Demos test the wrong thing
Every CMS demo shows an editor changing a heading and clicking save. They all do that well. The differences appear six months later, when marketing wants a new content type, or a developer needs to change a template without breaking three hundred pages.
Evaluate on the things you will do repeatedly after launch, not on the thing you will do once in a sales meeting.
Start with the content model
Before evaluating any system, write down your content types and how they relate: pages, posts, case studies, team members, services, locations, and which reference which.
Most CMS regret traces back to modelling content as pages when it was really structured data. A case study is not a page; it is a record with a client, a sector, a set of metrics and a narrative — and treating it as such is what lets you list, filter and reuse it later.
The five questions that predict regret
- Can a non-developer add a new content type? If not, every future requirement is a ticket.
- How does it handle relationships between content? Weak relational support means duplication.
- What does the editing experience look like for someone non-technical? Test with the actual person who will use it.
- How do you get your content out? Export in a usable format, or you are locked in permanently.
- Who can maintain it in three years? Availability of skills matters more for small businesses than feature lists.
Traditional, headless or hybrid
| Traditional | Headless | Hybrid | |
|---|---|---|---|
| Editor preview | Built in | Needs building | Usually good |
| Front-end freedom | Constrained by templates | Complete | Good |
| Multi-channel content | Awkward | Native | Good |
| Setup cost | Lower | Higher | Medium |
| Best for | Content-led sites | Multiple front ends | Most growing businesses |
Headless is frequently chosen for reasons that do not apply. If you have one website and no immediate plans for an app or kiosk, the extra complexity buys little.
Test with the people who will use it
Before committing, have your actual content editor build a real page in a trial. Not a demo, a real page with the awkward bits: an image gallery, a table, a call to action, a link to another content item.
Their frustration in that hour predicts three years of adoption more reliably than any feature comparison.
Plan the exit before you enter
Ask how content is exported, in what format, and whether relationships and media survive the export. Systems that make leaving difficult will eventually make you pay for that.
This is a five-minute question during evaluation and a six-month problem later if you skip it.