Skip to content

ADR-0019 — NGO-TARGETS broad-roster (ShareAction) pass — assessed and declined for Phase 1; seed complete at cycle-15a high-confidence state

Field Value
Status Accepted
Date 2026-05-23
Supersedes —
Superseded by —
Related ADR-0018 (SFDR/EU-Taxonomy assessed-and-declined — same no-defer close pattern); quarterly source-instrument continuity review; open NGO-TARGETS RAN BoBC/BOCC provenance item

Context

At cycle 15a, the NGO-TARGETS seed was refreshed to 19 high-confidence in-universe entries, adding 3 net-new backslider entries (Nordea E7.3; BlackRock E7.3+S6.2) from the ShareAction Asset Manager Survey 2025. The high-confidence seed covers institutions confirmed by the operator as active NGO campaign targets or backsliders. run_id=35 confirmed all three new signals at conf=1.0.

The full ShareAction broad roster — leaders, the asset-manager survey ranking table, and lower-confidence matches — was left undone at cycle 15a. ShareAction's report pages are rate-limited in a way that blocks programmatic fetch (same blocker as the cycle-12 LobbyMap acquisition); the broad-roster pass was noted as requiring a Claude-in-Chrome session and recorded as deferred to that future session.

A separate open item — the RAN BoBC-vs-BOCC seed labelling and provenance question — was also recorded as parked for the next quarterly source-instrument continuity review, pending resolution before any further seed expansion.

This ADR records the scoping decision on the broad-roster pass.

Decision

Do not run the ShareAction broad-roster pass. The NGO-TARGETS seed is complete for Phase 1 at the cycle-15a high-confidence state (19 entries). No Chrome session, no seed expansion, no migration.

The item is assessed and declined, not deferred. It is not on the Phase 1.5 carry list.

Rationale

Three independent grounds, each sufficient on its own:

1. Thin and no-op marginal value on a deduction source. NGO-TARGETS feeds deduction rules E7.3 (financials) and S6.2 (financials) — it records campaign targets and backsliders, not leaders. The cycle-15a pass already captured the high-confidence in-universe backsliders. The uncaptured broad roster skews toward (a) ShareAction leaders — institutions rated well, which under neutral-prior produce a not-found / no-deduction outcome that is scoring-identical to absence — and (b) lower-confidence cross-universe matches. The overwhelming majority of additional entries would not move any score.

2. No-defer principle. The broad-roster pass can only run via Claude-in-Chrome (rate-limit blocker). Queuing it as a "future Chrome session" is a deferral to a no-context future session. Per operator decision, items are either done now or skipped — not parked for an unscheduled session. No Chrome slot is being booked, so the honest action is to close the item, not carry it. This is the same close pattern as ADR-0018 (SFDR/EU-Taxonomy assessed-and-declined rather than deferred to SFDR 2.0 finalisation).

3. Open provenance question. The RAN BoBC-vs-BOCC seed labelling/provenance question remains open, parked for the next quarterly source-instrument continuity review. Expanding the seed before that is resolved would propagate the labelling ambiguity across more entries. The seed is better treated as complete at the 15a high-confidence state until the provenance question is addressed.

Consequences

  • NGO-TARGETS E7.3 and S6.2 coverage is final for Phase 1 at 19 seed entries (cycle-15a state). No further expansion is planned before Phase 1 close.
  • Any future expansion of the NGO-TARGETS seed should be a fresh scoping decision taken alongside resolving the RAN BoBC/BOCC provenance question — ideally at the quarterly source-instrument continuity review — not reverse-engineered from this deferral.
  • The ShareAction broad-roster rate-limit blocker is not a carry item; if the pass is revisited, it starts from a fresh assessment of whether the marginal value justifies the Chrome session.
  • The open NGO-TARGETS time-window expansion question (whether to broaden to a 24-month lookback window at annual refresh) is a separate item and remains parked at the quarterly review — it is not folded into this close.

Alternatives considered

Run the broad-roster pass via Claude-in-Chrome — declined. The Chrome session is not scheduled and would require operator coordination. More importantly, even if run, the majority of entries would produce no-deduction outcomes on a deduction-only source, making the marginal value thin relative to the session cost and the open provenance risk.

Defer to Phase 1.5 pending provenance resolution — declined. Same reasoning as ADR-0018: deferral without a declining decision leaves the item open indefinitely. Assessed-and-declined is cleaner than a carry with no expected resolution date.

References

  • Cycle 15a NGO-TARGETS seed refresh (d6f2636) — 16 → 19 entries; run_id=35 confirms all signals conf=1.0; broad-roster pass first recorded as deferred-to-Chrome
  • ADR-0018 — SFDR/EU-Taxonomy assessed-and-declined (same no-defer close pattern)
  • Quarterly source-instrument continuity review (parked queue, backlog)
  • Open NGO-TARGETS RAN BoBC/BOCC provenance item (parked at quarterly review, backlog)
  • NGO-TARGETS time-window expansion (separate parked item, not closed by this ADR)