Guide: no-code without illusions

Choosing a tool does not start with comparing features, it starts with answering what you are building. There are three jobs, and they need different things.

1. A one-page site. One or two pages: who you are, what you do, the price, a contact form. Carrd, Tilda, Framer, Wix and WordPress with Elementor all do this. The rest of this guide mostly talks about them.

2. A site plus a database. You need a list of clients, enquiries, bookings or products, and pages built from that list. Here you assemble “a table plus a shop window”: Airtable or Notion holds the data, a builder on top displays it.

3. An app with its own logic. Client accounts, calculations, roles, payments. That is a different class of tool and a different budget, and it is exactly where no-code most often hits its ceiling.

If yours is the first job, do not read reviews written for the third. Half the disappointment in no-code comes from choosing a tool for a job you do not have.

Checked in August 2026. Free plans are built so that trying is comfortable and working is impossible.

Carrd — up to three sites, but no custom domain, no forms and a badge. Webflow — two pages and a webflow.io address. Framer — a subdomain; the custom domain is paid only. Wix — a subdomain, a banner and the platform’s own favicon. Tilda — one site with no custom domain. Squarespace has no free plan at all, only a trial.

Hence a practical conclusion: comparing platforms by their free plans is pointless. Compare them by the cheapest plan that allows a custom domain, has working forms and drops the badge. That is the real cost of entry.

A small thing people get caught by: on some platforms the badge only disappears with annual billing and stays on monthly. Check that before you pay.

The cheapest usable plan — custom domain, no badge. Prices as of August 2026, billed annually; check on the platform’s own site, they change.

Carrd — about $19 a year, the cheapest usable option for a single page. Tilda — about $120. Framer — about €120 (with a page limit on the entry plan). Squarespace — about €144 plus VAT. Webflow — about $180. Wix — about €200. WordPress with Elementor — €140–200 a year from the second year on: hosting, domain and the plugin subscription.

Separately, the renewal trap, and it catches everyone. With hosts the first year is promotional: €1 a month against €11 from the second year, €2.99 against €5.99, and the cheapest offers want three or four years paid up front. The price you see in the advert is almost never the price you will be paying.

Count two years, not one month. Over two years the gap between the “cheap” and the “expensive” option usually turns out to be half what it looked like at first.

What follows is in no English-language review, and for a micro-entrepreneur in France it weighs as much as how pleasant the editor is.

Platforms registered in the EU — Squarespace in Ireland, Framer in the Netherlands, Elementor in Cyprus — accept your VAT number, check it and charge no VAT; you account for the purchase under the reverse charge. Routine.

Platforms outside the EU — Webflow and Carrd in the US, Wix in Israel, Tilda in the UAE — count as importing a service from a third country, with its own treatment in your return. French hosts such as OVH or o2switch simply issue an invoice with French VAT, and there is nothing to work out.

A word on Tilda, because a Russian-speaking entrepreneur thinks of it first. The platform has split its regions: EU users contract with a Dubai entity, the invoice arrives in dollars and without a European VAT number. The servers, meanwhile, are in Germany and the GDPR representative is in Ireland. You can work this way, but the bookkeeping will not look like it does with a French host, and that is better known in advance than in April.

And a practical detail: not every platform accepts SEPA direct debit — some take cards only.

Choosing a platform is choosing how expensive it will be to leave it. The export question belongs before you pay, not on the day you need to go.

Wix cannot be exported as a whole, and blog posts cannot be exported at all. Importing from WordPress, on the other hand, works — the traffic is one-way by design. Framer officially offers no HTML export. Squarespace hands over a WordPress-format file, but with losses: some sections, the styling and custom code do not come along. Webflow exports code only on a paid workspace plan, and forms stop working on the exported site. Tilda exports only on its higher plan. Carrd — only on the most expensive one.

With WordPress, honestly: the files and the database are entirely yours, and moving them to another host is an ordinary procedure. But layout built in Elementor is stored inside the post as the builder’s own data, and without the builder it does not render. That is a dependency too — just on a plugin rather than a platform, and it is cheaper to get out of.

The rule is simple: what you cannot export is not yours.

This is not grumbling, it is a factor in the calculation. Over the last eighteen months Webflow, Squarespace, Framer, Airtable, Make and Zapier have all restructured their plans; some popular cheap hosting offers simply closed.

A restructure usually does not mean a five per cent rise, it means features moving between tiers. That is how multilingual support at one platform ended up in a paid add-on — for a site in four languages that changes the whole calculation. At another, once you move to a new plan you cannot go back to the old one.

Two practical consequences. Do not pay three years up front for a discount: in three years the pricing will be rewritten twice. And before changing plan, find out whether you can move back — often you cannot.

If you keep clients, bookings or orders in a spreadsheet and copy them onto the site by hand, what you need is not a second page but a database.

The combination that works looks like this: Airtable or Notion holds the data, and on top sits a tool that shows it to visitors as ordinary pages. An enquiry from a form lands straight in the table rather than in an inbox you then have to copy from.

Here is where the ceiling is: on free plans what is limited is not so much the number of records as the automations — enough for dozens of runs a month, not thousands. And storing data in an EU region is generally offered only on the expensive enterprise plans. If you keep clients’ personal data in that base, it is not a detail: you will remember from the website guide that your privacy policy has to name who you pass data to.

A European alternative exists: Baserow, openly licensed and hosted in the Netherlands, close to Airtable in how it works.

Start with the unpopular part: for a one-page site with a form and an email address, automation is not needed at all. People set it up because it is much written about, then pay a subscription for a single scenario doing what a plain email already did.

It is needed when there is a repeated action between different services: an enquiry from the site has to land in a table, then in a calendar, then in a message to the client. There are three tools: Zapier, the simplest and the dearest, Make, cheaper and easier to read, and n8n, which you can put on your own server.

They count differently — one counts tasks, one operations, one credits — and their billing models have themselves changed over the past year. Estimate by the number of runs per month, not by the price of the plan.

One widespread misconception about n8n: it is called open source, but its licence is not an open one. Installing it for your own purposes is fine; selling clients n8n hosting as a service breaches the licence. If you plan to make a living automating other people’s processes, that is something to read before rather than after.

These days a non-programmer goes here first: describe the site in words and receive it finished. You do receive something — just not what the advertising promises.

What comes out well: a draft in fifteen minutes. Structure, placeholder text, filler images, the general look. As a starting point that saves an evening.

What comes out badly: everything else. The text is generic and identical for everyone, because the model knows neither your price, nor your town, nor what makes you different. Google, meanwhile, warns explicitly about mass-produced content with no value to the reader — such pages get demoted.

The economics are not obvious either. Free plans give you something like five to seven actions a day, and a failed attempt costs the same as a successful one: the money goes not into generating but into the repair loop, where you ask for a fix and it breaks the thing next to it. Check the export separately: some of these services hand over no code on any plan.

This is the one place in this guide where the subject is not convenience but liability.

Applications assembled from a description regularly go live with an open database: large-scale sweeps of publicly published projects turn up thousands of leaked access keys and exposed personal data, down to medical records and bank details. The cause is always the same — the generator produced a working interface, and nobody configured the rules governing access to the data. Someone who is not a developer cannot check this: from the outside everything looks fine.

The practical conclusion is a hard one. A prototype built by AI can be shown. Taking client enquiries through it, let alone storing their data, cannot be done until a human who understands where the database is has checked it. Under the GDPR you are the one answerable for a leak, not the service you built the page with.

For a one-page site the risk is small: there is no database. The moment a client account or a client list appears, the rule above comes into force.

Three things will not appear by themselves on any platform, whatever the plan costs.

The legal pages. Mentions légales cannot be assembled by a generator as a matter of physics: it has nowhere to get your SIREN, your SIRET and your host’s details. A cookie policy some plugins will produce; a privacy policy almost none. What has to be on those pages is set out in the website guide.

Your offer and your structure. A template sets the look, not the order of the conversation with the client. No builder will decide for you what to say first, which price to show, and how you differ from the person down the street. That is where a month gets lost — not in choosing the platform.

Your own words. Neither a template nor an AI knows what people pay you for. It is the one part of the job that can be neither bought nor generated.

Six points to run through inside the account itself, during the trial. Twenty minutes now against a year on the wrong platform.

Can you connect your own domain on the plan you intend to pay for. Does the platform badge disappear — and does it disappear on the billing cycle you have chosen. Do the forms work and does the message reach your inbox. What exactly can be exported, and on which plan. Is your VAT number accepted, and what currency is the invoice in. Can you go back to the previous plan if the new one does not suit.

If the platform answers yes to all six, pay with a clear head. If two or more come back no, you have picked the wrong one.

What to do next: build one page and show it to a living person rather than polishing it. After this guide you will pick the platform yourself; what you will not see from the outside is your own structure and your own offer. That takes a second person — and that, rather than a list of tools, is what mentoring is paid for.