OPDA to a Maturity-based Accreditation Directory that replaces the current binary pass/fail Conformance Scheme with a richer two-axis published view:

• Assurance Level (existing AL1–AL4) — describes a claim, per-claim property.
• Capability score (new, three dimensions: Engagement / Process / Evidence) — describes a firm, per-firm property.

The high-level decision was made in ADR-0001 §“Wave 2 — Maturity-based accreditation”. This ADR specifies the artefact: where it lives, what it contains, who sees it, how member firms feed it.

The directly-related parallel workstreams (per ADR-0001 §“Vote and Dissent” newly-resolved #3):

• Data Quality framework (Compliance & Risk WG) — defines the dimensions the directory scores firms on for the DQ capability area.
• Data Security framework (Compliance & Risk WG) — same, for the Security capability area.
• PDTF overlay attachment policy (Technical WG) — provides attachment-handling rules referenced by the Directory’s evidence requirements.

The Directory acts as the integrator of these parallel workstreams: it provides the meta-framework into which DQ + Security capabilities are slotted, and exposes member firms’ scores against those capabilities.

Decision Drivers

1. Replace binary signal with maturity signal. A list of compliant firms doesn’t differentiate between a firm that just-passed and one that’s operationalised at scale.
2. Consumer + regulator legibility. Buyers, sellers, lenders, FCA, ICO need to be able to read a firm’s status without specialist knowledge.
3. Member-firm comparability. Firms want to know how they compare; the Directory provides that without needing OPDA to issue a league table.
4. Audit trail. Scores must be defensible — backed by artefacts when they matter, not just self-declared.
5. Operational pragmatism. OPDA is a small standards body. The Directory must not require a full-time directory manager.
6. Resolve the parallel-execution cross-dependency. Since Wave 2’s three workstreams run concurrently (per ADR-0001 §“Vote and Dissent” newly-resolved #3), the Directory must define its schema such that DQ + Security fill in their respective capabilities as they ship, without blocking Directory delivery.

Considered Options

• A — Extend the existing Conformance Scheme page. Add maturity scores as columns to the existing Conformance Scheme presentation. Smallest change. Risk: couples two distinct concepts (the scheme vs the directory of accredited firms) into one page that becomes hard to evolve independently.
• B — Separate Directory page, hybrid scoring (chosen). New page at /governance/accreditation-directory. Per-firm cards showing AL coverage + capability scores. Detailed per-capability data behind each firm; summary roll-ups on the public page. Best for readability + maintainability.
• C — Separate Directory site. Standalone site (directory.opda.org.uk) outside the Knowledge Base. Maximum independence; significant new infrastructure for OPDA to host. Overkill for current scale.

Decision Outcome

Chosen option: B — Separate Directory page, hybrid scoring, because of clean separation of concerns (Conformance Scheme defines the rules; Directory shows who meets them and at what maturity), reuse of the KB chrome and navigation, and no new infrastructure. The seven sub-decisions follow.

1. Hosting

New KB page at /governance/accreditation-directory. Lives in the Governance section’s “OPDA’s own rules” group, sibling to Conformance Scheme. Generated as a normal Astro page that consumes data from src/data/accreditation/ (see §5).

2. Scoring granularity — hybrid

Capabilities scored per-capability; public view rolls up to per-section.

• Underlying data: every capability the framework defines gets a score per firm (Engagement / Process / Evidence). ~30–50 capabilities total once DQ + Security ship.
• Public view: firm card shows section averages (~8 sections — Governance, Architecture, DQ, Security, Operating Model, Business Data Knowledge, etc.), with a “drill down” link to the full per-capability detail.
• Section averages: simple unweighted mean of capability scores within the section, rounded to one decimal.
• No overall “firm maturity score”. Resisted explicitly — overall scores invite gaming and obscure trade-offs. Per-section + per-capability is the deepest aggregation.

3. Visibility — tiered

Public tier suffices for consumers, regulators, comparison-shoppers. Member-firm portal can be added later when needed; this ADR doesn’t block on it. Requirements stub at docs/specs/member-firm-portal-requirements.md.

4. Update cadence — quarterly, with on-demand re-scoring

• Default cadence: quarterly. Firms submit their self-attested scores by the 15th of the quarter-end month; OPDA refreshes the Directory by the 1st of the following quarter.
• On-demand re-scoring: a firm can request mid-quarter re-publication when they pass a meaningful milestone (e.g. achieving AL3 issuer status, passing an audit). C&R WG approves the publish.
• Stale flag: firms whose data is more than 9 months old (missed two quarters) get a “stale” badge until they submit fresh data. Protects the Directory from showing rotted scores indefinitely.

5. Data schema — VC-backed self-attestation, JSON for the render

Two-format design:

• Source of truth: each firm submits a W3C Verifiable Credential signed by their accredited issuer DID, attesting to their capability scores. Stored under source/04-governance-bodies/accreditation/credentials/{firm-did}/{YYYY-Qn}.json.
• Render input: a build-time aggregator script (scripts/build-accreditation-directory.mjs) reads all current-quarter VCs, validates signatures, and produces a flat JSON file at src/data/accreditation/current.json consumed by the Directory page.

This means:

• Firms have cryptographic skin-in-the-game for their published scores.
• The Directory cannot diverge from what firms have actually signed.
• Anyone can independently verify a firm’s score by checking the VC against the Trust Registry.

Capability VC payload shape (sketch):

{
  "@context": ["https://www.w3.org/2018/credentials/v1",
               "https://opda.org.uk/pdtf/harness/accreditation/v1"],
  "type": ["VerifiableCredential", "AccreditationStatement"],
  "issuer": "did:web:firm.example.com",
  "issuanceDate": "2026-06-15T00:00:00Z",
  "credentialSubject": {
    "firm": "did:web:firm.example.com",
    "quarter": "2026-Q2",
    "assuranceLevels": ["AL2", "AL3"],
    "capabilities": [
      { "key": "dq.accuracy",
        "engagement": 5, "process": 4, "evidence": 3,
        "evidenceRefs": ["https://firm.example.com/audits/dq-2026-q1.pdf"] }
    ]
  },
  "proof": { }
}

6. Evidence requirements per score level — three tiers

Maps to the Engagement / Process / Evidence dimension scores (1–6 scale per ADR-0001 reference to DCAM):

Member firms self-rate; OPDA spot-checks. A spot-check finding an inflated rating (no artefact backing a score-4 claim, or no audit backing a score-6 claim) drops the score and adds an audit-trail entry visible in the drill-down view.

7. Naming — “Accreditation Directory”

Selected: “Accreditation Directory”. Continuity with the existing “Accreditation Scheme” terminology (from the Articles of Association). Reads naturally: the Scheme defines the rules, the Directory shows who meets them. Alternatives considered:

• “Maturity Register” — too generic; doesn’t communicate “this is the list of accredited firms”.
• “Conformance Directory” — overloaded with “conformance” which is binary in current usage.
• “Member Capability Profile” — accurate but cumbersome.

Resolving the parallel-execution cross-dependency

ADR-0001 §“Vote and Dissent” newly-resolved #3 committed DQ, Security, and Maturity Accreditation to running in parallel under Compliance & Risk WG. This creates a real cross-dependency:

• Maturity Accreditation defines the shape of capability scoring (Engagement / Process / Evidence at 1–6).
• DQ + Security define the content — which capabilities exist within their domains, what each measures.

Resolution: the Directory ships first with a skeleton schema; DQ and Security each contribute a “capability bundle” PR.

Phases 1–3 run in parallel; Phase 4 depends on all three completing.

Consequences

• Good, because member firms get a quantifiable maturity story — not just a pass/fail badge.
• Good, because consumers + regulators get a richer signal — they can distinguish “early adopter” from “mature operator”.
• Good, because self-attestation is cryptographic — VC-signed claims are harder to repudiate or inflate without detection.
• Good, because the spec resolves the parallel-execution cross-dependency — DQ + Security can ship without blocking on Directory finalisation.
• Good, because continuity with existing language — “Accreditation Scheme” + “Accreditation Directory” reads as one coherent governance machinery.
• Bad, because of VC infrastructure dependency. The Directory requires the Trust Registry to be reliable for signature verification. If the Trust Registry has downtime, Directory rebuilds fail. Mitigation: snapshot the most recent valid build state and serve that as a fallback.
• Bad, because of self-attestation gaming. Firms have an incentive to inflate. Mitigation: spot-check process + visible audit-trail when discrepancies are found.
• Bad, because of the recruitment dependency for parallel execution. Cross-references ADR-0001 §“Vote and Dissent” newly-resolved #3 — if Engagement WG can’t recruit C&R members fast enough, the parallelism collapses back to serial.
• Bad, because quarterly cadence may feel slow to high-velocity firms. On-demand re-scoring mitigates; if pressure mounts, can move to monthly later (no ADR needed — just a cadence change).
• Bad, because no “overall maturity score” may frustrate consumers wanting one number. Deliberate. If pressure mounts, a weighted overall could be added — but the weighting becomes politicised. Better to refuse.
• Neutral, because schema design (VC + JSON) is moderate cost to change. Once firms have submitted VCs, the credential format must be backward-compatible or migrated.
• Neutral, because page structure / visibility tiers and cadence are low cost to change; naming is low cost early but high cost once external citations reference “Accreditation Directory”.

Confirmation

• EC ratification of this ADR is Phase 0; subsequent phases (per “Resolving the parallel-execution cross-dependency” table) tracked in ADR-0005 sections A and C.
• Acceptance test for first quarterly publish: the Directory page renders ≥1 firm card from a verified VC, the drill-down view exposes per-capability scores, and an independent verifier can fetch the VC and reproduce the signature check against the Trust Registry.
• Spot-check process documented as part of Phase 1 (placeholder pending Phase 1 kickoff); spot-check findings logged in the audit-trail surface of the drill-down view.

More Information

Open questions

1. OPDA-listed audit partners. Phase 4 requires a list of accreditation-audit-eligible DIDs (firms that can sign the score-5-or-6 audit attestations). Who’s on the initial list? Likely the Big 4 + a few specialists (e.g. KPMG, Deloitte, BDO + property-data specialists). Engagement WG action — scaffold at docs/recruitment/opda-listed-audit-partners.md.
2. Conformance Scheme integration. The existing page at /governance/conformance-scheme will need an update to point at the Directory and explain the AL × Capability relationship. ADR for the Conformance Scheme update — out of scope for this ADR.
3. Member-firm portal (deferred tier). When does the private detailed view get built? Probably triggered by the first member firm asking for it; ADR at that point. Requirements stub at docs/specs/member-firm-portal-requirements.md.
4. Quarterly publishing automation. Step 5 of Phase 4 needs an automated pipeline (cron, GitHub Action, etc.) to refresh the Directory. Tooling decision — out of scope for this ADR; an Implementation ADR if needed.
5. What happens to a firm that leaves OPDA membership mid-quarter? Stays in Directory as “Former member, last data YYYY-Qn” or drops out immediately? Membership ADR territory — draft scoping at docs/adr/_drafts/membership-lifecycle.md.

References

Primary sources

• ADR-0001 — selective adoption of DCAM / DAMA-DMBOK; Wave 2 “Maturity-based accreditation” commits to this artefact.
• ADR-0002 — folder hierarchy under which /governance/accreditation-directory will live.
• ADR-0003 — Astro architecture this page builds on (typed site.ts, build-time chrome).
• ADR-0005 — Deferred work register; tracks Phase 1–4 progress.

OPDA source files referenced

• src/pages/governance/conformance-scheme.astro — sibling page; will need update to reference the Directory.
• source/03-standards/trust-framework/docs/governance.md — Trust Registry mechanics that VC verification relies on.
• source/01-organisation/constitution-and-policies/Articles of Association 2026.pdf — defines the Accreditation Scheme (extends naturally to the Directory).
• source/01-organisation/accreditation/Accreditation Scheme 2026.docx — current Scheme document; will need a small update to acknowledge the Directory.

External references

• W3C Verifiable Credentials Data Model 2.0 — credential format.
• ISO 19011 — guidelines for auditing management systems (audit-partner eligibility benchmark).

Amendments

• 2026-05-25 — Refactored to canonical MADR 4.x format. Bullet-list metadata moved to YAML frontmatter; section names aligned to the MADR vocabulary; “Open questions” preserved under ## More Information. Filename gained the ADR- prefix per the ruflo-adr adr-create skill. Substance unchanged.