What this part delivers, and why
The Roadmap is the engagement timeline — the spine that sequences everything else. It answers a different question from the other run books: not "how do we model content?" or "how do we govern AI?", but "how does a consultant run the whole thing, end to end, so it lands and proves out?" Three phases — Clarity, Build, Scale & Prove — wrapped in program management: governance, stakeholder buy-in, status reporting, and the risk & decision discipline that keeps a multi-month programme on the rails.
- 1 · Mobilise & plan — stand up the programme: governance, RACI, success metrics, and the phased plan built from the Assess findings.
- 2 · Stakeholder & change management — keep sponsors and the content team bought in; manage the people side throughout.
- 3 · Run Phase 1 · Clarity (~60 days) — where the Assess diagnostic runs, plus editorial strategy and the business case.
- 4 · Run Phase 2 · Build (~90–120 days) — engineer the foundation (Pipeline, Operating, Governance run books) and pilot the AI layer.
- 5 · Run Phase 3 · Scale & Prove (ongoing) — roll out across teams/channels, deliver agentic/MCP where ready, prove ROI (Measurement run book).
- 6 · Program governance — steering cadence, status reporting, risk & decision logs, and the feedback loop that closes it.
Phase 1 · Clarity is where the Part 1 Assess diagnostic is run — it is not a second, separate diagnosis. The Roadmap is the companion to Part 1: Assess is the deep diagnostic method; Phase 1 is the slot in the timeline where you run it. Don't re-scope or duplicate the maturity scoring here — point to the Assess run book and execute it.
Click any bar or milestone to inspect the phase: its timebox, owner, what runs and what it produces. The dashed purple bars are the two continuous workstreams that span every phase.
The same journey read top-to-bottom — the progress line fills to the current phase as it scrolls into view.
Mobilise & plan
Stand up the programme so it can actually be run: agree governance and decision rights, lock a RACI across all workstreams, define the success metrics the board will judge it by, and convert the Assess findings into a phased, sequenced plan with milestones.
- Confirm the goal & the success metrics — what "proven ROI" means here (30m)
- Governance model — steering committee, cadence, decision rights (30m)
- Program RACI — sponsor, program lead, workstream leads, consultant (30m)
- Convert Assess findings → a phase plan (phase × workstream × milestone) (60m)
- Risk & decision logs stood up; status-report template agreed (20m)
- Confirm cadence, channels and the first steering date (10m)
- Assess findings & prioritised entry point
- Approved Phase 2 scope & budget
- Org chart / who's available
- Phase plan (phase × workstream × milestone)
- Program RACI & governance model
- Success-metrics baseline + status template
- Risk & decision logs (live)
The Assess readout gives you the what; mobilisation is where you turn it into a when and a who. Resist starting any build work in this fortnight — a programme that begins before the plan and governance exist spends month two unpicking month one.
Stakeholder & change management
Keep the people side healthy for the length of the programme. Content engineering changes how a team works day to day — the technical foundation only pays off if creators, editors and ops adopt it. Protect sponsor commitment, manage resistance, and make the change feel like something done with the team, not to it.
- Maintain the stakeholder map — who matters, their stance, what they need from this (revisit each phase)
- Keep the sponsor close — a standing 1:1; never let them be surprised in a steering meeting
- Bring the content team along — show early wins, explain the "why now," involve them in the model
- Surface and address resistance early — usually fear of being automated away; name it directly
- Communicate progress visibly — small, frequent proof beats one big reveal
- Stakeholder map (template below)
- Phase milestones & early wins
- Maintained sponsor commitment
- Team adoption of new ways of working
- Resistance surfaced & managed
The content team's real worry is rarely the taxonomy — it's "is this how you replace me with a model?" Answer it out loud, early, with the truth from the Governance run book: AI drafts, humans stay accountable. The people who feared the AI layer most become its best advocates once they've used it to skip the boring 60%.
Run Phase 1 · Clarity
Get to a clear, agreed picture and a signed-off plan. Clarity is the phase where the Assess diagnostic runs — the maturity scoring, content audit and gap analysis from the Assess run book — alongside the editorial strategy and the business case. This phase is diagnosis and decision, not build.
- The Assess run book (Part 1) — kickoff & scope, interviews, content audit, maturity scoring, gap & AI-readiness analysis, readout. Run it as written; don't re-invent it here.
- Editorial strategy — what content earns its place, for whom, and why — the strategy layer the engineering will serve
- Business case — cost of inaction, expected outcomes, the investment to approve Build
- Phase-1 readout & sign-off — the gate that authorises Phase 2
- Mobilised programme & plan
- The Assess run book method
- Maturity score + prioritised gaps (from Assess)
- Editorial strategy
- Approved business case
- Signed-off scope for Build
Phase 1 is not a "discovery" you can compress to a week to get to the fun part. If Clarity isn't genuinely signed off — score agreed, gaps prioritised, business case approved — Phase 2 has nothing solid to build against. A rushed Clarity is the most common reason a Build phase drifts.
Run Phase 2 · Build
Engineer the foundation the diagnosis called for, stand up the operating model and governance to run it, and pilot the AI layer on one controlled workflow. This is the heaviest phase — three run books run in parallel here, sequenced against the phase plan.
- Pipeline run book — build the content model, taxonomy and structure (Plan → Model → Structure); fix the copy-paste sprawl the audit found
- Operating run book — stand up roles (incl. the modular content architect), team topology and RACI for day-to-day delivery
- Governance run book — risk-tiering policy, human-in-the-loop approval flows, machine-readable brand rules, governed prompt library
- Pilot the AI layer — one workflow, human-in-the-loop, measured — proof before scale, never agents on a chaotic estate
- Signed-off Clarity outputs
- Pipeline / Operating / Governance run books
- Engineered content foundation (model + structure)
- Operating model & governance live
- One AI-layer pilot, measured, with results
Pick the pilot workflow for how cleanly it proves the point, not how big it is. One workflow where the structure already half-exists, run with a human in the loop, gives you a defensible result in weeks. A flagship workflow chosen for visibility tends to be the messiest — and a messy pilot teaches the client the wrong lesson about the AI layer.
Run Phase 3 · Scale & Prove
Take what the pilot proved and roll it out — across teams, content types and channels — while standing up the measurement that turns activity into evidence of ROI. This is where agentic/MCP delivery comes in where the foundation is ready, and where the engagement shifts from consultant-led to client-owned.
- Roll the engineered model and AI-layer pattern out across teams, content types and channels
- Agentic / MCP delivery where ready — expose content as data via API/MCP, scale agents on workflows that have earned it
- Measurement run book — stand up the scorecard, AI Share of Voice, and the "Can You Tell?" pass rate; report against the baseline
- Prove ROI — tie content to pipeline and outcomes, not just clicks; this is the proof the business case promised
- Transition to client ownership — capability transfer, the consultant moves to advisory
- Proven pilot & live foundation
- Measurement run book & baseline
- Rolled-out delivery across the estate
- Agentic/MCP delivery where ready
- ROI evidence vs the business case
- Client-owned capability
"Scale" tempts everyone to turn on agents everywhere at once. Scale the foundation first; let agentic/MCP delivery follow only on the workflows whose structure and governance can carry it. Scaling the AI layer ahead of the structure is just amplifying the mess, faster — the exact failure the whole engagement exists to prevent.
Program governance
Keep the programme decidable and visible. This is the management layer that runs across all three phases: a steering cadence that makes decisions, status reporting that tells the truth, logs that capture risks and decisions, and the feedback loop that feeds measurement back into the plan.
- Steering cadence — fortnightly steering committee; agenda is status, risks, decisions needed, next milestones
- Status reporting — a one-page RAG status each cycle (template below); honest amber beats false green
- Risk log — risk, owner, likelihood/impact, mitigation, status — reviewed every steering
- Decision log — what was decided, by whom, when, and why — so nothing gets re-litigated
- The loop — measurement insight (Phase 3) feeds back into the plan; the roadmap is a loop, not a line
- Phase milestones & metrics
- Status, risk & decision templates
- Steering decisions on record
- Current RAG status each cycle
- Live risk & decision logs
RACI & effort summary
Who does what across the programme. R Responsible · A Accountable · C Consulted · I Informed.
| Activity | Sponsor | Program lead | Workstream leads | Lead consultant | Content team |
|---|---|---|---|---|---|
| Mobilise & plan | A | R | C | R | I |
| Stakeholder & change | A | C | C | R | C |
| Phase 1 · Clarity | A | C | I | R | C |
| Phase 2 · Build | I | A | R | R | C |
| Phase 3 · Scale & Prove | A | R | R | C | R |
| Program governance | A | R | C | C | I |
| Phase | Focus | Indicative timeline | Consultant posture |
|---|---|---|---|
| Mobilise | Governance, RACI, metrics, phase plan | Week 0–2 | Lead |
| Phase 1 · Clarity | Assess diagnostic, editorial strategy, business case | ~60 days | Lead |
| Phase 2 · Build | Engineer foundation, operating model, governance, pilot AI layer | ~90–120 days | Lead + workstream leads |
| Phase 3 · Scale & Prove | Roll out, agentic/MCP where ready, measurement, prove ROI | Ongoing | Advisory · client-owned |
Step through each phase
Switch between phases to see who leads, what runs, and what each one must produce before the next can start.
The artifacts you use and leave behind
Five core program-management templates are spelled out below; the full set produced in this part is indexed at the end.
Phase × workstream × milestone
| Workstream | Phase 1 · Clarity | Phase 2 · Build | Phase 3 · Scale & Prove |
|---|---|---|---|
| Diagnosis & strategy | Assess diagnostic; editorial strategy; business case | — | Refresh strategy on the loop |
| Engineering | — | Content model, taxonomy, structure built | Roll out across the estate; API/MCP where ready |
| Operating model | — | Roles, topology & RACI stood up | Embedded & client-owned |
| Governance | — | Risk-tiering, HITL flows, prompt library | Agentic guardrails as agents scale |
| AI layer | — | Pilot one workflow, measured | Scale where foundation is ready |
| Measurement | Baseline success metrics | Instrument the pilot | Scorecard, AI SoV, ROI proof |
Hang dated milestones off each cell. The empty cells are deliberate — they show the client what is not happening yet, and why.
Who decides, does, and is told
- List every recurring programme activity down the left (the RACI table above is the starting set).
- Across the top: Sponsor · Program lead · Workstream leads · Lead consultant · Content team — add client-specific roles as needed.
- Exactly one A per row — accountability cannot be shared. If two people are A, the decision will stall.
- Keep R lean — one or two doers per activity; everyone else is C or I.
- Review the RACI at each phase boundary — accountability shifts from consultant-led to client-owned across the timeline.
Fortnightly RAG status
- Overall RAG — Red / Amber / Green, with one sentence of why
- Phase & milestone — where we are vs the plan
- Done since last — 3–5 bullets of what actually shipped
- Next cycle — 3–5 bullets of what's planned
- Risks & issues — top 3 from the risk log, with owner
- Decisions needed — what steering must decide this cycle
- Metrics — movement against the success baseline
Two linked registers, reviewed every steering
| Register | Capture per entry |
|---|---|
| Risk log | ID · risk · owner · likelihood × impact · mitigation · status (open/closed) |
| Decision log | ID · decision · made by · date · rationale · what it supersedes |
The decision log is the antidote to re-litigation — when a settled question comes back, you point to the entry rather than re-debating it.
Who matters, their stance, what they need
| Stakeholder | Interest / power | Current stance | What they need from this | Engagement plan |
|---|---|---|---|---|
| Executive sponsor | High / High | Champion | ROI proof for the board | Standing 1:1; no surprises |
| Content lead | High / Med | Cautious | Team not destabilised | Co-own the model; early wins |
| Content creators | Med / Low | Anxious | Reassurance on AI & roles | Involve in pilot; name the fear |
| Martech / IT | Med / High | Gatekeeper | Security & access clarity | Bring into Build early |
Revisit at each phase boundary — stances move as the programme delivers (or stalls).
Entry & exit gates
The quality bar that says the programme is genuinely ready to run, and genuinely delivering. (Each phase also has its own internal gate — these are the programme-level gates.)
- Assess findings in hand; entry point & phasing agreed
- Executive sponsor committed; program lead named
- Governance model, RACI & success metrics agreed
- Phase plan, risk & decision logs stood up
- Clarity signed off before Build started; Build before Scale
- AI-layer pilot proven before any scale-up
- Measurement live; ROI evidenced against the business case
- Capability transferred; consultant in advisory; loop running