06-reference

commoncog executing on becoming data driven

Tue Apr 14 2026 20:00:00 GMT-0400 (Eastern Daylight Time) ·reference ·source: Commoncog ·by Cedric Chin

“Executing on Becoming Data Driven: The Politics” — @CedricChin

Why this is in the vault

This is the “Politics” companion to Chin’s Becoming Data Driven first-principles essay — where that piece defines what the discipline is, this one explains why installing it is mostly a change-management problem, not a technical one. Directly relevant to RDCO’s agent-deployer positioning and how we scope phData/MG-style engagements: the Mulally/Ford BPR case is a pre-canned sales narrative for why tooling-alone engagements fail.

The core argument (paraphrased)

Every analytics project has two parts — the technical build, and getting the organization to actually use the data — and the second part is the hard one.

Chin’s framing: instrumentation, pipelines, warehouses, dashboards — “relatively straightforward”. He concedes “straightforward does not mean easy”, but the technical piece is tractable engineering with well-known patterns. The industry nevertheless obsesses over it: vendors, thought leaders, LinkedIn screeds, conference talks almost exclusively cover the stack.

The part that actually breaks projects is organizational: getting executives and line managers to use the data to gain knowledge (Deming-sense causal models that update against reality). That requires culture change. And on culture change, Chin quotes a data leader whose assertion has haunted him:

“Either the CEO leads the charge… or every leader who isn’t data driven gets kicked out.”

Chin has searched for counter-examples and found none. Every successful “became data driven” transformation took one of those two paths. No third road. No bottom-up grassroots victory. No “we built a great dashboard and people came around.”

The Mulally/Ford case study (the full piece’s centerpiece). When Alan Mulally took over Ford from Boeing, the board told him he could fire anyone. He declined — not out of softness, but because he knew his Business Plan Review (BPR) mechanism would do the self-selection for him. The BPR was a weekly, mandatory, executive-attended meeting where each leader personally presented their own numbers. No subalterns answering for them. No side discussions. No phones. No jokes at others’ expense. “The data sets you free”, Mulally told them, smiling.

The mechanism did three things simultaneously:

  1. Extreme accountability with no hiding place. Executives had to know their own numbers. Those who couldn’t — or wouldn’t stop lying — embarrassed themselves publicly until they self-selected out. One exec tried to skip the meetings citing Asia factory work; Mulally replied “you can’t be part of the Ford team if you don’t” — that exec was gone shortly after.
  2. Forced cross-functional visibility. Every executive saw every other function’s reality every week. Side-channel politics (“where the real business of the company was conducted”) collapsed because the BPR surfaced everything.
  3. Reward-structure inversion. In old Ford, “truths too painful to put in PowerPoint” were whispered in hallways. In Mulally’s Ford, the reward for telling the truth exceeded the reward for lying. The culture flipped within months — technical data cleanup (Quality took years to get clean global metrics) continued in parallel, but the behavioral shift happened first.

Chin’s takeaway: the BPR is the change-management mechanism. The technical work (clean metrics, dashboards, pipelines) is necessary but insufficient. Without Mulally in that room setting the norms, “all the data work Ford could have done would’ve been for nothing.”

The implicit consulting lesson: if neither condition is present at a prospective client — no executive running the charge, no willingness to let non-believers self-select out — the engagement will fail regardless of technical excellence. Most startup operators, Chin notes, are “too busy with the day-to-day grind” to fight prevailing cultures in their companies. That’s the trap.

Mapping against Ray Data Co

Mapping strength: strong. This is the posture-defining article for how RDCO qualifies and scopes client engagements, and the Mulally story is a ready-made narrative weapon for the sales conversation.

1. Qualification gate for MAC + agent-deployer engagements. Chin’s two-paths rule is a qualification filter. Before RDCO takes a phData/MG-style engagement to install MAC (Model Acceptance Criteria) and the agent-deployer role, we should diagnose which path exists: (a) executive sponsor actively leading the shift, or (b) willingness to replace non-aligned leaders. If neither, we’d be installing tools that rot when we leave. Add to the ../04-tooling/rdco-state-ownership-architecture qualification checklist as a hard gate, not a soft preference.

2. “Technical part is the easy part” is the thesis behind state-ownership. RDCO’s bet that the client owns vault + skills + MAC matrix — while models remain commodities — echoes Chin’s point: the technical stack is not where the leverage lives. The leverage is in the organizational discipline that uses the stack. State-ownership is RDCO’s mechanism for persisting that discipline: the vault IS the culture artifact. Without it, the client reverts to pre-engagement behavior within months of our exit — the same pattern Chin warns about.

3. MAC is the BPR for the AI era — and needs Mulally-equivalent executive cover. The MAC severity tiers (Stop/Pause/Go) and the expectation that an operator personally owns their model’s behavior map 1:1 onto Mulally’s BPR: each exec presents their own numbers; no hiding place; embarrassment-driven self-selection for leaders who refuse discipline. The agent-deployer role per 2026-04-14-levie-agent-deployer-role-jd is the MAC-equivalent of Mulally’s BPR chair. But — critically — an agent-deployer hired three layers down with no executive air cover is exactly the “third road” Chin says doesn’t exist. RDCO’s placement playbook must insist on exec sponsorship or decline the engagement.

4. Coaching > tooling in the Playbook + Coaching posture. Posture 2 (hand over pre-packaged practice + coach 3-6 months) needs to weight coaching HIGHER than tooling delivery. Chin’s argument — and the Ford case — implies 70%+ of engagement value is in the change-management work with leadership, not MAC matrix artifacts. Engagement pricing should reflect this: the coaching hours are the moat. Risk: we sell what’s legible (MAC installation, harness setup) instead of what works (weekly exec-attended meetings where the agent-deployer presents reality unflinchingly).

5. The Mulally/Ford case is a ready sales narrative. Chin’s extended BPR excerpt from American Icon is a pre-packaged analogy RDCO can use verbatim in sales conversations: “Here’s what a successful ‘became data driven’ transformation actually looked like — weekly meetings, no phones, no hiding, self-selection. MAC is the same mechanism for AI. If your CEO won’t run this, don’t hire us.” Concrete, narrative, memorable — worth extracting into a sales-asset in ../02-business.

One warning this article plants for RDCO: we should resist the consultant’s temptation to sell the technical build (MAC installation, harness setup, agent-deployer tooling) as the deliverable. Chin’s entire essay says that’s the part clients can get anywhere, and it’s not what determines success. The real deliverable is the culture shift — harder to scope, harder to price, harder to prove — but the only thing that works. Mulally didn’t sell Ford a dashboard; he sold them a weekly meeting with enforced honesty. RDCO should sell the MAC equivalent.