What marketing operations is and is not
Every marketing team has a "what": the campaigns, content and channels it runs. Marketing operations is the "how": the way a request becomes a brief, a brief becomes a build, a build is checked, approved and launched, and the results are recorded in a form someone can trust next quarter. When the how is left to habit, the team depends on a few people who remember how things are done, and every departure or busy week costs a launch. When the how is designed, the team can add people, channels and markets without the error rate climbing with them.
The function has three parts, and this hub gives each a section of its own. Process is the workflow and the procedures inside it. Systems are the places where work, assets and data live. Governance is the set of decision rights and checks that decide what may go out and who may change what. Each part has a measure: a process has cycle timeThe elapsed time from an accepted request to live work, read stage by stage to show where work waits; a process measure, not a target for individuals. and rework, a system has adoption and completeness, and governance has the share of work that passed its gates first time. Together they feed the team's reporting, because a KPIA metric tied directly to an objective. is only as reliable as the operations behind the number.
Three neighbouring disciplines are deliberately outside this hub. The design of the stack itself, which tools to buy and how they connect, belongs to the marketing technology guide. Tracking, naming conventions and data-quality diagnostics belong to the analytics root. Building and governing automated workflows belongs to the automation root. Operations is the layer that makes those three usable by a team, and this hub links to each where the boundary falls.
Process
Process is where most operations work starts, because it is where the team feels the pain. A request arrives by email, a brief is half written, creative is built against the wrong size, the tracking is added the night before launch and the campaign goes live with a mistake that a ten-minute check would have caught. The process hub starts from the core of the fix: a campaign workflow with named stages and gates, and standard operating procedures for the tasks that recur inside it.
A workflow is the route from intake to launch and the decisions taken along it: what information a request must carry before anyone starts, who signs the brief, what "ready to build" means, which reviews a piece of work must pass, and what happens after launch. The campaign workflow guide draws the route and its gates. A procedure is the written way of doing one recurring task inside that route, such as setting up a paid campaign or publishing a page, and the SOP tutorial writes one step by step. The published campaign launch QA checklist is the kind of artefact a good procedure ends in: a short, ordered list that a new starter can follow without asking. Beside the workflow and SOP guides, the mapping tutorial draws a process as it really runs before anyone changes it, and the error review tutorial traces a fixed mistake to the checks that missed it. Further guides in the section cover service levels for requests, prioritising operations work, limiting work in progress, handing over work when someone leaves and reducing key-person risk.
Process also has a measure. Cycle time, the elapsed time from request to live, shows where work waits; rework, the share of items sent back from a gate, shows where upstream stages are weak. Both are cheap to record in the work system and both move when the process improves.
Systems
Work that lives in inboxes and chat threads cannot be managed, only remembered. The systems hub covers the operational systems a team needs regardless of which tools it buys: a work management system where every piece of work has an owner, a stage and a due date; an asset library where approved creative, copy and brand files live with versions and rights; and the data operations that keep campaign names, tracking parameters and audience lists consistent across platforms.
The marketing calendar sits across all three as the single view of what is going out and when, and the work management guide explains how to build and keep it. Data operations is the quieter half. A campaign that is named one way in the ad platform, another in the analytics property and a third in the CRM cannot be reported on without manual matching, and manual matching is where numbers go wrong. The data operations guide sets the taxonomy and the routines that keep it clean, including the discipline behind every UTMURL tags identifying traffic source. and the hygiene of the audience lists that a CDPUnified first-party customer profiles. or CRM depends on. Further guides in the section cover withdrawing an asset from every channel when a licence ends or a claim stops holding, and an onboarding pack for new marketers.
What this hub does not do is choose the tools. Which work management product, which asset library and whether a customer data platform is warranted are stack decisions, and the guide to designing a marketing stack starts from use cases rather than from a vendor list. Operations decides what the systems must hold and who keeps them honest; the stack decides what runs them.
Governance
Governance answers the questions a workflow raises but cannot settle: who may approve a claim, who may spend above a threshold, who may publish to the main account, who may change the tag container and what evidence of each decision must be kept. The governance hub covers decision rights and approval matrices, brand and regulatory compliance reviews, access control and audit trails, and the operating model that decides whether operations is centralised, decentralised or a hub with spokes.
The compliance part is general direction rather than legal advice, and it is wide. Advertising standards decide what a claim may say; data protection and direct marketing rules decide who may be contacted and on what basis, which is why the operations function usually keeps the consent records and the suppression lists rather than leaving them to each campaign owner. The guidance that regulators such as the UK information commissioner publish on direct marketing is the kind of source an approval matrix should point at, so that a reviewer checks against a standard rather than a memory. The approvals and compliance reviews guide builds the matrix by risk. Further guides put it to work: a claims library and substantiation file lets routine work reuse approved claims, an approval audit checks that live work had the reviews the matrix required, a guide to handling changes after approval says which review an edit to approved work reopens, an operations charter sets out what the function owns and which decisions it takes, a change freeze holds risky changes back around peak periods, and a guide to hiring and evaluating a marketing operations consultant says which decisions and records stay with the business when an outside provider does part of the work.
Governance also reaches into the technical layer. The tag container (GTMTag management system.) has its own change rules, automated workflows have theirs, and dashboards and content have theirs; each lives in its own root. The automation governance guide is the closest neighbour, because an automated journey that sends the wrong message at scale is the most expensive governance failure a marketing team can have.
The operations maturity ladder
Most teams can place themselves on a ladder with a few rungs. The stage is not a score to chase; it tells you which problems to expect and which fix to make next. Moving one rung is usually a quarter's work, and skipping rungs rarely holds.
| Stage | Process | Systems | Governance | What it feels like |
|---|---|---|---|---|
| Ad hoc | Work arrives by any channel and is done from memory | Files on personal drives; the calendar is in someone's head | Approvals are verbal and depend on who is in the room | Launches depend on a few people; mistakes repeat |
| Documented | A written workflow and a few procedures exist | One work tool, partly adopted; a shared drive with folders | An approval list exists but is applied unevenly | Fewer surprises, but the documents drift from practice |
| Managed | Every item passes the same gates; procedures are owned and versioned | Work, assets and calendar in agreed systems; naming rules enforced at intake | An approval matrix by risk, with evidence kept | Cycle time is known; rework is visible and falling |
| Measured | Cycle time, rework and first-time-pass rates are reviewed monthly | Taxonomy shared across platforms; data checks run on a cadence | Periodic audits of access, consent and claims | Decisions about process are made from data, not anecdote |
| Optimising | Procedures are retired and rewritten on a schedule | Reporting comes straight from the work system | Governance scales to new markets and teams without redesign | Adding a channel or a market does not add risk |
The honest placement is usually lower than the team's self-image, because documentation that exists but is not followed counts as the rung below. Judge by what happened on the last few launches, not by what the wiki says.
Operations and the rest of the stack
Operations borrows from its neighbours and gives something back to each. From the stack it takes the systems it works in, and gives back the requirements: what the work tool must record, what the asset library must hold, what the integration between the ad platforms and the CRM must carry so that reporting does not depend on a spreadsheet. The marketing technology hub owns those decisions.
From analytics it takes the naming convention and the tracking plan, and gives back enforcement. The UTM and naming conventions guide defines the convention; operations is the function that makes sure every campaign leaving the workflow follows it, by building the check into intake and launch rather than trusting each owner. The marketing data quality guide diagnoses what goes wrong in the data; the operations routines in this hub are the cadence that stops the same faults recurring.
From the frameworks library it takes two structures that make reporting coherent. The measurement plan framework decides which metrics matter and who owns them, which is the input to every dashboard the operations team maintains; the KPI tree framework links the outcome a leader watches to the inputs the team controls, which is how an operations lead explains why a process metric such as cycle time deserves a place on the report. Automation, finally, is where operations design becomes unattended execution, and the governance for that belongs to the automation root.
A first operations review
A new operations lead, or a marketing lead who has just realised that nobody owns the how, can run a first review without buying anything. It takes an afternoon and produces a short list of fixes in order.
- Trace the last few campaigns backwards. From live back to the first request: where did each wait, what was redone, which gate caught the mistake and which let it through. The mapping tutorial turns the trace into a map of steps, handovers and waits.
- List every place work currently lives. Inboxes, chat, spreadsheets, the work tool, personal drives. The length of the list is the first finding.
- Collect the names of one campaign across every platform. Ad platform, analytics, CRM, the calendar. If they differ, the taxonomy comes before anything else.
- Ask who may approve what, and write down the answers. Disagreement between two people's answers is a governance gap. The charter tutorial records which of those decisions operations takes alone and which it only recommends.
- Pick the two fixes that remove most rework and give each an owner and a date. Resist fixing everything; the second review is in a quarter. The guide to prioritising operations work keeps the rest in one backlog, sorted by effect, risk and effort.
A home-insurance comparison site's first review
Mallowquay Compare, a home-insurance comparison site with a marketing team of eleven, ran its first operations review after a launch went out with the previous quarter's offer in the ad copy. Tracing the last three campaigns showed an average of 23 working days from request to live, of which 14 were spent waiting for review or for assets. Work lived in seven places, including two personal drives. The same campaign had four different names across the ad platforms, the analytics property and the CRM, so the monthly report took two days of manual matching. Nobody could say who was allowed to approve a price claim; three people believed it was them. The team chose two fixes: an intake form that refused requests without a brief and a campaign name in the agreed format, and a one-page approval matrix for claims and offers. At the next review, cycle time had fallen to 15 days and the report took an afternoon.
The three hubs below this one, process, systems and governance, each end with a table that routes a common problem to the guide that solves it. Start with whichever table names the problem the review found.
When the review has already found a specific symptom, such as campaign names that differ in every system or a monthly report that takes days, the marketing operations troubleshooting section works through its causes with a check, a fix and a prevention habit for each.
