How easy is your website for AI search to read?
Paste the address. We check structured data, machine readability, content format, technical setup, entity and language, and show what pulls the score down and how to fix it.
Free. No email, no account.
How we measure
21 checks in 6 categories. Each category has a fixed weight, and the weights add up to 100. A check that does not apply to your website, such as opening hours for an online shop, is left out of the calculation instead of counting against you.
Structured data 25 points
- Readable structured data7
We check that every block of structured data (JSON-LD) on the page can be read without errors.
- Business type8
We check whether the structured data says what kind of business this is, such as a hair salon, plumber or restaurant.
- Name, address and phone in structured data10
We check whether the business has a name, a full postal address and a phone number in the structured data.
Machine readability 15 points
- AI crawlers are allowed5
We read robots.txt and check whether the crawlers that fetch pages for AI answers and search may read this page.
- llms.txt5
We check whether the site has an llms.txt, a short summary written for language models, in the right format.
- Sitemap5
We check whether the site has a valid sitemap that crawlers can find, and whether this page is listed in it.
Content format 20 points
- Questions and answers8
We look for questions with full answers in the visible text, or questions and answers marked up as an FAQ in the structured data.
- Visible address4
We check whether the address appears in the text on the page, not just hidden in the code.
- Opening hours4
We look for weekdays with times in the text, or opening hours in the structured data.
- Services and offers4
We look for services or offers in the structured data, or a heading or menu link that says what you offer.
Technical 20 points
- Text in the HTML6
We count how many words are in the HTML the server sends, before any JavaScript has run.
- Response time5
We measure how long it takes before the server starts to respond, once, from our server in Europe.
- HTTPS5
We check whether the page ends up on a secure https address.
- Mobile-friendly4
We check whether the page declares that it is built for screens of every size (the viewport tag).
Entity 10 points
- Organisation number4
We look for a valid Norwegian organisation number on the page and check that it is active in the Brønnøysund Register.
- Name as registered3
We compare the name in the Brønnøysund Register with the name the page uses for itself.
- Address as registered3
We compare the postcode and street name on the page with the addresses in the Brønnøysund Register.
Language and meta 10 points
- Page title3
We check whether the page has a title of a length that shows in full in search results.
- Meta description3
We check whether the page has a description of a length that fits in search results.
- Language tag2
We check whether the page states a valid language, and whether it matches the language version the page claims to be.
- Language versions2
We check whether the language versions of the page link to each other, including to themselves (hreflang).
The check follows fixed rules, not an AI model. Entity is only assessed when the website looks Norwegian (a Norwegian domain, Norwegian language or an organisation number), and only when the Brønnøysund Register Centre responds: otherwise the score is calculated from the other categories, so a site is never marked down because a register did not answer.
About our crawler
The check is fetched by AventlyBot. It identifies itself with this user agent: AventlyBot/1.0 (+https://avently.no/ai-sjekk)
It reads robots.txt first and respects it. If robots.txt blocks AventlyBot or all crawlers (*) for the page, we do not fetch the page, and the result says the site is blocked. A page marked noindex gets no score either.
It only fetches the address you enter, plus robots.txt, llms.txt and the sitemap. It sends no cookies, and we limit how often the same website can be checked.
To block it, add this to robots.txt:
User-agent: AventlyBot
Disallow: /Example of a result
The website below is fictional: the business, the address and the domain are made up. The result is calculated by the same engine that checks your website.
kvernvold-ror.example
75 out of 100 for how easy the page is for AI search and search engines to read and understand.
The most important things to fix
llms.txt is missing
Could add 5 pointsAn llms.txt is a proposed standard that gives language models a short, tidy overview of the site and its most important pages.
How Avently handles itAvently generates an llms.txt for every site automatically, in all the site's languages.
Only one question, or a link to the FAQ page
Could add 4 pointsPeople ask AI services in full sentences, and a clear question with a direct answer is easy to pick out.
How Avently handles itThe FAQ section in Avently shows the answers as regular text and marks them up as an FAQ in the structured data.
The organisation number on the page is not valid
Could add 4 pointsAn organisation number that matches the Brønnøysund Register makes it clear which business the page belongs to.
How Avently handles itAvently writes the organisation number in the footer and in the structured data.
The six categories
- Structured data23 of 25
- Machine readability9 of 15
- Content format16 of 20
- Technical18 of 20
- Entity0 of 10
- Language and meta8.8 of 10
All findings
Missingllms.txt is missing
0/5
An llms.txt is a proposed standard that gives language models a short, tidy overview of the site and its most important pages.
How Avently handles itAvently generates an llms.txt for every site automatically, in all the site's languages.
PartialOnly one question, or a link to the FAQ page
4/8
People ask AI services in full sentences, and a clear question with a direct answer is easy to pick out.
How Avently handles itThe FAQ section in Avently shows the answers as regular text and marks them up as an FAQ in the structured data.
MissingThe organisation number on the page is not valid
0/4
An organisation number that matches the Brønnøysund Register makes it clear which business the page belongs to.
How Avently handles itAvently writes the organisation number in the footer and in the structured data.
MissingCould not be compared: the business was not found in the register
0/3
The same name on the page and in public registers makes it easy to see that it is the same business.
How Avently handles itWhen Avently builds the site, the name and organisation number are taken from the Brønnøysund Register.
MissingCould not be compared: the business was not found in the register
0/3
The same address on the page and in the register makes it easy to see that it is the same business, and if they differ it is worth checking whether the register is out of date.
How Avently handles itWhen Avently builds the site, the address from the Brønnøysund Register is included, and you can correct it in the settings.
PartialSome of the name, address and phone is missing
8/10
When the contact details are machine-readable, AI services do not have to guess them from the text or take them from other sources.
How Avently handles itAvently adds the name, address and phone from your settings to the structured data on every page.
PartialThe server took 1.3 s to respond
3/5
Crawlers that fetch pages while someone waits for an answer have short time limits, and a slow server may be skipped.
How Avently handles itPublished Avently pages are cached on Vercel's network, so they respond quickly.
PartialThe description is 177 characters
2/3
A good description says briefly what the page is about, and is often used when the page appears in search and is shared.
How Avently handles itAvently gives every page its own description.
PartialOnly crawlers that collect training data are blocked
4/5
A page the crawlers may not read is a page they cannot take content from when they build answers.
How Avently handles itThe robots.txt on Avently sites explicitly lets AI crawlers in.
In placeStructured data can be read
7/7
Structured data tells search engines and AI services directly who you are, and a block with errors cannot be read at all.
How Avently handles itAvently generates JSON-LD automatically from your settings and adds it to every page.
In placeThe business type is stated
8/8
A precise type states clearly what kind of business you are, so AI services do not have to guess from the text.
How Avently handles itAvently uses the business type you choose in the settings as the type in the structured data.
In placeThe sitemap is in place and lists this page
5/5
A sitemap tells crawlers which pages exist, including those that are hard to find through links.
How Avently handles itAvently generates a sitemap of every published page, and on your own domain the site's robots.txt points to it.
In placeThe address is on the page
4/4
When the address is visible and matches the structured data, people and machines get the same information.
How Avently handles itAvently shows the address from your settings in the footer of every page.
In placeOpening hours are in place
4/4
“Are they open now?” can only be answered correctly when the opening hours are written somewhere readable.
How Avently handles itAvently shows the opening hours in the footer and adds them to the structured data.
In placeThe services are clear
4/4
For the site to come up when someone asks for a service, it has to say clearly that you offer it.
How Avently handles itAvently has dedicated sections for services, prices and products, and you choose the heading yourself, such as “Services” or “Prices”.
In placeThe text is in the HTML
6/6
Several AI crawlers do not run JavaScript, so they only see the text that is in the HTML.
How Avently handles itAvently pages are built on the server, so all the text is in the HTML the crawlers fetch.
In placeThe page uses HTTPS
5/5
Browsers mark pages without HTTPS as “not secure”, and that is the first thing visitors see.
How Avently handles itEvery Avently site gets HTTPS automatically, including on its own domain.
In placeThe page is mobile-friendly
4/4
Search engines assess the mobile version of the page first, and without the viewport tag the page shows as a shrunken desktop page.
How Avently handles itAvently pages are mobile-friendly and get the viewport tag automatically.
In placeThe title has a good length
3/3
The title appears in search results and browser tabs, and is often the first thing read about the page.
How Avently handles itAvently gives every page its own title.
In placeThe language is tagged correctly
2/2
The language tag tells search engines, screen readers and AI services which language the text is in.
How Avently handles itAvently tags the content of every page with the language of its language version.
Not relevantLanguage versions
–
With links between the language versions, search engines and AI services know which version to show to whom.
How Avently handles itAvently links the language versions to each other with hreflang, including to the page itself.
About the result
Calculated by ai-sjekk-1, the same engine that checks your website. The business and the address are fictional.
Questions about the AI check
What does the AI check measure?
We fetch the page you enter and go through 21 checks in 6 categories: structured data, machine readability, content format, technical setup, who is behind the site (entity) and language and metadata. Each category has a fixed weight, and together they make a score from 0 to 100.
Does the check use AI to assess the site?
No. The check follows fixed rules, described on this page, so the same page is always assessed by the same rules. What can vary from one check to the next is how fast the page responds, and whether the Brønnøysund Register, llms.txt and the sitemap respond in time.
Does a high score mean AI services will mention me?
Not necessarily. The score says how easy it is for search engines and AI services to read and understand the website. Which sources a service chooses to show or mention is up to the service, and nobody can promise that.
Why did my website get no score?
If robots.txt blocks our crawler or all crawlers, or the page is marked noindex, we respect that and show no score. If the page does not respond, or we hit a page meant to stop bots, we cannot measure it either.
What do you store when I check a website?
The address without its query string, the domain and the result, for 30 days. We do not store your IP address, email or browser details with the result, and we do not ask for your email.
Why do I get the same result when I check again?
The same address checked within the last 24 hours gives the same result, so a score you have shared holds. The person who ran the check can ask for a new measurement once something has been fixed.
Can I have a result deleted?
Yes. All results are deleted automatically after 30 days. If it is your website and you want the result removed sooner, email kontakt@avent.no with the link and we will delete it.
Does the check cost anything?
No. The check is free and requires neither an account nor an email address.
Would you rather see what your website could look like? Build a website with Avently →