Content Management Systems

Forrester publishes two evaluations with almost identical names and almost no vendors in common.

The Forrester Wave: Content Management Systems, Q1 2025, authored by Chuck Gahun, covers thirteen providers and names Optimizely and WordPress VIP among them.

The Forrester Wave: Content Platforms, Q1 2025, authored by Cheryl McKinnon, covers twelve providers including Box, Hyland, IBM, iManage, Laserfiche, M-Files, Microsoft, NetDocuments, Newgen, and OpenText.

Both categories are about managing content. Neither list contains the other's vendors. Different analysts, different criteria, different buyers, and a naming overlap that produces genuine confusion in procurement.

Getting that distinction right is the first useful thing to establish, because the two categories solve opposite problems.

Two content markets

A content management system, in Forrester's usage, publishes content to audiences. Web pages, campaign landing pages, product content, marketing sites. The buyer is marketing or digital experience, the content is created to be seen, and the value is in how quickly a team can get something live and how well it performs once it is.

A content platform manages documents inside the organisation. Contracts, records, case files, engineering drawings. The buyer is IT, legal, or operations, the content is created to be retained and controlled, and the value is in governance, retention, retrieval, and compliance. This is the market that used to be called enterprise content management.

Publishing outward against controlling inward. Both are content, and a vendor built for one is nearly useless for the other.

The reason the confusion persists is that both markets use the word content and both involve storing files with metadata and workflow. The reason it matters is that organisations regularly shortlist across the boundary, evaluate a document management platform for a marketing website, and discover the mismatch during implementation.

Inside The Forrester Wave: Content Management Systems, Q1 2025

The evaluation scored thirteen providers, selected against inclusion criteria covering experience with enterprise clients, CMS revenue, mindshare, innovation, and momentum.

Optimizely placed as a Leader with the highest possible scores in criteria including vision, innovation, content generation and editing, and content model, taxonomy and search, alongside five more. Forrester's assessment describes it delivering on an ambition to be a comprehensive content orchestration platform spanning ideation, planning, creation, and digital delivery, and notes a roadmap extending toward agent-like interfaces capable of managing campaigns and social posts, with agentic frameworks for external interactions.

WordPress VIP was positioned by Forrester for enterprises and government agencies in North America and EMEA that need to publish at scale through an intuitive interface, with recognition for the strength of its third-party plugin and add-on community.

That community point is worth pausing on, because it is a genuine structural difference rather than a feature. A platform built on an open ecosystem inherits extensions built by thousands of agencies and developers. It also inherits their maintenance burden and their security posture, which is the trade every organisation running that stack has made knowingly or otherwise.

Forrester's earlier Q2 2023 Landscape mapped thirty two CMS vendors, and a Q3 2023 Wave preceded the current edition. The category has had regular coverage, which reflects a mature market with steady buying activity rather than an emerging one.

Headless, and what it actually costs

The architectural argument that has dominated this category for a decade is worth stating plainly, because vendors present it as a technical choice and it is an organisational one.

A traditional CMS couples content management to presentation. Authors work in an interface that resembles the published page, templates control layout, and what you see while editing is close to what ships.

A headless CMS separates them. Content is stored as structured data and delivered through an API, and a separate front end renders it. That means one content repository can serve a website, a mobile application, a kiosk, a partner portal, and now an AI system, without duplicating the content for each.

The benefit is real and the cost is consistently understated. Decoupling presentation removes the author's ability to see and control the result. A marketer who could previously adjust a layout now files a ticket with engineering, and the speed advantage the CMS was bought for moves to a different team's backlog.

The market's response has been hybrid: structured content underneath with visual editing on top. Optimizely's recognition for a visual builder alongside its SaaS CMS, and its scores in content model, taxonomy and search alongside content generation and editing, describe exactly that combination.

The practical question for a buyer is not headless or traditional. It is who will be making changes, how often, and whether they need to see what they are doing. An organisation with a strong front-end engineering team and many delivery channels benefits enormously from decoupling. One where marketing needs to ship a landing page on Thursday does not.

What a website is for now

The uncomfortable context for this category is that the traffic it was built to serve is declining.

Forrester reports B2B organic traffic down between ten and forty percent over the past year, and that buyers researching via AI are roughly one tenth as likely to click through to a site.

That does not make a CMS less necessary. It changes what it is for.

If a substantial share of buyers encounter your content as a summarised passage inside a generated answer rather than as a page they visit, then the content still has to exist, be accurate, be current, and be structured well enough to be extracted correctly. What changes is that the presentation layer, the thing a traditional CMS optimises, is doing less of the work, and the structure and accuracy of the underlying content is doing more.

Which is an argument for structured content that has nothing to do with multichannel delivery. A model extracting an answer from a page does better with clear, self-contained, well-labelled content than with a narrative that requires reading the whole page to understand. The same properties that make content reusable across channels make it extractable.

There is a second consequence that is easy to miss. When traffic falls, the metrics a CMS reports on stop describing performance. Page views declining is not evidence the content failed, and a team that responds to that decline by publishing more is treating a channel shift as a volume problem.

The agentic roadmap, and one reason for caution

Forrester's assessment of Optimizely notes a roadmap toward agent-like interfaces managing campaigns and social posts, with agentic frameworks for external interactions. Similar ambitions appear across the category.

Worth holding that against a separate Forrester prediction for 2026: that content technology vendors will roll back some of their AI agent plans.

Both statements can be true. Roadmaps describe intent, predictions describe what analysts expect to survive contact with customers. The mechanism behind the retreat is not hard to imagine, since autonomous publishing means a system making judgement calls about claims, tone, and accuracy at volumes that exceed review capacity, with the organisation's name attached to whatever ships.

The practical position for a buyer is the same one that applies whenever a category is selling ahead of itself. Evaluate what is generally available and running in production at other customers. Treat agentic roadmap items as directional rather than as capability you are purchasing.

Where the content model earns its keep

If there is one criterion in this evaluation that predicts whether a CMS implementation succeeds, it is the content model.

A content model defines what types of content exist, what fields each has, and how they relate. Get it right and content is reusable, queryable, and adaptable to channels that do not exist yet. Get it wrong and every future requirement becomes a migration.

The common failure is modelling around pages rather than around meaning. A content type called landing page with a field for hero text encodes a presentation decision permanently. A content type called product with fields for its actual attributes can be rendered as a landing page, a comparison table, a feed for a partner, or a structured answer to a question.

That work happens at implementation, is invisible in a demonstration, and is usually done by whoever is available rather than by whoever understands the content. Organisations that treat it as a technical configuration step rather than as an information architecture decision spend the following five years working around it.

Optimizely's maximum score in content model, taxonomy and search is a signal that Forrester weights this, and it deserves more attention in evaluations than it usually gets.

What to keep in view

Three things distinguish a good decision in this category from an expensive one.

Know which market you are buying in. If the requirement involves records retention, legal hold, or case management, the content platforms category is the relevant research and the CMS vendor list will not serve you. The reverse is equally true.

Decide the authoring question honestly. Headless architecture delivers genuine flexibility and moves control toward engineering. Whether that suits depends on how your teams actually work rather than on which architecture is currently fashionable.

And model content for meaning rather than for pages, because that decision outlives the platform. A well-modelled repository can be migrated to a new CMS. A repository modelled around one system's page templates has to be rebuilt.

The category itself is mature and the vendors are competent. What separates outcomes is almost entirely decided in the first three months of implementation, before anyone publishes anything.

Forrester does not endorse any vendor named here, and tier placement should not be read as a recommendation to buy.