let's talk

Custom or Platform

We build custom sites in Next.js. This is the post about when not to buy one.

Most marketing sites should not be custom-built, and the ones that should usually know it for reasons that have nothing to do with design.

Four cases where it genuinely pays, four where it does not, and what you take on the day you own the stack.

Custom website development: when it actually pays

When the site is part of a product. If the marketing pages and the application share components, a design system and a deploy, splitting them across two stacks costs more than the platform ever saved. One codebase, one set of buttons, one place a change happens.

When something has to be computed on the page. A configurator, a live price, a search across real data, anything that has to answer while the visitor waits. A platform can embed that; it cannot be that, and the seam shows.

When performance is a commercial requirement rather than a preference. High-traffic pages where a hundred milliseconds is measurable revenue, or an audience on slow connections you have actually measured rather than imagined.

And when the design does something no platform renders. Not “our brand is unique” — every brand says that. A specific technique: type that responds to the viewport in a non-linear way, a canvas or WebGL scene, a scroll built as an animation timeline rather than as reveals.

The i4B homepage: an electric car seen from above with a dotted circuit outline traced around it.

When it is vanity

“We want full control.” Control is not free — you are buying a maintenance obligation, a deploy process and a dependency on whoever wrote it. Most teams asking for control want a design a platform can already build, in the same way most teams asking for an app want a website that works properly on a phone.

“Platforms are slow.” Modern platforms are not slow. Sites are slow, usually because of an oversized hero and three forgotten scripts, and those travel with you into a custom build unchanged.

“We will need it later.” Later is a real thing and it is also the most expensive assumption in this category. If the feature is eighteen months away and undefined, build the site you need now and plan the split when the shape is known.

And the one nobody says out loud: a custom build sounds more serious in a board meeting. It is not, and the invoice is where that becomes clear.

What you take on

Marketing loses the keys unless somebody builds them back. On a platform, a non-technical editor changes a headline in thirty seconds. In a custom build that is a pull request, unless a CMS was designed in from the start, which is work that has to be quoted rather than assumed.

Dependencies age. A framework, its ecosystem and the packages around it all move, and a site nobody has touched in two years is not stable, it is out of date. Somebody has to own the upgrade, and “our developer” is not a plan if there is only one of them.

Hosting, environments, error monitoring and a deploy that somebody other than the author can run. None of this is difficult. All of it is a running cost that a platform folded into one monthly number, and it is the difference between two budget bands.

That is the honest trade, and it is not an argument against custom. It is an argument against buying it for a reason that would not survive being written down.

A software company that still needed pages

i4B builds software for electric-vehicle manufacturers. The software is the product; the site's job is to explain it and to recruit the people who build it.

That is as technical as a client gets, and the site is still pages. There is nothing on it that has to compute, nothing a visitor logs into, and no reason for the marketing team to queue behind a deploy to change a headline.

The reflex in that room is to build the site in whatever the product is built in, because it feels consistent. It is the most expensive consistency available, and the useful question is not what the company builds — it is what this particular set of pages has to do.

Questions we get asked

When should a website be custom-built instead of using a platform?

Four cases: the site shares code with a product, something on the page has to be computed live, performance is a measured commercial requirement, or the design uses a technique no platform renders. Wanting control, disliking platforms, or expecting an undefined future need are not on that list.

Is a custom Next.js website faster than Webflow?

It can be, and it usually is not by default. Both render server-side and both serve from a CDN. What decides speed is image weight, font loading and third-party scripts, and all three travel intact from one stack to the other. A custom build gives you a higher ceiling, not a faster starting point.

Can non-technical people edit a custom-built website?

Only if a CMS was designed in from the start. That is real work and it has to be in the quote. Without it, every copy change is a developer task, which is the most common and most avoidable regret in this category.

How much more does a custom website cost?

The build is typically one band higher than the platform equivalent, and the difference that matters is afterwards: hosting, environments, monitoring, dependency upgrades and someone who can deploy. Budget for the second year, not just the launch.

Get a free
audit of your
website

Speed

SEO

Animation

Design

Conversion

get my free audit
Get my website
View: CasesCases

You

Click

Drag