← Playbook hub
algomarketing
Run Book · Parts 3 & 6 of 7
Parts 3 & 6 · Operating Model & Stack · Delivery Run Book

Operating Model & Stack

Two parts that belong together: the team that runs content engineering, and the stack it runs on. Work through it to stand both up in the right order — shape the Content Services Organization, write the roles (the keystone being the modular content architect), resource it for the scale you actually have, then choose the 5-layer stack to fit and keep it running day to day.

Duration
3–5 weeksinside Phase 2 · Build
Effort
~10–15 dayslead consultant + architect
Client team
HR & leadership+ martech & content leads
Output
Team + stackorg design, roles, stack decisions
Overview

What these parts deliver, and why

Assess told the client where to start; the pipeline told them what to build. These two parts answer the next question: who runs it, and on what? We design the operating model — the Content Services Organization on three durable practices, strategy + engineering + operations — write the roles around it, resource them realistically for the client's scale, and only then select the stack to match. The order matters: the model comes before the tools, every time.

The five steps at a glance
  • 1 · Design the operating model — charter the Content Services Organization; pick the shape: centralised, hub-and-spoke, or federated.
  • 2 · Define roles & RACI — the role set, with the modular content architect as the keystone hire; clear accountability per activity.
  • 3 · Skills & resourcing plan — hire vs train vs borrow; the realistic team for mid-market vs enterprise.
  • 4 · Select the tech stack — the 5-layer reference stack, evaluation criteria, MACH / composable, build-vs-buy.
  • 5 · Operating-model governance — the Content Services council, cadences, how it runs day to day.
Start with strategy, power with engineering, wrap in operations.
the Content Services Organization in one line · engineering is the keystone, not the afterthought
1

Design the operating model

Week 1
Timebox · ~2 daysLead: ConsultantFormat: half-day workshop + write-up
Objective

Decide the shape of the content function before naming a single tool or person. Charter a Content Services Organization on the three durable practices — strategy, engineering, operations — and pick the topology that fits this client's size, brand spread and politics.

Workshop agenda (half day)
  • Map today's reality — who owns content, where it's fragmented, who's already doing engineering work without the title (30m)
  • The three practices — agree what strategy / engineering / operations each owns, and that engineering is the keystone (30m)
  • Pick the shape — centralised (one team owns it), hub-and-spoke (govern & enable at the centre, execute at the edge), or federated (standards-led, autonomous teams) (45m)
  • Decide what lives at the centre vs the spokes — model, taxonomy and governance almost always central (30m)
  • Draft the charter — mandate, decision rights, what success looks like (20m)
Inputs → outputs
Inputs
  • Org chart & current content ownership
  • Assess findings (maturity, gaps)
  • Brand / business-unit structure
Outputs
  • Operating-model canvas (template below)
  • Chosen topology + rationale
  • CSO charter (1 page)
◆ From the field

Nine times in ten the right answer is hub-and-spoke, and nine times in ten the client wants either a single empire or total autonomy. Draw the line where the model, taxonomy and governance sit at the centre and execution stays local — and put it in writing before anyone's reporting line is on the table, because that's when the conversation stops being about content.

Pick the shape — interactive
Three topologies for the Content Services Organization
Toggle between the three operating-model shapes to see how the centre relates to the edge. The centre always owns the model, taxonomy and governance; what changes is how much execution sits local.
Hub-and-spoke

Govern and enable at the centre; execute at the edge. The centre holds the model, taxonomy and standards; business-unit spokes run their own delivery within those guardrails. The default answer for most clients.

Centre owns
Model · taxonomy · governance
Edge owns
Local execution
2

Define roles & RACI

Week 1–2
Timebox · ~2 daysLead: Consultant + ArchitectOutput: role set + RACI
Objective

Turn the model into named roles and clear accountability. Write the role set as roles, not people — then map who's responsible and accountable for each content activity. The modular content architect is the keystone: ontologies, metadata, brand-as-graph, MCP-ready structured content — the role the market now needs and our defensible wedge.

The role set
  • Modular content architect ★ — the keystone. Owns the content model, taxonomy, semantic/graph layer and AI-readiness; the bridge between content strategy and the stack
  • Content engineer — implements models, reuse, metadata, workflow and tech selection; the business↔IT translator
  • Content strategist — the "CEO of content": audience, journeys, editorial plan
  • Taxonomist — the controlled vocabulary, tagging and classification
  • Information architect — structure, navigation, findability
  • Content designer — the content people actually read and use
  • Content ops manager — intake, workflow, cadence; runs the engine room
  • Knowledge-graph engineer — entities, relationships, the intelligence layer (enterprise scale)
Inputs → outputs
Inputs
  • CSO charter (Step 1)
  • Existing job descriptions
Outputs
  • Role descriptions (architect spelled out below)
  • RACI across content activities
▲ Watch out

RACI is roles, not names. Insist on at least one R and exactly one A per activity, and never let accountability be shared or delegated — "we're all accountable" means no one is. The fights happen over the A column; that's where they should happen.

The role set — interactive
Click a role to reveal what it owns
Roles split into People work — strategy, design, the human craft — and Process work — the structure, models and pipes. The modular content architect is the keystone that bridges the two.
People · strategy & craft Process · structure & engineering ★ Keystone hire
3

Skills & resourcing plan

Week 2
Timebox · ~2 daysLead: Consultant + Client HR
Objective

Decide, role by role, whether to hire, train an existing person into it, or borrow it (us, an agency, or a contractor) — and size the team honestly for the client's scale. The mid-market team and the enterprise team are different animals; don't sell one the other's org chart.

Activities
  • Skills-gap pass — map each role to the people who could grow into it vs the skills genuinely missing
  • Hire / train / borrow decision per role — be honest about which specialist skills are scarce in the market
  • Right-size for scale (see below) — mid-market wears multiple hats per person; enterprise can afford dedicated specialists
  • Phasing — which roles in the first 90 days vs which can wait for Phase 3 · Scale
Mid-market vs enterprise — the realistic team
Mid-market (lean)
  • One content engineer / architect wearing both hats
  • Strategist & ops manager shared with the wider marketing team
  • Taxonomy & graph work borrowed from us or a contractor
  • ~2–4 people who also do other things
Enterprise (specialised)
  • Dedicated modular content architect + content engineers
  • Standalone taxonomist, IA and knowledge-graph engineer
  • Content ops manager running the function full time
  • A genuine team, often hub-and-spoke across BUs
◆ From the field

Everyone wants to hire the "unicorn" content architect — the one who does ontologies, metadata, the graph, the stack and writes a clean brief. You won't find that person, or you'll find one and they'll leave in a year. Assemble the skills across two or three people and let us hold the rarest piece while they grow into it. A job ad for a unicorn is a vacancy you'll still be carrying at Christmas.

4

Select the tech stack

Week 3–4
Timebox · ~3 daysLead: Consultant + ArchitectOutput: stack decisions
Objective

Choose the stack against the model, layer by layer, using the 5-layer reference stack — Structure → Repository → Orchestration → Intelligence → Delivery. Each layer rests on the one beneath; the foundation (structure) is where content engineering lives, and nothing above it works if that base is soft. Default to MACH / composable; decide build-vs-buy per layer.

The 5-layer reference stack (foundation up)
  • 1 · Structure — content model, taxonomy, metadata, schema. The foundation; where content engineering lives
  • 2 · Repository — the source of truth: headless CMS, CCMS, DAM, PIM
  • 3 · Orchestration — content marketing / ops platform (CMP); workflow and intake
  • 4 · Intelligence — knowledge graph, vector DB, AI generation grounded via RAG
  • 5 · Delivery — DXP / composable front end, APIs, and MCP as the agent contract
The 5-layer stack — interactive
Hover or click a layer to see what runs there
Each layer rests on the one beneath. Structure is the foundation — where content engineering lives — and nothing above works if that base is soft. Vendor names are illustrative, never endorsements.
Layer 1 · Foundation
Structure
The content model, taxonomy, metadata and schema. This is where content engineering lives — build it first, build it hardest.
Tools & artifacts here
Content modelTaxonomyMetadata schemaJSON-LD / schema.org
◆ The foundation — score it first and hardest
Evaluation criteria (score per layer)
  • Fit to the content model & AI-readiness (structured fields, schema, API/MCP access)
  • MACH / composable alignment — API-first, headless, swappable, not a walled suite
  • Interoperability with the layers above and below it
  • Total cost & lock-in risk; build-vs-buy economics
  • Team fit — can the resourced team in Step 3 actually run it?
Inputs → outputs
Inputs
  • Content model & taxonomy (from Build / Pipeline)
  • Current tooling inventory & contracts
  • Resourcing plan (Step 3)
Outputs
  • 5-layer stack scorecard (template below)
  • Build-vs-buy matrix per layer
  • Recommended stack + sequencing
▲ Watch out

The shiny headless CMS is the trap. A client will fall in love with a demo and want to sign before there's a content model to put in it — and a headless system with no model underneath is just a more expensive way to store blobs. Tools are layers 2–5; layer 1 is yours to build first. Don't buy down the stack until the foundation is real.

Tools don't fix an operating model. A new platform on a broken model just gives the chaos a faster engine and a bigger licence fee.
the line for the procurement-eager sponsor · the model decides the stack, never the reverse
5

Operating-model governance

Week 4–5
Timebox · ~1.5 daysLead: Consultant + Sponsor
Objective

Give the operating model a heartbeat so it runs after we leave. Stand up the Content Services council — the standing body that holds the model, taxonomy and stack decisions — set its cadences, and wire it to the day-to-day so the function is governed, not just designed. (This is the organisational governance; the AI risk & control stack lives in Part 4.)

Activities
  • Charter the Content Services council — who sits on it, what it decides, where it can say no
  • Set cadences — weekly ops standup, monthly model & taxonomy review, quarterly stack & maturity review
  • Define decision rights — what the centre owns vs the spokes, change-control for the model
  • Wire it to delivery — intake, prioritisation and escalation paths the team actually uses
Inputs → outputs
Inputs
  • CSO charter, RACI, stack decisions
Outputs
  • Council charter & membership
  • Cadence calendar & decision-rights map
Roles & effort

RACI & effort summary

Who does what across these two parts. R Responsible · A Accountable · C Consulted · I Informed.

ActivitySponsor / LeadershipHRContent leadMartech / ITLead consultantContent architect
Operating model designACCIRC
Roles & RACICCCIRC
Skills & resourcingARCIRI
Stack selectionCICRAR
Operating-model governanceAICCRC
R Responsible — does the work A Accountable — owns the outcome (exactly one) C Consulted — two-way input I Informed — kept in the loop
WeekFocusConsultant days
Week 1Operating model design, start roles & RACI~3.5
Week 2Finish RACI, skills & resourcing plan~3
Week 3–4Stack evaluation, build-vs-buy, recommendation~4.5
Week 4–5Council & cadences, handoff into Build~2
Templates & worksheets

The artifacts you use and leave behind

Five core templates are spelled out below; the full set produced across these two parts is indexed at the end.

Template 1 · Operating-model canvas

Content Services Organization — on one page

DimensionCapture
Mandate & purposeWhy the function exists; the one outcome it owns
Three practicesWhat Strategy / Engineering / Operations each owns — engineering as the keystone
ShapeCentralised · hub-and-spoke · federated — and why this one
Centre vs edgeWhat's owned centrally (model, taxonomy, governance) vs executed locally
Decision rightsWho decides the model, the standards, the stack — and who can override
InterfacesHow it works with brand, IT, BUs and agencies
Success measuresHow the model itself is judged (ties to the Part 1 maturity re-score)
Template 2 · Role description — Modular Content Architect (keystone)

The role the market now needs

  • Mission — own the structure behind content so it's reusable, governed and AI-ready: the model, taxonomy, semantic/graph layer and MCP-ready access.
  • Owns — content model & component library · taxonomy & metadata schema · brand-as-graph / ontology · AI-readiness of the estate.
  • Partners with — strategist (what & why), content engineer (implementation), taxonomist & knowledge-graph engineer (the semantic layer), martech/IT (the stack).
  • Skills — structured-content modelling, taxonomy/ontology, schema & semantic markup, API/MCP literacy, and the rare ability to translate between content and engineering.
  • Seniority — senior IC or lead; the bridge role, not a junior tagging seat.
  • Anti-pattern — do not write this as a "unicorn" spec that also owns writing, design and the CMS admin. Split the rare skills across people (see Step 3).
Template 3 · RACI grid

Content activities × roles — roles, not names

ActivityStrategistArchitectEngineerTaxonomistOps manager
Editorial strategy & planningACIIC
Content model & componentsCARCI
Taxonomy & metadataCACRI
Workflow & intakeCICIA
Stack & toolingIRACC

Rule: ≥1 R and exactly 1 A per row; accountability is never shared or delegated. Fill against the client's actual roles, then reconcile to named people.

Template 4 · 5-layer stack evaluation scorecard

Criteria × layer — score each cell 1–5

LayerModel / AI-ready fitMACH / composableInteroperabilityCost & lock-inTeam fit
1 · Structure__/5__/5__/5__/5__/5
2 · Repository__/5__/5__/5__/5__/5
3 · Orchestration__/5__/5__/5__/5__/5
4 · Intelligence__/5__/5__/5__/5__/5
5 · Delivery__/5__/5__/5__/5__/5

Score the foundation (layer 1) first and hardest — a weak score there caps everything above it. Vendor names are illustrative, never endorsements; fit to the model wins.

Template 5 · Build-vs-buy matrix

Per capability — build, buy, or borrow?

CapabilityBuild when…Buy when…Lean
Content model & taxonomyAlways — it's yours and bespoke to the businessNever wholesale; tools support it, don't replace itBuild
Repository (CMS/DAM/PIM)Genuinely unique needs & the team to maintain itMature category, MACH options existBuy
Orchestration / CMPLightweight, tooling already covers itScale & workflow complexity justify itBuy
Intelligence (graph / RAG)Differentiating IP; in-house skill existsOff-the-shelf graph & vector tools sufficeMix
Delivery (front end / MCP)Bespoke experience or agent contractComposable DXP covers the channelsMix
Full template index for these parts
Operating-model canvas — CSO on one page (above)
Topology decision sheet — centralised / hub-spoke / federated trade-offs
CSO charter — mandate, decision rights, success measures
Role descriptions — full set; architect spelled out (above)
RACI grid — activities × roles (above)
Skills-gap & resourcing plan — hire / train / borrow per role
Mid-market vs enterprise team model — two right-sized shapes
5-layer stack scorecard — criteria × layer (above)
Build-vs-buy matrix — per capability (above)
Stack recommendation & sequencing — what to stand up, in order
Content Services council charter — membership & decision rights
Cadence calendar — ops, model, stack & maturity reviews
Done criteria

Entry & exit gates

The quality bar that says these parts are genuinely ready to start, and genuinely finished.

Before you start (entry)
  • Assess complete; entry point & roadmap phase agreed
  • Sponsor + HR/leadership engaged on team design
  • Content model & taxonomy work underway (so the stack has something to fit)
Before you finish (exit)
  • Operating model chosen, charter signed, topology agreed
  • Roles defined with a clean RACI; resourcing plan owned by HR
  • 5-layer stack decided with build-vs-buy rationale
  • Content Services council stood up with cadences running
algomarketing
Run Book · Parts 3 & 6 · Operating Model & Stack (v0.1) · the team and the tech, designed in that order. ← back to the playbook hub