What can Google and ChatGPT actually see on your website?
Paste your address below. This runs twelve checks on the page a search engine and an AI assistant would receive — whether your text is readable without JavaScript, whether your prices and hours are marked up, whether AI crawlers are blocked, and how fast the server answers. No sign-up, no email, results on this page.
This one needs JavaScript, because it has to go and fetch your site while you wait. Everything else on this page reads fine without it — and you can always send the address in an enquiry and I will run it myself.
Type your website address first.
That address is too long to be real. Try the plain domain.
That does not look like a web address. Try something like yourbusiness.com.
Only http and https addresses can be checked.
That address points somewhere private rather than at a public website.
Nothing answered at that address. Check the spelling, and that the site is up.
The site answered with an error rather than a page. That is worth looking into on its own.
That address redirects too many times to follow.
That is enough checks for one hour. Try again later, or send an enquiry and I will look at it myself.
Something went wrong at this end. Try again in a moment.
What the check found
Each finding is what a crawler received, not an estimate. The ones marked in red are the ones costing you visitors.
—
out of 100
Secure connection
GoodNeeds workProblemServed over HTTPS.
Reachable, but not consistently over HTTPS.
Not served over HTTPS. Visitors see a "not secure" warning, and Google ranks the site lower for it.
Measured:
Browsers mark a site without HTTPS as "not secure", in front of the visitor.
Server response
GoodNeeds workProblemThe server answered quickly.
The server was slow to answer. On mobile data that is felt.
The server took over a second and a half to answer. Most visitors on a phone will have given up before anything appeared.
Measured:
How long the server took to answer, and how heavy the page is. This is not a full speed score — that needs real visitors — but a slow answer here is slow for everybody.
Built for phones
GoodNeeds workProblemThe page tells the phone how to size itself.
The mobile setting is present but unusual.
No mobile viewport tag. On a phone the page loads zoomed out and every tap target is too small.
Measured:
Most small business traffic is on a phone. Without this tag the page is rendered at desktop width and shrunk.
Page title
GoodNeeds workProblemA title of a sensible length.
The title is there but the wrong length — too short says nothing, too long gets cut off in results.
No page title at all. Google will invent one from whatever text it finds.
Measured:
The blue line in a Google result, and the words an AI assistant uses to say what your site is.
Search description
GoodNeeds workProblemA description of a sensible length.
The description is the wrong length and will be cut off or padded out by Google.
No description. Google will pick a sentence off the page, and it is rarely the one you would have chosen.
Measured:
The grey text under your Google result. It does not affect ranking; it affects whether anybody clicks.
Readable without JavaScript
GoodNeeds workProblemThe page text is in the HTML, so search engines and AI assistants can read it.
There is very little text in the HTML. Some of the page is probably being built in the browser.
Almost no readable text in the HTML. Google and every AI assistant see a blank page where your content should be.
Measured:
The single most important check here. If your text is assembled in the browser, a crawler receives an empty shell — and no amount of other optimisation survives that.
Structured data
GoodNeeds workProblemStructured data is present, and it says what kind of business this is.
There is structured data, but nothing that identifies the business itself.
No structured data. Google and AI assistants have to guess what this business is from the prose.
Measured:
The machine-readable version of what you are and what you sell. It is how you become an entity rather than a string of words.
Opening hours
GoodNeeds workProblemOpening hours are marked up so search engines and maps can read them.
No opening hours found, and nothing on the page says this business has premises. If people visit you, add them. If they do not, this one does not apply.
No machine-readable opening hours. Search and maps cannot tell anybody whether you are open.
Measured:
Marked-up hours are how "open now" works in search and in maps. Being shown as closed while you are open is the most expensive error on this list.
AI assistants allowed
GoodNeeds workProblemThe AI crawlers are allowed to read the site.
No robots.txt was found. Not a problem in itself, but nothing is being said either way.
Your robots.txt blocks AI crawlers. When somebody asks ChatGPT for a business like yours, you cannot be one of the answers.
Measured:
ChatGPT, Perplexity, Claude and Google’s AI answers each use a named crawler. Plenty of sites block them without knowing — often a setting somebody turned on years ago.
Sitemap
GoodNeeds workProblemA sitemap is published at the usual address.
Something is at the sitemap address but it is not a sitemap.
No sitemap at /sitemap.xml. Search engines have to discover your pages by following links and will miss some.
Measured:
A list of your pages for search engines, so a page is not found only if something happens to link to it.
One-tap contact
GoodNeeds workProblemThere is a number or a WhatsApp link a visitor can tap.
A contact route exists but may not be tappable.
Nothing on the page that dials or opens WhatsApp with one tap. Every enquiry has to survive somebody copying a number by hand.
Measured:
A phone number that dials, or a WhatsApp link that opens the chat. A number typed as plain text is not either of those on a phone.
Image descriptions
GoodNeeds workProblemEvery image carries a description, or is marked as decorative.
Some images have no description.
Most images have no description. They are invisible to search and to anybody using a screen reader.
Measured:
Alt text is what a screen reader announces and what Google Images ranks on. An image beside a link that already names it should carry an empty description on purpose — that counts as done here, because it is.
Questions people actually ask
Is this really free?
Yes, and there is nothing to sign up for. No email, no account, no report sent to you later. The results appear on this page and that is the whole thing.
What does the check actually do?
It fetches one page of your site the way a crawler would, plus your robots.txt and your sitemap, and reports what it received. Nothing is stored, and the request identifies itself in your logs as MabniSiteCheck.
Why does it not give me a PageSpeed score?
Because a lab score measured from one machine is not what Google ranks on — it ranks on field data from real visitors, which no tool can produce from a single fetch. This reports the server response time, which is real, and says so rather than dressing it up as something bigger.
My site scored badly. Is it beyond saving?
Almost certainly not. Most of these are settings rather than rebuilds — a missing tag, a robots.txt line, hours that were never marked up. The one that usually means a rebuild is text that is not readable without JavaScript, because that is a decision made by whatever built the site.
Want these fixed rather than listed?
Send the results with your enquiry. If your site is close, I will tell you what to change and you can do it yourself. If it is not, I will price a rebuild.
Send an enquiry