· 3 min read
A website check should say what it could not read
When a tool can't read part of your site, the honest result is couldn't check. Calling it a fail, or folding it into one score, mixes findings with guesses.
A website check should mark any item it could not read. When a site blocks the tool, the honest result is couldn't check. Counting that as a fail, or folding it into one score, mixes real findings with guesses. SignalGrade, the free homepage check from Gman Creations, is built this way.
Why a check gets blocked
Some firewalls block automated tools. That is a setting on the site. It says nothing about the page itself. A check that hits one has learned nothing about your title, your headings or your business details.
What one number hides
Some website checks hand you a single score. A single score adds up items that may not have been checked. From the total alone, you can't tell which parts came from your page and which parts were estimated.
That matters when you decide what to work on first. A low number might point at a real gap. It might also point at a page the tool never read.
How SignalGrade sorts results
SignalGrade checks whether search engines and AI tools can read your homepage basics. It puts each basic in one of three groups. Looks fine, needs attention, or couldn't check.
Each result shows what was read, why it matters and what to do next, with a link to the source. The results are not added up into a grade. When a site blocks the check, the report says so and says why. It does not score that as a failure.
What it reads
It reads three public addresses once, when you ask. Your homepage, robots.txt and llms.txt. Scripts aren't run, so you see the page the way tools that don't run scripts see it.
It looks at seven basics. The page title and description. One main heading and section headings. Structured data that names the business. Question headings and FAQs. The viewport setting for phones. The noindex and canonical settings. The crawler rules in robots.txt.
llms.txt is a newer, optional file. It is reported on its own and never counted against you.
What it does not do
It does not ask AI assistants about you, track rankings, crawl the whole site or check that your facts are true. No one can promise a spot in search results or AI answers, and SignalGrade does not try to.
What to do with a couldn't check
Treat it as an item still open. The report says why it could not read it. A person can then look at that item by hand on the live site.
What to do next
The check is free and needs no account. When you want a person to confirm the findings, the SignalGrade Audit is $497. A person checks your report against the live site and confirms each finding. Then they write a fix brief, most important first, for whoever builds your site. Talk it through with us first, and we will say if it fits: cal.com/shohgman/discovery.
The next time a tool hands you a number, look for the items it never read.
FAQ
Is couldn't check the same as a fail?
No. It means the tool could not read that item. A firewall that blocks automated tools is one cause. SignalGrade marks the item, says why, and does not count it against you.
Why doesn't SignalGrade give a score out of 100?
A single score adds up items that may not have been checked. SignalGrade shows each basic on its own and marks anything it could not read.
What does SignalGrade read?
Your homepage, robots.txt and llms.txt, once, when you ask. It does not run scripts.
Will getting these basics right get my business into AI answers?
No one can promise that. AI tools and search engines decide their own answers. SignalGrade checks the basics on your site that those tools can read.