How to compare website builders

Why feature lists mislead

Every builder lists hundreds of features and you will use perhaps a dozen. A longer list signals nothing about whether the twelve you need are any good.

Comparison tables are also written by the people selling. The rows chosen are the rows where the seller wins, which is why every table shows a different winner.

So compare on use rather than on capability. The question is not what it can do; it is what you will do and how easily you will do it.

Reviews have the same problem. Most are written in the first week of use, which is exactly the period before the difficult parts appear, and almost none are revisited a year later.

Build one real page in each

Shortlist two or three and build the same actual page in each, with your own words and your own photographs. An hour each settles more than a week of reading.

Use a page you genuinely need rather than a test. The friction only appears when the content is real and the layout has to accommodate something awkward.

Then change it. Add a section, move something, replace a photograph. Most builders are pleasant to start with and differ enormously in what editing feels like afterward.

Do it on the device you will actually use. Owners who edit from a phone between jobs have a completely different experience from the one shown in any demonstration recorded on a large screen.

Check what is hard to change later

Address structure, because changing it later costs every link you have. Our page on address structure covers what a sensible one looks like.

Export, because a platform you cannot leave is a decision you are making permanently. Ask specifically what an export contains and try it before committing.

And who owns the domain. If the builder registers it for you, confirm it is in your name and transferable. That is the detail that traps people years later.

What actually decides satisfaction

The list below is what we would test, in this order. None of it appears on a comparison table and all of it decides whether you are still content in two years.

• How long publishing a real page takes, start to finish
• How painful changing that page is a week later
• Whether the pages pass a structural check
• How the site behaves on a phone on a poor connection
• What support actually answers, and how fast
• What an export contains, tested rather than described
• The total annual cost including everything that renews

Test the output, not the editor

Publish your test page and check what it produced. Heading order, alt text, page weight, and whether the content arrives in the markup where an answer engine can read it or is assembled by a script.

Our site check reports that in one pass, and running it on a page from each shortlisted builder is the most objective comparison available.

This is where builders differ most and market least. A pleasant editor producing poor structure is a common combination and it is invisible until somebody measures.

Support, tested rather than promised

Ask a real question before buying, ideally an awkward one, and see what comes back and how quickly. That single test predicts the next two years better than any feature.

Look for whether the answer is useful or a link to a help article you had already found. The difference matters most on the day something is broken.

And check what hours support covers. A business that works weekends and a provider that answers weekdays is a mismatch that surfaces at the worst moment.

Cost over three years

Work out the total for three years including the domain, any add-ons, the renewal price and anything that becomes chargeable as the site grows.

Introductory pricing is common and the second-year figure is the real one. Ask for it specifically rather than assuming the first year continues.

Then weigh that against the hours. A builder that saves two hours a month is worth a great deal more than the price difference between most options.

Include the cost of leaving as well. A platform with no usable export has a hidden price that only appears on the day you want to go, and it is usually paid in rebuilt pages.