Scoring

How we score a domain.

4 weighted pillars, 15 checks with their points, the exact grade bands, and the cases where we refuse to give a grade. Every rule on this page is read from the same file the scorer runs, so it cannot drift from what we compute. Current criteria version 1.0.

Overview

One number, built from things you can check.

Every website you verify gets a GEO score out of 100 and a letter grade. It is a weighted sum of 4 pillars, each scored from 0 to 100. Page health is a calculation over your audited pages. The other three are lists of pass or fail checks whose points add to 100.

A domain needs at least 3 audited pages before we show a grade. Before that you see each page's own score and the pillars we can measure, with a message saying why there is no grade yet.

domain score = page health × 0.55
+ site technical × 0.20
+ consistency × 0.15
+ geo readiness × 0.10

Weights out of 100.

The four pillars

What each pillar measures, and why it matters.

55

Page health

How the individual pages we have audited score, blended so that systemic problems cannot be hidden behind one good page.

This pillar is a calculation rather than a checklist. Each page you audit already gets its own score from our page-level rules; the domain figure blends them:

70%

the average of your page scores

30%

the 10th-percentile page, your weak end

2×

weight of the home page, the one answer engines reach first

A site with one excellent page and nine broken ones cannot average its way to a good grade.

20

Site technical

Whether the site as a whole is reachable, crawlable, and open to the AI crawlers that answer engines use.
Points inside the pillar total 100. Some checks are scored proportionally: if eight of ten pages point at themselves correctly, that check earns eight tenths of its points.
30
AI crawlers are not blockedrobots.txt does not disallow the crawlers that answer engines use. Blocking them is the single most damaging thing a site can do to its AI visibility.
20
A valid XML sitemap existssitemap.xml is reachable, parses as XML, and lists at least one URL.
15
No audited page returns an errorEvery page we have audited answered with a 2xx or 3xx status.
15
robots.txt is readablerobots.txt either serves cleanly or is absent. A server error here leaves crawlers guessing what they may fetch.
10
robots.txt points to the sitemapA Sitemap: line in robots.txt is how most crawlers discover it.
10
The home page serves over HTTPShttps://your-domain/ answers successfully.
15

Consistency and coverage

Whether pages describe themselves distinctly and point at themselves correctly, and whether we have seen enough of the site to judge it.
Points inside the pillar total 100. Some checks are scored proportionally: if eight of ten pages point at themselves correctly, that check earns eight tenths of its points.
30
Every page has its own titleNo two audited pages share a title. Duplicates make pages look interchangeable.
25
Pages point at themselvesEach page has a canonical URL, and it is the page itself.
25
Every page has its own descriptionNo two audited pages share a meta description.
20
Enough of the site has been auditedMeasured against a target sample rather than the whole site, so a large site is never penalised for being large.
10

GEO and entity readiness

Whether machines can tell what this site is, who publishes it, and which parts are worth quoting.
Points inside the pillar total 100. Some checks are scored proportionally: if eight of ten pages point at themselves correctly, that check earns eight tenths of its points.
35
The home page identifies the organisationOrganization or LocalBusiness structured data on the home page tells an answer engine who publishes the site.
25
Structured data is used across the siteScored on the share of audited pages carrying any JSON-LD.
20
Quotable content is marked upAt least one page uses FAQPage, Article, HowTo, Product or BreadcrumbList — the types answer engines quote most readily.
10
The site name is consistentPages that declare og:site_name agree on it.
10
At least one page is published to an AI directoryA frozen snapshot at a stable URL that AI crawlers are explicitly allowed to read.
Grades

Six bands, no curve.

A+
95 and up
Excellent
A
85 to 94
Strong
B
75 to 84
Good
C
65 to 74
Fair
D
50 to 64
Weak
F
under 50
Failing
Ceilings and refusals

Some problems cannot be averaged away.

A number we cannot stand behind is worse than no number, so there are cases where we cap the score, or say plainly that we will not give one.

49

Home page cannot be indexed

It fails to return a success status, or it is marked noindex. The domain score is capped at 49.

59

Every AI crawler blocked

robots.txt blocks every AI crawler we check for. The score is capped at 59. For a tool about generative engine optimisation this is the most damaging thing a site can do, so we say so loudly.

3

Pages needed for a grade

With fewer we show each page's own score and no domain grade. An average of one or two samples is noise. If we cannot load your home page at all, you get one clear message instead of a breakdown.

A pillar we cannot measure is dropped, not scored zero, and the remaining weights are rescaled. The crawlers we check: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Google-Extended, PerplexityBot.

A real example

What one page audit actually returns.

This is the page-level breakdown from a real audit of example.com, shown as a fixed illustration. Page scores feed the page-health pillar above, and a page needs 90 or more to be published to an AI directory.

example.com, one page83/100
Indexability82
Metadata77
Content72
Technical95
Social90
Accessibility100
High, penalty 18: Very little indexable text was found. Add useful, original copy that answers the visitor's main questions.
Every check

The full catalogue, filterable.

Generated from the scorer's own rules, so it cannot drift from what we run.

PillarCheckWhat passesPoints
Site technical AI crawlers are not blocked robots.txt does not disallow the crawlers that answer engines use. Blocking them is the single most damaging thing a site can do to its AI visibility. 30
Site technical A valid XML sitemap exists sitemap.xml is reachable, parses as XML, and lists at least one URL. 20
Site technical No audited page returns an error Every page we have audited answered with a 2xx or 3xx status. 15
Site technical robots.txt is readable robots.txt either serves cleanly or is absent. A server error here leaves crawlers guessing what they may fetch. 15
Site technical robots.txt points to the sitemap A Sitemap: line in robots.txt is how most crawlers discover it. 10
Site technical The home page serves over HTTPS https://your-domain/ answers successfully. 10
Consistency Every page has its own title No two audited pages share a title. Duplicates make pages look interchangeable. 30
Consistency Pages point at themselves Each page has a canonical URL, and it is the page itself. 25
Consistency Every page has its own description No two audited pages share a meta description. 25
Consistency Enough of the site has been audited Measured against a target sample rather than the whole site, so a large site is never penalised for being large. 20
GEO readiness The home page identifies the organisation Organization or LocalBusiness structured data on the home page tells an answer engine who publishes the site. 35
GEO readiness Structured data is used across the site Scored on the share of audited pages carrying any JSON-LD. 25
GEO readiness Quotable content is marked up At least one page uses FAQPage, Article, HowTo, Product or BreadcrumbList — the types answer engines quote most readily. 20
GEO readiness The site name is consistent Pages that declare og:site_name agree on it. 10
GEO readiness At least one page is published to an AI directory A frozen snapshot at a stable URL that AI crawlers are explicitly allowed to read. 10
Method and versions

What this score is not.

It is a measure of how ready your site is to be read, understood and quoted by search engines and AI answer engines. It is not a prediction of traffic, and it is not a guarantee of ranking, indexing or citation, because those depend on systems run by other companies that change without notice. See our Terms of Service for the full position, and how it works for the audit itself.

Criteria 1.0
September 24, 2026. First release: four weighted pillars, fifteen checks, six grade bands, two ceilings, and a three-page minimum for a grade.
Why does page health count the weakest pages?

Because averages hide problems. 30% of the pillar comes from the 10th-percentile page, so broken pages cannot sit behind a few good ones.

Is a big site penalised for being big?

No. Coverage is measured against a target sample of 25 pages, not against your whole sitemap. A site with fifty thousand URLs and 25 audited pages scores full marks for coverage.

Why is there no score for my new site?

A grade needs at least 3 audited pages. We would rather show nothing than a number we cannot stand behind.

Which signals are scored?

Only what we can observe from a page we audited or a file we fetched. Categories are averaged only over the pages that reported them, so commerce checks apply where we detected commerce and not to your about page.

Who can see my score?

Only you. A domain scorecard is visible only to the verified owner unless that owner chooses to publish it.

Score your site in minutes.

Run a free audit