How this table works. The AI CMO reads each artefact against the categories in column 1. Every category that applies triggers the sign-off line(s) in column 3, and combines to give the quorum level in column 4. The review pack the AI CMO produces cites the row(s) that fired, so the four humans in the WhatsApp group know why their line is being asked to sign. Categories are cumulative — an artefact that touches two categories triggers both sets of sign-off lines.
Companion pages. /marketing-operators-manual (the shared-gate playbook · how the pack moves onto WhatsApp) · /marketing-review-checklist (the 7-checkbox card the AI CMO runs as a pre-check BEFORE this table).
Order of operations. First the AI CMO runs the 7-checkbox card (compliance-craft pre-check). If the card fails, the artefact goes back to the specialist — no review pack, no domain-line trigger, no WhatsApp post. When the card passes 7/7, the AI CMO then runs this assignment table (governance-domain lookup) to build the review pack. Corrina posts the pack to the WhatsApp group. Humans sign the triggered lines. Publish when quorum is met.
Contents
- 1. Legend — the four sign-off lines + routine (no-line)
- 2. Quorum levels — what "shipped" requires
- 3. The assignment table — every artefact category
- 4. The review pack — what the AI CMO puts in the WhatsApp message
- 5. Quorum-completion SLAs
- 6. Tie-break rules — when signers disagree
- 7. Change control
1. Legend
Routine — no domain lines
No domain-line trigger. Ships on the Marketing-Craft line alone (Corrina) after the AI CMO's 7-checkbox card passes.
Evidence & Brand — Laurence
Co-founder & CEO. Evidence integrity, spokesperson quotes, brand voice, operating narrative, hard-constraint items, journalist-response substance.
Philanthropic — Lisa
Co-founder & Head of Philanthropy. Donor / funder / trust / corporate messaging. Funder-facing landing pages. Grant/trust submission copy.
Financial & Regulatory — Carla
PMO/CFO (COSAI, external). $ figures and cost ratios. ACL/ABN/ACN/DGR/PBI/ACNC copy accuracy. Tech-that-runs-LWD. Google Ads admin.
Marketing-Craft — Corrina
External CMO. Voice fit, calendar-fit, channel-fit, ad-copy quality, execution judgement. Signs on almost every artefact (the constant of the shared gate).
All 4 humans — constitutional hard case
Real-beneficiary anything, hard-constraint page changes, material narrative shifts. Laurence + Lisa + Carla + Corrina all sign, regardless of which domain lines are nominally triggered.
2. Quorum levels
Four quorum levels, based on how many domain lines an artefact triggers. This is what "shipped" requires under the shared gate.
| Quorum level | Lines triggered | Required to publish | Notes |
|---|---|---|---|
| Routine | 0 domain lines | Marketing-Craft (Corrina) alone | Approved-calendar posts, ad-copy variants in pre-approved bounds, community replies in SLA template, typo/formatting fixes, weekly analytics summaries. |
| Domain-touching | 1 domain line | Domain owner + second-eye (any other human, brief acknowledgement) | Second-eye is not a sign of the domain line — it's a "read, no cross-domain concern" acknowledgement. Two humans confirm they've looked. |
| Multi-domain | 2+ domain lines | Every triggered line signs (no separate second-eye needed — the multiple domain signers ARE multiple eyes) | Common combos: Y1 number in donor page = Evidence & Brand + Philanthropic + Marketing-Craft. $ figure in press release = Financial & Regulatory + Evidence & Brand + Marketing-Craft. |
| Constitutional | Hard case (see rows 20–22 below) | All 4 humans sign, regardless of nominal domain-line trigger | Real-beneficiary anything, hard-constraint page changes, material narrative shifts. No "one line handles this" — all four look together. |
3. The assignment table
Every artefact category the AI agency produces, with its sign-off-line trigger and quorum level. If a row seems ambiguous, the AI CMO triggers the higher line-count — over-triggering costs a few extra minutes of a human's WhatsApp attention; under-triggering is the failure that hurts.
| Artefact category | Domain lines triggered | Quorum | Reason |
|---|---|---|---|
| LinkedIn post inside the approved calendar & voice (no new claims) | Marketing-Craft | Routine | Calendar + voice pre-approved. Post is an execution of an approved plan. Corrina signs Marketing-Craft; ships. |
| Ad-copy variant inside pre-approved creative bounds (headline / body / CTA re-mix on approved themes) | Marketing-Craft | Routine | Bounds pre-approved. Variant testing is a craft decision. Corrina signs Marketing-Craft; ships. |
| Community reply inside SLA template (comment / DM response to a routine query) | Marketing-Craft | Routine | Template pre-approved. Reply is an execution of a template. Corrina signs Marketing-Craft; ships. |
| Typo / formatting / broken-link fix on a published page (no content change) | Marketing-Craft | Routine | No semantic change. Corrina signs Marketing-Craft; ships. |
| Weekly analytics summary (Friday CMO Summary numbers section, no new claims) | Marketing-Craft | Routine | Reporting on already-published performance. No new external artefact. Corrina signs Marketing-Craft; ships (with all humans acknowledging on WhatsApp). |
| New use of a Y1 audited number (any artefact citing 31 households, $1.82M, 23-day, 52% K10, $1,840, 87c) | Evidence & Brand Marketing-Craft | Domain-touching | Evidence integrity is Laurence's line. He verifies the number is used with the outcome caveat and in a defensible context. Corrina signs Marketing-Craft. Second-eye acknowledgement from either Lisa or Carla. |
| Spokesperson quote for publication (anywhere — press release, LinkedIn, landing page) | Evidence & Brand Marketing-Craft | Domain-touching | Spokesperson decisions are CEO decisions. Words attributed to a named LWD person must be approved by that person or by Laurence on their behalf. Corrina second-eye plus Marketing-Craft. |
| Brand voice change / new operating narrative / new strategic messaging (a shift in how LWD describes itself) | Evidence & Brand Marketing-Craft | Domain-touching | Brand voice and operating narrative are CEO-owned. Corrina signs Marketing-Craft for execution quality. If the shift is material, this escalates to the constitutional row 22 below. |
| Response to a journalist query (any substantive answer — not the holding line) | Evidence & Brand Marketing-Craft | Domain-touching | Substantive journalist responses are spokesperson decisions. Holding-line acknowledgement is Corrina alone (30-min SLA). The substantive line always Laurence. If numbers cited, Carla joins (multi-domain). |
| Donor / funder / trust / foundation / corporate-partner message (any direct communication) | Philanthropic Marketing-Craft | Domain-touching | Philanthropic-network relationships are Lisa's line. She approves what goes to her network and how. Corrina signs Marketing-Craft. Second-eye acknowledgement from Laurence or Carla. |
| Update to funder-facing landing pages (/for-major-donors, /for-trusts-and-foundations, /donate, /funder-faq — Sprint 55 pages) | Philanthropic Marketing-Craft | Domain-touching | Lisa owns funder-facing surfaces. Content ships only on her sign-off. Corrina signs Marketing-Craft. If the page cites Y1 numbers or $ figures, this becomes multi-domain (add Evidence & Brand and/or Financial & Regulatory). |
| New grant application or trust proposal copy | Philanthropic Marketing-Craft | Domain-touching | Grant/trust submission is Lisa's line. Corrina signs Marketing-Craft for compliance-craft. If numbers or regulatory copy involved, add the relevant lines (multi-domain). |
| Any specific $ figure or cost ratio (financial claim beyond the Y1 canonical set) | Financial & Regulatory Marketing-Craft | Domain-touching | Financial integrity is Carla's CFO line. Y1 canonical numbers pass Box 02 of the pre-check — new $ figures need CFO sign-off. Corrina signs Marketing-Craft. |
| Admin-ratio, efficiency, or cost-of-fundraising claim (including any restatement of 87c-per-dollar in a new context) | Financial & Regulatory Evidence & Brand Marketing-Craft | Multi-domain | Regulator-adjacent financial claims. ACNC guidance on admin ratios must be honoured (Carla). The claim itself is evidence-facing (Laurence). Corrina Marketing-Craft. |
| Copy referencing ABN, ACN, ACL 387398, DGR/PBI/tax-status, or ACNC registration (any change to how these are stated) | Financial & Regulatory Marketing-Craft | Domain-touching | Registration/regulatory copy accuracy is Carla's PMO line. Even small phrasing changes need her verification against current registration status. Corrina Marketing-Craft. Second-eye Laurence. |
| Tech / infrastructure / website-integrity claim ("our site uses X", "our data is stored Y", "our hosting is Z") | Financial & Regulatory Marketing-Craft | Domain-touching | Tech-that-runs-LWD is Carla's line (she owns the tech constraint including the cosai-proposal.html hash). |
| Google Ads account admin (account creation, application to Ad Grants, billing, permissions) | Financial & Regulatory Marketing-Craft | Domain-touching | Account admin sits with Carla as PMO. Campaign design and copy is Paid-Ads specialist → Marketing-Craft; account admin also triggers Financial & Regulatory. |
| DGR / PBI / tax-status messaging (pre-DGR: none allowed; post-DGR: any wording) | Evidence & Brand Financial & Regulatory Marketing-Craft | Multi-domain | Regulator-facing status claim (Laurence CEO accountability) AND registration-copy accuracy (Carla PMO). Both must sign. Corrina Marketing-Craft on execution. |
| Complaint response draft (substantive response, not the same-day acknowledgement) | Evidence & Brand plus relevant domain: Philanthropic if donor / Financial & Regulatory if regulatory Marketing-Craft | Multi-domain | FIA Code complaint response is CEO-authored/authorised. Domain lead reviews for their specialty. Corrina Marketing-Craft. |
| Real LWD beneficiary imagery, quote, name, or identifying case detail (any form, any channel) | All 4 humans sign | Constitutional | Constitutional hard case. Requires written consent from the beneficiary AND all 4 humans sign, regardless of which line is nominally triggered. Stock imagery with faces is NOT this category — that's routine (Marketing-Craft only). |
Any change to cosai-proposal.html or artefact making a claim contradicting the hard-constraint page |
All 4 humans sign | Constitutional | Hard-constraint page is a constitutional item. Carla (Financial & Regulatory) verifies the sha256 hash and rolls back if the change breaks the constraint. Laurence's Evidence & Brand line carries decisive weight; all 4 sign. |
| Material shift in the LWD operating narrative (a new "here's what we do" that departs from the constitutional docs) | All 4 humans sign | Constitutional | Constitutional-level change. All four humans need to agree before it lands. If the shift is minor and only affects execution voice, treat as row 8 (Evidence & Brand + Marketing-Craft) instead. When in doubt, treat as constitutional. |
| Crisis event coordination (breaking news, beneficiary incident, external attack on LWD) | All 4 humans on WhatsApp within the hour; Evidence & Brand line is decisive on the public response | Constitutional | Marketing follows the CEO in crisis. Marketing does not lead. Corrina stops in-flight publishes and posts to the WhatsApp group within the hour so all four humans have the same situation picture. |
| New campaign concept or new audience segment (a whole new marketing direction, not an execution) | Evidence & Brand plus Philanthropic if donor-facing Marketing-Craft | Multi-domain | Strategic decisions get the sign-off before the AI is briefed to produce anything. Corrina proposes on Monday, humans sign the direction on WhatsApp, then the AI produces execution. |
| Paid-media spend authorisation over $500 in a week or $2,000 in a month | Financial & Regulatory Marketing-Craft | Domain-touching | Budget-authorisation limits are set by the CFO. Below these thresholds, Corrina authorises under Marketing-Craft. Above, Carla's Financial & Regulatory sign-off adds. Pre-DGR, live paid-media is minimal. |
The default when a row seems to fit but the AI CMO isn't sure. Trigger the higher line-count. Over-triggering costs a few extra minutes of a human's WhatsApp attention (annoying, not damaging). Under-triggering is the failure that hurts — publishing something Laurence would have flagged, or contradicting something Carla would have caught, or breaking a relationship Lisa was managing. The AI errs upward.
4. The review pack — what the AI CMO puts in the WhatsApp message
Every review pack from the AI CMO arrives as a structured six-item message. Not "can you look at this when you get a chance". The pack contains six items:
- The artefact itself. Inline for short artefacts (LinkedIn post, ad copy, email); a link to a shared doc for long ones (landing page, press release, proposal), with a one-line summary.
- 7-checkbox pre-check result. "Pre-check: 7/7 clean" — or, if a box needed inline fix, "Pre-check: 7/7 clean after inline fix — [what changed]" — or, if the AI wants to flag a soft signal, "Pre-check: 7/7 clean, flag for humans: [what to notice]".
- Domain lines triggered (row cite). Each triggered line named and the row from the table above quoted, so humans see exactly why they've been asked to sign.
- Quorum level required. Routine / domain-touching / multi-domain / constitutional. Names which humans' sign-offs are needed to publish.
- Reversibility note. If this publishes and turns out to be wrong, can it be retracted? At what cost? (One sentence.)
- Deadline. When quorum needs to be met by (day + time AEST), and what happens by default if the deadline passes with quorum incomplete (usually: hold, not ship).
Six items, ≤200 words each, single WhatsApp message when possible. Structured so a human can read in 3 minutes and reply with a signed line in 30 seconds. The review pack is the discipline that makes 30 min/week per human possible — without it, humans spend all their time reconstructing context.
5. Quorum-completion SLAs
Humans depend on the AI CMO delivering packs promptly, and on each other responding on WhatsApp promptly enough for the weekly rhythm to close. These are the response times each signer commits to.
| Pack type | Quorum-completion target | Default if quorum not reached |
|---|---|---|
| Routine (Marketing-Craft only) | Corrina signs same-day; ships | Held to next day; not blocked (Corrina is the only signer needed) |
| Domain-touching (weekly-planned, non-time-critical) | Within 48 business hours of the WhatsApp pack posting | Corrina @-mentions the un-signed line owner at 48h; if still no response at 72h, artefact is held, not shipped by default |
| Multi-domain (time-sensitive with a publish date this week) | Within 24 business hours of the WhatsApp pack posting | Corrina @-mentions all un-signed line owners at 24h; if still incomplete at 48h, escalate to Laurence for a call on hold-or-withdraw |
| Journalist response (holding line already sent) | Within 4 business hours (Evidence & Brand line first) | Corrina sends a second holding line to the journalist; call Laurence directly if under 60 min to the journalist's deadline |
| Constitutional hard case | Within 24 business hours (all 4 humans required) | Held indefinitely until all 4 sign. Withdrawal is a valid outcome — a constitutional item stalling means the artefact should not exist yet. |
| Crisis / beneficiary incident | All 4 humans acknowledge on WhatsApp within 1 hour | Corrina stops all live marketing activity and continues to attempt contact until reached; nothing publishes |
| Complaint acknowledgement (same day per FIA Code) | Corrina sends the same-day acknowledgement to the complainant without waiting; substantive response follows on multi-domain sign-off within 5 business days | Substantive response held; complainant is not left un-acknowledged (acknowledgement is not gated on quorum) |
Missed SLAs are a signal, not a failure. If a human misses a target repeatedly, it means their weekly load estimate is wrong — either what's being routed to their line is over-scope, or their availability has changed. Corrina raises it in the Friday CMO Summary on WhatsApp. The assignment table is revised, not the human hassled.
6. Tie-break rules — when signers disagree
Rare but real. Two humans give different verdicts on the same review pack (e.g. Lisa signs the Philanthropic line on the donor page, but Carla flags the ACL 387398 wording as needing a change). The tie-break rules:
- Same line, different views: not applicable. Each line has exactly one owner. If two humans disagree on the same artefact, it's because the artefact touches two lines — both must sign, for different reasons.
- Different lines, incompatible views: the artefact does not publish. Corrina rebriefs the AI CMO to produce a version that satisfies both signers. If no such version is possible, the artefact is withdrawn. Withdrawal is a decision, not a failure.
- Founder-line disagreement (Laurence's Evidence & Brand vs. Lisa's Philanthropic on the same artefact): the artefact does not publish. Laurence and Lisa resolve directly, off-WhatsApp if needed. Corrina and Carla do not attempt to mediate a co-founder disagreement — that's outside the contracted role of external contractors.
- External-contractor line vs. founder line (Carla's Financial & Regulatory vs. Laurence's Evidence & Brand; or Corrina's Marketing-Craft vs. any founder line): the founder line prevails. External contractors advise on their domain with full weight; founders carry the accountability. The disagreement is logged in Corrina's Friday summary on WhatsApp so the pattern is visible if it recurs.
- Deadlock over 5 business days: the artefact is withdrawn permanently (not "held"). Withdrawal is a decision. Sitting in a queue indefinitely is worse than deciding not to ship.
- AI CMO output disagrees with any human signer: the human is decisive. The AI CMO's Box 07 provisional voice-check does not override Corrina's Marketing-Craft line; the AI CMO's Y1-number verbatim check does not override Laurence's Evidence & Brand judgement on context. AI reports; humans decide.
7. Change control
This table is reviewed quarterly by all 4 humans on WhatsApp (one message each), and out-of-cycle when any of these three triggers fire:
- DGR granted. Row 18 (DGR / PBI / tax-status messaging) flips: pre-DGR "none allowed" becomes post-DGR "specific wording required" on the same multi-domain sign-off (Evidence & Brand + Financial & Regulatory + Marketing-Craft). Reviewed same-day.
- New audience, channel, or specialist charter launches. A new artefact type adds rows to this table. Assignment logic reviewed before the first artefact ships in that channel.
- A sign-off went wrong. Either something published on the wrong quorum (a domain line was missed), or something got over-triggered (a routine artefact hauled four humans in unnecessarily). The row is examined, categories tightened or loosened, update highlighted in the Friday summary on WhatsApp.