At organizational scale, the constraint on content quality is almost never writing ability. It’s that nobody can say who owns a page, how long approval takes, or which of the four versions in circulation is the one customers actually see.
That’s the operations layer, and it’s unglamorous enough that it usually gets addressed by buying something. I’ve watched several seven-figure platform implementations land on top of an unresolved ownership problem and make it more expensive rather than less. The invoice is the easy part to explain afterwards. The two years of elapsed time aren’t.
Content ownership is the thing that’s actually broken
Ask who owns a specific product page in a large organization and you’ll usually get three answers: product marketing wrote it, the web team publishes it, and regional marketing edits it without telling either. That isn’t a disagreement about roles. It’s the absence of a decision.
Ownership has to be assigned at a level that survives reorganization, which means by content type and business area rather than by page or by person. Every content type gets a named accountable owner who decides what exists and when it’s retired, a producer who makes it, reviewers with defined scope, and a publisher with CMS access.
- Write ownership down per content type. Product pages, solution pages, help centre, blog, case studies. Assign a role rather than a name, so it survives the next reorganization instead of walking out with the person.
- Give every page an explicit review date at creation, owned by the accountable owner, and treat a missed review as a governance failure rather than a housekeeping task.
- Define who can publish without review and who can’t, then enforce it in CMS permissions rather than in a policy document nobody has read.
- Set a retirement path for every content type up front (archive, redirect or delete), because an operation that can only add produces an estate nobody can audit inside three years.
- Give regional and local teams a defined scope of what they may change without central review. A rule that forbids everything gets routed around with a separate site nobody governs, and you’ll find it during an audit two years later.
Fix the review chain rather than removing it
In a regulated industry the review chain includes legal, compliance, and often medical, security or regulatory affairs. Those reviews aren’t overhead to be eliminated. They’re the cost of operating, and treating them as an obstacle is how you get a shadow publishing process outside the CMS.
What you can fix is the shape. Sequential review, where each reviewer waits for the last, turns a five-reviewer chain into six weeks of elapsed time with about four hours of actual work inside it. Run the reviews in parallel with a single named arbiter who resolves conflicts, because the second most common delay is two reviewers disagreeing with nobody empowered to decide between them.
Then apply service levels. Each reviewer gets a fixed window (three working days is usually achievable) and silence at the end of it counts as approval for anything outside their explicit veto scope. Legal and compliance veto scope is genuinely non-negotiable and should be written down precisely, so that everything not in it can move.
Measure elapsed time from brief to publish, by content type. It’s the single most diagnostic number in content operations and almost no organization tracks it. It also happens to be the number that wins the argument, because it converts directly into campaigns that missed their quarter.
Taxonomy and metadata discipline decay silently
Taxonomy is the part that rots without anybody noticing. It gets defined at implementation, maintained for about a year, and then degrades as people add values to free-text fields because the controlled list didn’t have what they needed.
Keep the controlled vocabulary small enough that people can hold it in their heads (product, audience, funnel stage, market, language, owner, review date) and make the fields required at creation rather than promised later. Retrospective tagging projects are where content operations budgets go to die.
Appoint someone to own the vocabulary and review requests to extend it monthly. That single role is the difference between a taxonomy that still works in year three and a field with 400 distinct values, 380 of which appear exactly once. The same discipline underpins the structured data on your public pages, where Google requires markup to be a true representation of the page content. Keeping a consistent Schema.org type per content type is far easier when the CMS already knows what each page is.
What content reuse actually requires
Reuse across channels (web, email, sales decks, in-product help, partner portals) is the usual business case for an enterprise platform, and it’s a real one. It requires something most organizations haven’t done, which is to stop storing content as pages.
Structured content means a product description exists once as a field, with its own approval state and review date, and gets rendered into the web page, the datasheet and the sales deck rather than copied into each. That’s a genuine architectural change: content modelling, a headless or API-first delivery layer, and a discipline about not pasting.
It isn’t free and it isn’t always right. A campaign landing page that will run for six weeks doesn’t need modelling. A product specification that appears in nine places and has legal consequences when it’s wrong absolutely does. Model the content that’s reused and regulated, and leave the rest as pages. Organizations that try to model everything spend two years on an implementation and publish nothing during it.
What a content platform can and can’t fix
An ECM or DAM platform is good at versioning, permissions, asset management, workflow enforcement, and making it possible to find the approved version of an image. Those are real problems and the tooling solves them.
It doesn’t decide who owns a page. It doesn’t shorten a review chain with five sequential approvers, it encodes that chain and makes it harder to change. It doesn’t make anybody define a taxonomy, and if you migrate an undefined taxonomy into a new system you’ve paid to preserve it. It certainly doesn’t make the content any good, because a well-governed, correctly tagged, properly versioned page can still be written for nobody.
The spend data suggests this happens a lot. Gartner’s 2025 marketing technology survey found martech utilization down to 49%, with just 15% of organizations qualifying as high performers. Half a stack, bought and not used. That lands in a year when Gartner also reported marketing budgets flat at 7.7% of company revenue, with 39% of CMOs planning agency cuts and 39% looking to reduce labour costs. Unused licences and headcount reductions in the same budget is not a comfortable thing to present.
So sequence it the other way round. Decide ownership, fix the review shape, define the vocabulary, and run the operation manually for a quarter. You’ll then know exactly which constraints the platform needs to solve, and the implementation will take less time and cost less, because you won’t be discovering your own process during a vendor engagement.
Where to start with enterprise content operations
Pick ten pages that matter commercially and try to answer three questions for each: who owns it, when was it last reviewed, and which system holds the authoritative version. If you can’t answer all three for most of them inside a day, that’s the problem, and no procurement cycle will solve it.
Then measure brief-to-publish elapsed time on the last twenty pieces. Those two exercises take a week and will tell you more about your content operation than a platform evaluation will. They’re where I start on content operations work, before anything touches the strategy itself.