Free tool

What does your website look like in the first four seconds?

A first-time visitor decides almost nothing about your work and almost everything about whether to bother. Paste a URL and get a score out of 100 across nine fixed checks, each one pass or fail, each one with the fix in a single line. No sign-up, no email, nothing stored.

Test a page

We fetch the page once, score it against nine fixed rules, and forget it. No account, no email, nothing stored.

The nine checks

The first screen says what you do

A heading that is your business name tells a stranger nothing. It has to name the service and, for a local business, the place.

There is an obvious way to book or ask

One clear action near the top. If the only route is a contact page three clicks away, most people never take it.

The phone number is tappable on a phone

Most local traffic is on a phone. A number that is not a tel: link is a number nobody dials.

A real business is visible behind the site

Address, phone, hours, email, a social profile. Buyers check that you exist before they call.

Getting in touch takes no hunting

A contact route in the header, not only in the footer. Friction here costs more than anything else on the page.

The page is not heavy before it starts

The document itself and how many scripts and stylesheets it pulls in before anything can render.

Nothing on the page looks abandoned

A stale copyright year or leftover placeholder copy reads as a business that has closed.

Someone other than you vouches for you

Reviews, testimonials or case studies. Your own claims are not evidence.

The phone number is text, not a picture

A number baked into an image cannot be tapped, copied or read by a search engine.

How the score works, and what it is not

Every check is a rule, not a judgement. The page gets fetched once, each rule either matches or it does not, and the weights add up to 100. No model is involved anywhere in the path, which is the whole reason the result is stable: run it twice on an unchanged page and you get the same number twice.

What it deliberately does not do is grade your design, your copy or your photography. Those matter enormously and they are not testable by rule. This measures the floor: the things that lose you a visitor for reasons that have nothing to do with how good you are at your job.

One honest limitation, stated rather than buried. The test fetches the document, so it sees your page the way a crawler or a slow connection sees it first. A site that paints its phone number in with JavaScript will score worse here than it looks in your browser. We think that is the more useful measurement, but you should know that is the choice being made.

What a buyer is actually judging

The written companion to this tool, the four second test, goes through what a visitor decides before they scroll and why the order of those judgements is not the order most businesses build for.

We also ran these nine checks across a real sample of local business sites and published the distribution, including the checks almost nobody passes: what fifty local websites scored.

If the page is fine and the problem is what happens after someone gets in touch, that is a different job. The automation scanner ranks which one to fix first.

The companion guide

What a buyer decides about your website before they scroll

Where this applies

AI systems for contractors and trades who are on a roof when the phone rings

AI systems for studios where the booking happens in the DMs

AI systems for studios and clinics where an empty slot is revenue that never comes back

Website test FAQ

What exactly does it check?
Nine things a first-time visitor judges before they scroll: whether the first screen says what you do, whether there is an obvious way to book or ask, whether the phone number is tappable, whether a real business is visible, how hard it is to make contact, how heavy the page is, whether anything looks abandoned, whether anyone vouches for you, and whether the phone number is selectable text rather than an image.
Is there AI involved in the score?
None. Every check is a fixed rule over the fetched page, which is deliberate: rules give the same page the same score every time and there is nothing for a model to invent. It also means the sample we published in the blog post can be reproduced by anyone.
Do you store the URL I test?
No. The page is fetched, scored and forgotten. There is no database behind this, no log of submitted addresses and no email capture. The only thing that outlives your request is a short-lived counter that stops one person running hundreds of tests.
How is the score calculated?
Each check carries a weight and the weights add up to 100. Offer clarity and a visible booking path are worth the most at 14 each, and the phone-as-text check is worth the least at 6, because it is a symptom of a bigger problem rather than the problem itself.
My site scored badly and I think it is wrong.
Read the detail line under each failed check, because it says what was actually found. The common honest disagreement is a site that renders its content with JavaScript: this fetches the page the way a first-time crawler sees it, and if your phone number only appears after a script runs, that is a real finding rather than a bug in the test.
It said my page could not be read.
Usually a bot protection layer refusing an unfamiliar visitor, a redirect chain longer than three hops, or a page that took more than eight seconds. All three are worth knowing about on their own.
Does a perfect score mean the site is good?
No. It means the site clears the floor. These are hygiene checks, not design or copy judgement, and a page can pass all nine and still fail to persuade anyone. What a low score reliably tells you is that you are losing people for reasons that have nothing to do with your business.