Lesson 08 — Provenance: Grant-Report Attribution Without Enterprise Audit Logs

Generation-time decisions (logged, per Playbook §3–5)

Learning objective

(Bloom's: Analyze) Determine what Bridgeway's Team-plan organization's actual provenance tools do and do not give you for grant-report attribution, and design the one record for your grant-report template that covers what the platform does not.

Standard named: AIHC.1.C1 (Provenance). Provenance is first-class (X5) — not an afterthought bolted on once someone asks. Note the boundary with Lesson 05 (B1, Verification): verification asked is this claim true; provenance asks who is accountable for it, and can I show that after the fact. By Lesson 05 you already know the Outcomes-section claims below check out against the export — this lesson is not re-checking their accuracy.

Cycle 1 — The assumption test: does "audit logs" solve this?

SAY — A common first instinct — "the platform logs everything, I'll just pull the audit log when someone asks" — is worth testing directly before you build anything, because it's a plan-tier assumption, not a settled fact. Per the doc-set's roles table, the ability to "Request audit logs" sits inside a permission area explicitly labeled "Security & data (Enterprise only)" — grouped there alongside "Manage SSO / auth," "Manage data retention controls," and "Manage feedback settings." Bridgeway is on a Team plan. This is not a setting you haven't found yet; it is a capability your plan tier does not have, named as such in the source itself.

SEE — (static table)

DO — Confirm, in your own words, one sentence: which of these four Enterprise-only tools your org does not have, and what that means for the assumption "the platform will log who did what." Recommended path: write it as a flat statement of fact for your rollout-plan.docx, not a complaint — you'll design the actual fix in Cycle 4.

Cycle 2 — What audit logs would (and would not) have told you anyway

SAY — Even if Bridgeway upgraded to Enterprise tomorrow, audit logs would not fully answer your real question. The doc-set names the exact fields captured: created_at, actor_info, event type, event_info, entity_info, ip_address, device_id, user_agent, client_platform, covering "authentication, project modifications, file uploads, SSO/JIT changes." And it names, just as explicitly, what is not captured: "title and content of chats and projects are not available to be exported in audit logs (only their unique identifiers will be exported)." Audit logs would tell you a project was modified. They would not tell you which sentence in your Q3 Family Support Outcomes Report came from which draft, or who verified the $145,000 figure against family_services_q3_export.xlsx. No plan tier gives you that from audit logs alone.

SEE — (static two-column comparison)

DO — For the Outcomes section's three real claims — "Served 3,200 families in Q3, a 14% increase over Q2," "89% of enrolled families completed at least one case-management session," "Distributed $145,000 in direct emergency assistance" — write the exact question you'd need answered for each if a board member asked "who's accountable for this line." Confirm, one more time, that no named platform tool answers it directly at any tier.

Cycle 3 — The tool that IS available: data exports

SAY — There is a real mechanism at Bridgeway's tier, and it's worth knowing precisely what it is (not more). The roles table lists "Request data exports" under the general "Security & data (Team/Enterprise)" row — available to Users, not gated to Enterprise. Unlike audit logs, a data export pulls actual chat/project content, not just identifiers. The doc-set doesn't further specify the export's format — treat it as the content-level source of truth you point to, not a finished attribution report on its own.

SEE — (static three-tier diagram)

Audit logs           →  Enterprise only · event/identifier level · NOT available to Bridgeway
Data exports          →  Team + Enterprise · actual chat/project content · available, but raw
Per-section attribution note  →  named nowhere in the doc-set · Devon has to build this himself

DO — Decide which real project or chat holds the Outcomes-section drafting work, and note where it lives (the specific Project name, per your own workspace) — that's the thing you'd data-export if a funder ever asked to see the source conversation.

Cycle 4 — Building the attribution record (BBQ pitmaster's log referent)

SAY — A competition pitmaster keeps a log for every entry — what rub went on, when it hit the smoker, when it came off, who tended it — not because the meat needs the paperwork, but because a judge can ask, and "trust me, it was good" doesn't survive that question. Bridgeway's platform gives you raw material (data exports); it gives you no report. Recommended path — a one-line footer convention on every grant-report section: [Drafted: Claude / <staffer name>, <date> · Verified against: <source file>, by <verifier name>] (Alternatives — a separate tracking spreadsheet, a project-per-report convention — available on request; not expanded here, per your option-suppression preference.)

SEE — (static before/after)

BEFORE:  "Served 3,200 families in Q3, a 14% increase over Q2."

AFTER:   "Served 3,200 families in Q3, a 14% increase over Q2."
         [Drafted: Claude / M. Alvarez, 2026-07-14 · Verified against:
          family_services_q3_export.xlsx, by D. Torres]

DO — Write the exact footer-convention text for your real grant-report template, and apply it to the "$145,000 in direct emergency assistance" line as a worked instance.

Independent at-bat

Apply the footer convention to the remaining two Outcomes-section claims in your real grant-report template, unscaffolded. Separately, for each claim, name which real Project you'd pull via data-export if a funder asked to see the underlying chat — no checklist provided this time.

Exit ticket (climbing to Analyze; graded against the doc-set)

1. (Remember) Name the two provenance-relevant permissions from the roles table and which plan tier each requires.
2. (Understand) Why wouldn't upgrading Bridgeway to Enterprise, by itself, solve the "who wrote this sentence" problem — what does audit-log export still not capture, per the doc-set?
3. (Apply) A board member asks who verified last quarter's volunteer-hours figure. Name the two things Devon checks — one platform tool, one convention he built.
4. (Analyze — objective level) One Outcomes-section paragraph was co-drafted by two staffers in the same Cowork session. Design how your footer convention should handle that case, and name, in one sentence, what the platform gives you no help with here at all — at any plan tier.

Ledger write

ledger_write:
  learner_id: L2-ADMIN-DEVON
  lesson_id: L2-admin-devon-08-c1-provenance
  anchors: [AIHC.1.C1]
  tags: [X5]
  exit_ticket:
    score: ""
    bloom_reached: ""
  auto_score: ""
  self_score: ""
  calibration_gap: ""
  journal_prompt: >
    Before this lesson, what did you assume "the platform will just log this" meant — and what's
    the one sentence you'd now correct that assumption with, for your own rollout-plan.docx?
  structure_used: PBL
  referents_used: [bbq-pitmasters-log]
  next_lesson_seed: "Lesson 09 (C2 Governance) takes the same enforceable-vs-not discipline from this lesson's audit-log finding and applies it to every rule in the data-handling-policy draft, not just attribution."

RUBRIC SELF-AUDIT (against Exemplar_and_Rigor_Rubric_Cowork_x_Nonprofits_2026-07-19_v01_I.md)

Escalation check: no load-bearing indicator fails → auto-ships per the versioned escalation policy (Playbook §5).