The right question is not which one is best, but which one will cost you least three years from now. What each family charges you once, every month, and on the day you decide to leave.
Almost everyone asks this backwards. People ask which technology is best, when the only question that matters is: which one will cost me least over three years, in money and in room to manoeuvre.
A website is not a purchase, it is a subscription in disguise. There is the sticker price, then what you pay every month without seeing it, then what you will pay the day you want to change. The three families barely differ on the first number. They differ radically on the other two.
Here are the criteria that decide, in the order they will catch up with you.
WordPress and its cousins. You install software on a host, add a theme, then plugins for every missing function. The ecosystem is vast, the skills are everywhere, and you move fast at the start.
The trade-off comes down to one word: assembly. Your site becomes the sum of thirty-odd components written by as many authors, each updating on its own schedule. As long as the assembly holds, all is well. The day it breaks, nobody owns the whole.
Webflow, Squarespace, Shopify and the rest. You manage no server, no updates, no backups. The visual editor produces clean results without a line of code, and the technical quality of the output is generally good.
The trade-off is that you rent everything: the tool, the hosting, and the right to carry on. Your content is yours, but its structure never leaves the platform. The day the price doubles or a feature you depend on disappears, you have no recourse.
The site is written for your need, with no generic layer to work around. You own the code, you choose the hosting, and nothing stops you changing supplier.
The trade-off is the entry cost and the dependence on a competent team. A badly built custom site is far worse than a well-built WordPress: it combines the high invoice with the impossibility of finding anyone willing to take the work over.
Before comparing features, ask one question: if I want to leave in two years, what do I take with me?
With an installed CMS you take the database and the files. That is a lot, and it is why these solutions remain popular. With no-code you take your text and images, but neither the structure, nor the integrations, nor the layout: the rebuild is total. With custom you take everything, provided the contract says so in writing and the code is readable.
This criterion feels remote when you start. It becomes the only one that matters the day your business changes, your volumes grow, or your supplier simply stops answering.
This is the figure discussed in meetings, and the least useful of the three. It varies mostly with design ambition and page count, not with technology. We broke it down line by line in our article on website budgets.
This is the one that catches up with you. Hosting, plugin licences, maintenance, fixes after updates, and above all the time somebody on your side spends on it. On an installed CMS this is rarely zero and rarely predictable: it depends on the health of the assembly. On a no-code platform it is predictable but rising, since it is a subscription indexed to your usage. On a well-built custom site it is low in steady state, but never zero.
Nobody prices this when choosing, and this is where the real difference lies. Work out what it would take to rebuild the same thing elsewhere. If that number approaches the price of the original site, you did not choose a technology: you signed up for a long stay.
Almost never the technology itself, but what gets piled on top. A stack of plugins loads scripts on every page, including those that need none. Uncompressed images often weigh more than everything else combined. Cheap shared hosting adds a delay before the page even starts to appear.
A well-built site is fast in all three families. A neglected one is slow in all three too.
Google does not know which technology you use and does not care. It measures loading time, visual stability, responsiveness to the first interaction, and it reads your content. The answer engines rising today add one requirement: that your page be readable and quotable without running a script.
That last point is where the three families part company a little, and it is the subject of our article on search and AI.
The attack surface is proportional to the number of third-party components. A very widespread CMS, with unevenly maintained plugins, accounts for most of the incidents we see: these are almost never targeted attacks, but bots exploiting a known flaw in a plugin nobody updated.
No-code platforms move that risk to the vendor, which is a genuine service and the strongest argument in their favour. Custom development shrinks the surface because it ships only what you need, provided its dependencies are kept current, which assumes somebody to do it.
In every case, one rule is not negotiable: automatic backups, stored somewhere other than the site's own server, and restored at least once. A backup you have never restored is not a backup, it is an intention.
Your priority is to learn fast and spend little. No-code exists for exactly this moment. Do not custom-build an idea you have not yet put in front of customers: you would be paying for the durability of a site you may throw away in six months.
You have traffic, enquiries, and limits that are starting to show. This is the right moment for a well-kept CMS or a targeted custom build, depending on how much of your business actually runs through the site. The trigger is not ambition, it is the number of workarounds you perform every week.
Customer area, booking, unusual pricing, connections to your internal tools: as soon as the site runs the business rather than presenting it, custom stops being a luxury. You are no longer paying for a shop window, you are paying for a production tool, and the criteria become those of software: reliability, room to grow, ownership.
Four signs come up again and again, and none of them is technical.
You turn down requests because "the site cannot do that". You pay monthly subscriptions for features you use ten per cent of. Every update frightens you because something breaks one time in three. And you enter the same information twice, once in the site and once in another tool.
The fourth is the most expensive, because it is paid in hours every day and never appears on an invoice.
Choosing the technology before the need. The right order is: what the site must do, then what it will have to do in two years, and only then how to build it.
Confusing price with cost. The lowest quote is often the most expensive, because it includes neither maintenance nor handover.
Stacking plugins instead of deciding. Every addition is free at the time and paid for in slowness, vulnerabilities and conflicts.
Signing without an ownership clause. Have it written down who owns the code, the content, the domain name and the credentials. That sentence is worth more than ten features.
Rebuilding the site instead of fixing what blocks. A full rebuild rarely repairs a positioning or content problem; it reproduces it with a fresh design.
No. Very well-ranked sites run on it. What hurts rankings is not the software but the pile of plugins, heavy themes and unoptimised images. A cleanly maintained WordPress performs well.
Yes, but the price varies tenfold between families. From an installed CMS, content comes across cleanly. From a no-code platform, you usually rebuild everything. Which is why it is better to estimate that cost before choosing than after.
It is for needs that justify it. A five-page brochure site gains nothing from it. A site that runs your business pays for itself within months, simply by removing manual work.
Look at them to spot what works, not to copy. You know neither their budget, nor their constraints, nor what their site actually costs them each month.
Three steps, in this order, before requesting a single quote.
First write down what the site must do, in plain sentences and without technical vocabulary: what a visitor must be able to accomplish, and what you must be able to change yourself. Then add what will change in two years, in volume and in features: that column alone eliminates half the options. Finally price all three costs for each candidate, entry, ownership and exit, over three years.
That sheet fits on one page and will save you far more than the time it takes. It also transforms the conversation with a supplier: you are no longer asking for a website, you are asking for a solution to a described problem.
If you want an outside view on that page, let's talk. We build on CMS platforms as readily as we build custom, and our first job is precisely to tell you which of the two you do not need to pay for.
Why the same "website" is quoted at $900 in one place and $80,000 in another, what you are actually paying for, honest ranges by project type, and the recurring cost almost no quote mentions.
Read full articleMost projects fail because they start with a tool instead of the bottleneck. Here is the order that holds, what each stage really costs, and how to tell which one applies to you today.
Read full articleLet's discuss your project and see how we can help you achieve your goals
Contact Us
