Most CMS decisions are made backwards. Someone picks the platform first, usually because their last agency used it or because a competitor’s site looks good, and then the brief gets bent to fit whatever that platform does well.

The order should be the other way around. Decide what the site has to do, who has to edit it, and how long it needs to last. The platform then narrows itself down to one or two credible options, and the argument mostly disappears.

This piece sets out the four realistic choices for a South African marketing site in 2026, what each one actually costs to run, and the specific situations where we would recommend each.

First, answer four questions

Before comparing platforms, get honest answers to these. They decide the outcome more than any feature list does.

Who edits the site, and how often? A site edited weekly by a marketing coordinator needs a different tool than one edited twice a year by whoever remembers the password. Frequent editing by non-technical staff pushes you toward visual editors. Rare editing pushes you toward whatever is cheapest and most stable.

How many page types are there? Five bespoke pages is a different problem to 300 product pages, a resource library and a multilingual careers section. Structured, repeating content is what content management systems are actually for. If your site is five handcrafted pages, you may not need a CMS at all.

What does it need to talk to? CRM, booking, payments, an inventory feed, a member database. Integration requirements eliminate more platforms than design requirements ever do.

How long does this build need to last? A three-year horizon and a ten-year horizon justify very different amounts of upfront investment. Most marketing sites get replaced every three to five years, and pretending otherwise leads to over-engineering.

WordPress

Still the default, still running a large share of the web, and still the right answer more often than the design industry likes to admit.

Where it wins. Enormous plugin ecosystem, so almost any requirement has an off-the-shelf answer. Deep local support – you can hire a WordPress developer in Cape Town, George or Johannesburg tomorrow. Clients often already know the admin interface. Good multilingual and e-commerce paths if you need them later.

Where it hurts. Maintenance is real and ongoing. Core updates, plugin updates, security patching, backups. A neglected WordPress site is a security liability, not just a slow one. Plugin sprawl also degrades performance quietly – ten plugins each adding a stylesheet turns a fast page into a three-second one.

Running costs in South Africa. Decent managed hosting from a local provider runs roughly R150 to R600 a month for a marketing site. Premium plugins and a theme licence add maybe R3,000 to R8,000 a year. Budget a further R1,500 to R4,000 a month if you want someone actively maintaining it rather than reacting when it breaks.

Pick it when you need a large content operation, you have or can hire technical support, or your requirements are common enough that plugins genuinely solve them.

Webflow

The visual builder that grew up. Real CSS output, a proper CMS layer, hosting included.

Where it wins. Marketing teams can edit and even build new pages without a developer, once the system is set up properly. Design fidelity is excellent – what the designer draws is what ships. Hosting, CDN and SSL are bundled, so there is no separate maintenance contract. No plugin security surface.

Where it hurts. The pricing is in US dollars, which for a South African business means the cost moves with the rand. Content structure limits exist and you will find them if your content model is complex. Integrations beyond the common ones often need a third-party connector, which becomes another subscription. And you are renting the platform – migrating away later is a real project, not an export.

Running costs. A CMS site plan plus a workspace seat lands somewhere around R700 to R1,500 a month at 2026 exchange rates, before any connectors. Predictable, but not cheap over five years, and exposed to currency movement.

Pick it when design quality matters more than integration depth, the marketing team wants autonomy, and the content model is moderate rather than sprawling.

Squarespace and the hosted builders

Worth naming rather than dismissing. For a small business with a handful of pages and no in-house technical capacity, a hosted builder is frequently the correct commercial decision.

Where it wins. Cheapest total cost of ownership by a distance. Nothing to maintain. Reliable, fast enough, and the templates are no longer embarrassing.

Where it hurts. You will hit a ceiling on design and structure, and when you hit it there is no way through. Custom functionality is largely off the table.

Pick it when the site is genuinely simple, budget is tight, and the money is better spent on brand or content than on a bespoke build. We have told clients to use Squarespace and put the saved budget into photography, and it was the right call. If your budget is under R50,000 in total, this is usually where we point you – the same logic we set out in what a brand refresh costs in South Africa.

Headless and static site generators

A content API (Sanity, Contentful, Storyblok, or a git-based setup) feeding a front end built in Astro, Next.js or similar, deployed to a global edge network.

Where it wins. Performance is in a different class – pages are pre-rendered and served from an edge location, so a visitor in Durban and a visitor in London both get a fast page. Total design freedom. Security surface is close to zero because there is no database to attack on the public site. Content can feed a website, an app and a screen in a showroom from one source. Hosting is often free or close to it.

Where it hurts. It needs a developer. Not for content editing, but for structural change – a new page type is a code change, not a click. The initial build costs more, typically 20 to 40 percent above an equivalent WordPress build. And if your developer relationship ends badly, you need someone who knows that specific stack.

Running costs. Hosting frequently under R200 a month, sometimes zero. The CMS layer runs from free tiers up to around R800 a month for a small team. The real cost is developer availability, not infrastructure.

Pick it when performance matters commercially, you have a long-term technical partner, or the content genuinely needs to serve more than one destination.

The honest comparison

Strip away the platform loyalty and it comes down to a trade between three things: how much you can edit yourself, how much you pay every month, and how far you can go before you hit a wall.

Hosted builders – highest editing ease, lowest cost, lowest ceiling.
WordPress – high editing ease, moderate cost, high ceiling, ongoing maintenance obligation.
Webflow – high editing ease for pages, moderate to high cost, moderate ceiling, no maintenance.
Headless – moderate editing ease, low running cost, highest ceiling, needs a developer.

No row in that list is wrong. The mistake is picking a row that does not match how your organisation actually works. A headless build handed to a company with no technical partner becomes a frozen site nobody can change. A WordPress site handed to a company with no maintenance budget becomes a security incident.

We build most client marketing sites on a static or headless stack, and we are open about the reason: for the kind of work we do, the performance and security wins are worth the developer dependency, and we stay on as that developer. This site runs on exactly that setup.

But we do not recommend it universally. In the past year we have specified WordPress for a client with a large editorial team and an existing in-house developer, and Squarespace for a two-person consultancy who needed a credible site and had better uses for the rest of the budget. Recommending the platform we like best regardless of fit would be order-taking with extra steps.

The questions that actually decide it are the four at the top of this article. If you can answer those clearly, any competent studio will arrive at roughly the same shortlist.

One thing to avoid

Do not choose a platform before you have agreed the content model. The number of page types, how they relate, and what repeats – that structure is the real architecture of a website, and it is far more expensive to change after the build than before it. We do the content model in week one of every project, before a single design decision, for exactly this reason. It is the same discipline that makes a project run to schedule rather than slipping, which we covered in how long a website project actually takes.

Get the model right and the platform question mostly answers itself.

If you are weighing up a platform for an upcoming build and want a second opinion that is not tied to selling you a particular stack, send a short note with what the site needs to do and who will be editing it. We will tell you which way we would go, even when the answer is not us.