Sixty Stakeholders Is Not a Stakeholder Plan
Designing a four-tier engagement model when every executive wants a steering seat and every external party wants weekly reviews. Plus, why Firebase Push is not an internal broadcast channel.
Everyone Wants a Seat
Kickoff on 7 February had sixteen people in the room. By 12 February, the program manager's inbox had twenty-three separate requests for different kinds of involvement. By 18 February, a working count of the stakeholder universe sat at sixty-one people.
Samuel Osei (Head of Retail) wanted a weekly update directly from the program manager. The three mobile network operator relationship managers each wanted a seat on steering. Two of the three central-market regulators wanted observer status. The CFO's deputy wanted to attend steering in the CFO's place. The Head of Corporate Banking, whose unit was explicitly out of scope, wanted "just a quarterly briefing so I can tell my team what's going on." The platform integrator had three account managers, each of whom wanted fortnightly reviews with the program manager.
None of these requests were unreasonable. Each one, taken alone, had a credible justification. Taken together, they described a governance structure with sixty-one active participants, which is not a governance structure at all.
The requests were not really for governance. They were for proximity. The program was visibly important, board-approved, budget-funded, and on the CEO's public roadmap. Being adjacent to it was career-useful. A lot of the requests for "a seat" or "a weekly update" were politics presented as process.
Reading this correctly mattered. Stakeholders with legitimate governance needs had to be treated as such. Stakeholders chasing proximity had to be kept informed without being given decision authority. The two groups required different engagement mechanisms, not the same one with different labels.
The program manager had five working days to propose a stakeholder engagement model before the 20 February steering meeting. Whatever got approved would shape every governance conversation for the next 18 months.
Four Tiers, Not One Committee
On Day 30, the program manager sketched a four-tier engagement model on a whiteboard in the PMO room and walked Ahmed Hassan through it. Hassan made one suggestion (rename Tier 4 from "Passive" to "Broadcast" to avoid the implication that these stakeholders did not matter), and the model went onto a Confluence page that afternoon.
The model was not sophisticated. Its value was in being explicit about who fell into which tier and why. Implicit tiering, which is what most programs actually operate, creates ambiguity that stakeholders exploit to upgrade their own status.
Tier 1 — Direct and Govern (5 people). Steering committee. Decision authority on scope, budget, schedule escalations. Monthly meeting, fortnightly readouts.
Tier 2 — Consult and Influence (15 people). Working groups. Shape the work, no decision authority on program-level matters. Fortnightly cadence per group.
Tier 3 — Inform and Contribute (20 people). Subject-matter consultation as needed. No regular cadence; engagement is issue-driven.
Tier 4 — Broadcast and Aware (21 people). Monthly Confluence digest plus email notification. Read-only. No meeting cadence.
The tier model also forced a question the program had been avoiding: how would Tier 4 actually receive communication? Twenty-one people, spread across three markets, representing the central bank, the regulators, the data protection authorities, the board observers, and the adjacent business units. A monthly digest needed an actual channel.
Dmitri Volkov, the platform integrator lead, made a suggestion in the 12 February architecture review that deserved its own decision. He proposed using the production Firebase Push infrastructure, which the program was standing up for customer notifications, as the delivery mechanism for Tier 4 internal updates.
Volkov's logic was efficiency: the infrastructure was being provisioned anyway, the in-country residency relays would be in place, and tokenized delivery was already solved. Why stand up a parallel internal channel?
The answer is that Firebase Push is a customer-facing production surface, subject to central bank audit obligations on message logging and PII controls. Running internal governance comms through it would pull every Tier 4 digest into the audit perimeter. Every broadcast would require the compliance review that customer push messages require. The operational cost of mixing traffic exceeds the capital cost of standing up a second channel.
Firebase Push stays locked to customer use, tokenized, no PII, routed through the in-country relay. Tier 4 internal communication runs on Atlas Bank's existing corporate infrastructure: a shared Confluence space for program updates, with a monthly email digest dispatched from the PMO.
Tier Classification for Sixty-One Stakeholders
Assigning sixty-one stakeholders to four tiers by hand is a half-day of work and introduces inconsistency across the list. The program manager used an AI-assisted pass to produce a draft classification, then reviewed and corrected it before the model went to steering.
The prompt below follows the RACE structure: Role (experienced program manager classifying stakeholders in a regulated banking environment), Action (assign each stakeholder to one of four tiers with an engagement mechanism), Context (program state, charter commitments, regulated three-market environment), Expectation (tier plus rationale plus flag for ambiguous cases).
RACE is lighter than CO-STAR and CRAFT. For classification work where Style, Tone, and Target Audience add no value over Context, Action, and Expectation, the lighter framework produces cleaner output faster. Framework complexity should match task complexity.
Classification Draft, Sample
A sample of the classification output showing six representative stakeholders from across the four tiers, plus the three flagged cases the AI surfaced at the end. Full output covered all sixty-one entries; the structure and the flagged-cases section are what matter here.
| Stakeholder | Tier | Mechanism | Rationale |
|---|---|---|---|
| Fatima Idris CDO | TIER 1 | Monthly steering + ad-hoc escalations | Program sponsor; scope, budget, schedule authority. |
| Abebe Alemu CTO | TIER 1 | Monthly steering | Technology authority, co-signs Scale Plan. |
| Amara Okonkwo Head of Compliance | TIER 2 | Fortnightly compliance working group | Central bank interface; shapes work but not decision authority on scope. |
| Priya Raman Technical Lead | TIER 2 | Weekly engineering sync + fortnightly architecture review | Shapes technology direction, consulted on Scale Plan commitments. |
| MNO-A Account Mgr Mobile Operator A | TIER 3 | Integration checkpoints per milestone | Vendor-equivalent; input during integration phases only. |
| Head of Corporate Adjacent Business Unit | TIER 4 | Monthly Confluence digest | Out of scope; informed only, no cadence. |
1. Amara Okonkwo (Head of Compliance). Classified as Tier 2 based on "shapes the work but no decision authority." However, central bank certification is a go-live gate. If compliance sign-off fails, the program does not launch. Rationale for Tier 2 may be too structural; Tier 1 may be more accurate given the veto power this role effectively holds.
2. MNO-A / MNO-B Account Managers. Both requested steering seats. Classification as Tier 3 is structurally correct but may produce political friction. Consider whether a named Tier 2 slot for "mobile operator relationship" would resolve without upgrading any single operator.
3. CFO's deputy. Not in the source list, but the PM's notes indicate the deputy wishes to attend steering in the CFO's place. This is a delegation question, not a tier question, and should be resolved with the CFO directly before the model is published.
Stakeholder Register, Version 1.0
The final Stakeholder Register, published 20 February 2025 after human review of the AI draft. Representative entries shown per tier. Full register carries sixty-one entries with the same structure.
| Stakeholder | Role | Engagement Mechanism |
|---|---|---|
| Fatima Idris | CDO, Program Sponsor | Monthly steering chair; critical escalation SLA 4hrs |
| Abebe Alemu | CTO | Monthly steering, technology authority |
| Samuel Osei | Head of Retail Banking | Monthly steering, business owner for both channels |
| Nadia Benali | Chief Risk Officer | Monthly steering, risk and regulatory escalation point |
| CFO | Chief Financial Officer | Monthly steering, financial authority; delegation to deputy requires sponsor approval |
| Stakeholder | Role | Engagement Mechanism |
|---|---|---|
| Amara Okonkwo | Head of Compliance | Compliance WG (fortnightly); central bank sign-off holder |
| Ahmed Hassan | PMO Lead | Program review (fortnightly); daily PM sync |
| Priya Raman | Technical Lead | Architecture WG (fortnightly); weekly engineering sync |
| Jin-ho Park | Development Lead | Architecture WG (fortnightly) |
| Dmitri Volkov | Platform Integrator Lead (vendor) | Vendor WG (fortnightly); monthly contract review |
| Mobile Operator Relationship | Named slot, rotating across 3 operators | Vendor WG (monthly); quarterly full-operator review |
| +9 others | Market compliance leads, vendor technical leads, audit firm lead | Working-group cadences by domain |
| Category | Examples | Engagement Mechanism |
|---|---|---|
| Legacy Core Banking Team | Core banking integration leads (Atlas Bank) | Integration checkpoints per milestone |
| Mobile Operator Technical | 3 operators × 2 technical leads each | API integration sessions (on demand) |
| Market Operations | Ops leads per market (3) | Deployment readiness reviews per market |
| Security Audit Firm | External auditor engagement team | Quarterly audit checkpoints, ad-hoc queries |
| +~10 others | Vendor account managers, compliance specialists | On demand |
| Category | Examples | Engagement Mechanism |
|---|---|---|
| Atlas Bank Executive | Head of Corporate Banking, Head of SME, COO | Monthly Confluence digest + email |
| Regulatory Observers | Data protection authorities (3 markets) | Monthly digest; ad-hoc compliance sessions via Tier 2 |
| Board Observers | Non-EDTC board members | Quarterly board briefing, digest between quarters |
| +~15 others | Internal audit, corporate comms, HR business partners | Monthly digest |
What the Tiered Model Actually Did
On 18 February, two days before steering, the program manager walked the draft register through a one-on-one preview with Nadia Benali, the CRO. Benali stopped reading at Tier 1.
"Why is Amara at this table?"
The question was pointed but not hostile. Amara Okonkwo, Head of Compliance, reported into Benali's risk and compliance function. Placing her in Tier 1 alongside the CDO, CTO, CFO, and Head of Retail put her at peer level with Benali herself. That was not her organisational position.
"Because central bank certification is a go-live gate. If Compliance declines to sign, the program does not launch. A veto holder belongs in the room where the decisions they can veto are made."
Benali considered it. "You are telling me organisational hierarchy is not the right lens."
"I am telling you decision consequence is. If the gate were in your function directly, I would want you at Tier 1 regardless of seniority."
Benali did not endorse the classification in that conversation, but she did not block it either. At steering on 20 February, she raised it publicly in a way that gave Idris a chance to back the call. Idris did. Okonkwo stayed at Tier 1, the register was approved with one amendment (Idris requested a quarterly briefing for the Head of Corporate Banking, carried as a named exception in Tier 4), and the model went live.
The real test of the tiered design was not the mobile operators, the CFO deputy, or the Head of Corporate Banking. It was whether governance authority could be defended on decision-consequence grounds rather than hierarchy grounds. It could. But it took a one-on-one preview and a steering-table moment to get there.
- Amara Okonkwo moved from Tier 2 to Tier 1. The AI correctly flagged this as a review candidate. The PM agreed: central bank certification is a go-live gate, which means Compliance holds an effective veto. A veto holder belongs in the room where the decisions they can veto are made. Tier 2 framing would have left this authority ambiguous.
- Mobile operator relationship restructured as a named Tier 2 slot. The AI's flag was right that three individual Tier 3 classifications would produce political friction. The PM created a single "Mobile Operator Relationship" Tier 2 slot, rotating across the three operators on a quarterly cycle. Each operator gets structured input without any single one being upgraded.
- CFO deputy issue resolved upstream. The PM called the CFO directly and asked whether the deputy should attend steering. The CFO declined the delegation and committed to attending personally. This was a 20-minute phone call that the AI correctly flagged but could not resolve.
When the vendor dispute hit at month 7 (described in Post 01), the resolution meeting had five people in it: Fatima Idris, the program manager, Dmitri Volkov, Abebe Alemu, and Amara Okonkwo. Five. Not sixteen, not thirty. The tier model is why. Every other stakeholder was informed afterwards through the appropriate channel for their tier, but none needed to be in the room to make the decision. A decision that would have taken four to six weeks to coordinate across a sixty-person forum was made in seventy-five minutes.
A fictional case study for teaching purposes. Atlas Bank, Project Falcon and all named individuals are invented. Technologies are industry-standard and publicly available.