Skip to content
Sitequiry
Methodology

How Sitequiry scores a page

This page states how a free Sitequiry analysis is produced: what is fetched, which numbers come from Google's measurements, how each score is calculated and what the tool does not measure. The rules below are the ones the software applies; when they change, this page changes with them.

It covers the free analysis of one URL. The paid Site Report checks up to 100 pages of one site and adds checks between pages; it is not described here.

Updated Written by the GE-KO team

What the analysis fetches

  • The page you submit, with the user agent SitequiryBot/1.0. If the server answers 429 with a short wait, or the transfer times out or is reset, there is one more attempt. Any other failure ends the analysis and the result names the reason.
  • robots.txt and llms.txt of the same host, to report what they say. The request for the submitted page itself does not consult robots.txt: it is the one address you asked for.
  • The type and size of up to 18 files the page loads, such as stylesheets, scripts and images.
  • Two Google PageSpeed Insights runs, mobile and desktop. For these your server is visited by Google's Lighthouse, not by us.

The page must be public: nothing is logged in, and addresses on private networks are refused. A result is stored for 24 hours per language, so a second analysis of the same address in that time returns the stored result. One visitor can start up to 10 analyses per hour.

Three kinds of evidence

Every line in a report carries one of three labels:

  • Measured: a value returned by a measurement or a direct network check. PageSpeed scores and metrics, Chrome UX Report data, the response time of our own request, and what robots.txt and llms.txt answered.
  • Detected: something present or absent in the fetched HTML or in the HTTP headers. The title, the meta description, headings, the canonical URL, a noindex directive, structured data types.
  • Inferred: a judgement drawn from several signals, such as how clearly the page states its offer, and anything that could not be checked. A robots.txt that cannot be read is reported as unknown, not as open or closed.

The four scores and the overall score

Each score runs from 0 to 100.

How each score is calculated
Score Calculation
SEO 60% technical SEO plus 40% content and search clarity, rounded.
Speed 70% mobile plus 30% desktop PageSpeed performance score. With only one device measured, that device alone.
AEO (AI answer readiness) Starts at 100; points are deducted for missing signals, as listed below.
GEO (AI citability) Starts at 100; points are deducted for missing signals, as listed below.
Overall The plain average of the four scores, rounded.

Scores are shown in four bands: 0 to 49 critical, 50 to 69 needs work, 70 to 84 good, 85 to 100 strong. These are Sitequiry's own bands. The PageSpeed part of the report keeps Google's: 90 and above good, 50 to 89 needs improvement, below 50 poor.

SEO: a technical part and a content part

Technical SEO is the Lighthouse SEO score from PageSpeed, averaged over the mobile and the desktop run. Only when PageSpeed returns no result does Sitequiry score the HTML itself. Content and search clarity is always calculated from the HTML and cannot fall below 38. Both start at 100 and lose points as follows:

Deductions in the SEO score
Check Technical SEO (without PageSpeed) Content and search clarity
Title missing, shorter than 20 or longer than 70 characters 8 18
Meta description missing, shorter than 70 or longer than 170 characters 6 14
Not exactly one H1 8 16
No canonical URL 8 0
A noindex directive in the page or in the X-Robots-Tag header 26 0
No lang on the html element 5 0
Fewer than 70% of images with an alt attribute 10 6
No main element, or fewer than two headings or sections 4 8

HTTPS, security headers, structured data and internal links are reported as findings and do not change the SEO score. The character ranges are the limits for a warning, not a recommended length.

Speed

The Speed score is Google's Lighthouse performance score, 70% mobile and 30% desktop. The report also shows the lab metrics of each run (FCP, LCP, CLS, TBT and Speed Index) and up to eight opportunities per device.

Field data comes from the Chrome UX Report in the same PageSpeed response: for the URL when Google has enough visits, otherwise for the whole origin, and the report says which. Core Web Vitals are shown as passed only when Google's overall assessment of that data is "fast".

If neither PageSpeed run succeeds, the score is an estimate from our own request: server response time, page weight, the number of render-blocking files and similar signals. It cannot exceed 90 and is labelled "Estimate — not measured". It is not a Lighthouse score and should not be compared with one.

AEO: can AI search reach and read the page

The score starts at 100 and loses points for each of these:

Deductions in the AEO score
Finding Points
No structured data of any kind (JSON-LD, microdata or RDFa) 25
A JSON-LD block that cannot be parsed 8
No Organization or LocalBusiness type 10
No FAQ and fewer than two headings phrased as questions 8
Empty meta description 8
nosnippet or max-snippet:0 25
max-snippet below 50 characters 8
Open Graph title and description not both present 7
robots.txt closes the whole site to Googlebot 30
robots.txt blocks at least one AI search or user-request crawler 20, once

The deduction for missing questions does not apply to pages where a question-and-answer block is not expected: shop category listings, content portals, and contact, legal and checkout pages recognised by their address. The report says so when it applies. The weight was 12 until 10 November 2026 and is now 8. Microsoft says that direct questions with clear answers can often be lifted word for word into AI answers; Google states that no special optimisation is needed for its AI features.

The crawlers that count are OAI-SearchBot, Claude-SearchBot, PerplexityBot and DuckAssistBot (search) and ChatGPT-User, Claude-User and Perplexity-User (requests made for a user). Training crawlers such as GPTBot, ClaudeBot, Google-Extended and CCBot are listed in the report and never cost points: opting out of training is a legitimate choice. A site without a robots.txt (HTTP 404 or 410) is reported as having no restrictions. When the file exists but cannot be read, nothing is deducted and access is reported as unknown. llms.txt is requested and its presence is reported, but a missing file costs no points: no major search or AI provider has committed to reading it. The free AI crawler checker shows the same rules crawler by crawler.

GEO: is the content specific enough to cite

The score starts at 100 and loses points for each of these:

Deductions in the GEO score
Finding Points
Fewer than 150 words of visible text 25
150 to 399 words 10
Heading structure other than one H1 with at least one H2 6 to 20
No organisation named in structured data or in the text (not applied to articles) 15
No location stated (not applied to articles and products) 5
No clear offer or topic in the headings, the title or the description (not applied to articles) 8
No meta description, or an opening text under 120 characters 8
Article without an author 12
Article without a date 10

Every GEO factor is detected in the HTML or inferred from it. Nothing in this score is a measurement.

What Sitequiry does not measure

  • Whether an AI assistant mentions, cites or recommends a site. AEO and GEO are readiness signals only.
  • Rankings, impressions and clicks, and whether Google has indexed a page. Those are in Google Search Console.
  • Backlinks, search demand and competitors.
  • Legal compliance. The accessibility check is a separate test based on Lighthouse's automated accessibility audits and our own HTML checks; it is not part of the four scores, and a clean result means that automated rules found nothing, not that the page conforms to a standard.

Who is responsible and how to reach us

Sitequiry is built and operated by GE-KO, Maja Dolžan s.p., Cesta Toneta Tomšiča 71, 4270 Jesenice, Slovenija (registration number 8732043000, VAT ID SI33880263).

Questions about a result or about this method: info@ge-ko.si or +386 51 821 907. For work on your own website, use the inquiry form.

Frequently asked questions

Why does the Speed score differ from the number PageSpeed Insights shows me?

PageSpeed Insights shows one device at a time, while the Speed score blends the mobile and the desktop run 70 to 30. A Sitequiry result is also stored for 24 hours, so it can be older than a test you run now.

Does a high score mean the page will rank well?

No. The scores describe the page: its markup, its speed and how clearly it is written. Rankings also depend on competition, links and what people search for. See SEO score vs rankings.

Can a score change when nothing on my site changed?

Yes. Speed is measured again at every new analysis, so the numbers can differ. The rules on this page can also change; the date under the introduction shows the last revision.

See the method applied to your page

One public URL, the full result and a link you can share. No sign-up.