Methodology

How SEO health scores are calculated

The score is a transparent summary of the problems an investigation found. It is calculated the same way every time, from findings you can see in the report. It is our own measure and not a Google metric or a ranking prediction.

1. Every check produces one result

An investigation runs 53 checks, listed below. Each check reports at most one result for the page: passed, informational, or a finding with a severity. If the same problem affects many URLs (for example several broken links), it is one finding whose evidence lists every URL, so a single issue is not counted many times.

Findings are also labelled as a problem (a confirmed technical fault, such as a noindex directive or an invalid certificate) or an opportunity (an optional improvement or a heuristic, such as a short title or missing Open Graph tags).

2. Penalties per severity

SeverityMeaningPoints deducted
CriticalPrevents the page being reached, trusted or indexed30
HighSeriously harms crawling, indexing or user trust15
MediumMeasurably weakens how the page is understood or shown7
LowMinor issue or recommended improvement2
InfoUseful context only0
PassedNo issue found0

3. Category scores

Each of the four categories starts at 100 and loses the penalty of each finding in it, down to a minimum of 0. A category with no evaluated checks (for example on-page SEO when the page returned an error) is shown as “–” and left out of the overall score.

4. Overall score

The overall score is the weighted average of the evaluated category scores, rounded to the nearest whole number:

CategoryWeight
Technical SEO35%
On-page SEO30%
Performance15%
Security & Trust20%

Example: technical 70, on-page 98, performance 100 and security 85 give (70×35 + 98×30 + 100×15 + 85×20) ÷ 100 = 86.

Grades: 90–100 Excellent, 75–89 Good, 50–74 Needs work, below 50 Poor.

5. What we measure, and what we don’t

  • Measured directly: HTTP responses, redirects, headers, TLS certificates, robots.txt, sitemaps and the HTML as delivered by the server.
  • Server response time is a single measurement from our server. It is lab data that varies with network distance, not real-user data.
  • Not measured: Core Web Vitals, search rankings, traffic, backlinks or domain authority. We do not show estimates for these.
  • JavaScript is never executed. Pages that build content in the browser are analysed as their initial HTML.
  • Heuristics such as content length and title length are always labelled as opportunities, never as confirmed errors.

6. Site crawls

A site crawl investigates the start page with every check below, then follows internal links breadth-first (up to the chosen page limit) and adds 25 site-wide checks. Each site-wide check also reports once for the whole site, with every affected URL in its evidence, and uses the same penalties, so crawl scores are comparable with single-page scores.

  • Only URLs on the same site are crawled (www and non-www count as one site; other subdomains do not). robots.txt is honoured for TheSEODetectiveBot, and links marked nofollow are not followed.
  • Duplicate content is detected with a fingerprint of each page’s visible text (identical text) and a 64-bit SimHash (near-identical text, at most 3 differing bits).
  • Orphan pages are only reported when the crawl reached every discoverable page; otherwise sitemap URLs that were not reached are listed for information.
  • Response times are one measurement per page from our server, not real-user data.

7. Every check

Technical SEO

CheckWhat is examined
Page returns a successful response
http_status
The final HTTP status code of the investigated URL after redirects, and that it serves HTML.
No redirect loops
redirect_loop
Whether following redirects ever returns to a URL already visited, or exceeds 10 hops.
Short redirect path
redirect_chain
How many redirects are followed before the final page is reached.
HTTP redirects to HTTPS
http_to_https
Whether http:// requests to the host are permanently redirected to https://.
WWW and non-WWW are consolidated
host_consistency
Whether the www and non-www hostnames redirect to a single preferred host.
robots.txt is reachable and valid
robots_txt
The HTTP response and syntax of /robots.txt.
robots.txt allows the page
robots_allows_page
Whether robots.txt rules block Googlebot or all crawlers from the investigated page.
XML sitemap is available and valid
sitemap
Sitemaps declared in robots.txt, or /sitemap.xml: reachability, XML structure, namespace and URL entries.
Indexable (meta robots)
meta_robots
noindex, nofollow and none directives in <meta name="robots"> and <meta name="googlebot">.
Indexable (X-Robots-Tag header)
x_robots_tag
noindex, nofollow and none directives in the X-Robots-Tag HTTP response header.
Canonical URL declared
canonical_present
Whether the page declares a canonical URL with <link rel="canonical"> or a Link header.
One consistent canonical declaration
canonical_single
Whether more than one, or conflicting, canonical URLs are declared.
Canonical URL is valid and resolves
canonical_valid
Whether the canonical URL is an absolute http(s) URL that returns HTTP 200 without redirecting.
Page links to other internal pages
internal_linking
Count of internal links crawlers can follow, and internal links marked nofollow.
No broken internal links
broken_internal_links
HTTP status of internal links on the page (up to the probe limit), excluding URLs disallowed by robots.txt.
Internal links point to final URLs
redirecting_internal_links
Internal links that respond with a redirect instead of the content.

On-page SEO

CheckWhat is examined
Page has one title
title_present
Presence and count of the <title> element.
Title length is practical
title_length
Title length in characters; very short titles carry little information, titles over ~65 characters are often truncated in results.
Meta description present
description_present
Presence and count of <meta name="description">.
Meta description length is practical
description_length
Description length in characters (50–160 is a practical range for result snippets).
Page has a main (H1) heading
h1_present
Number of <h1> elements and whether they contain text.
Heading levels are in order
heading_hierarchy
Whether heading levels are skipped (for example an <h2> followed directly by an <h4>).
Images have alternative text
image_alt
Images without an alt attribute. alt="" is treated as a correctly marked decorative image.
Links have descriptive text
anchor_text
Links without any text, aria-label or image alt, and links with generic text such as "click here".
Page has substantive text
content_length
Visible word count, excluding scripts, styles and navigation markup. A heuristic, not a ranking rule.
Page language declared
lang_attribute
The lang attribute on the <html> element.
Mobile viewport configured
viewport
The viewport meta tag needed for mobile-friendly rendering, and whether it disables zooming.
Open Graph metadata
open_graph
og:title, og:description and og:image, used when the page is shared on social platforms.
Twitter/X card metadata
twitter_card
The twitter:card meta tag.
Structured data present
structured_data
Whether the page carries JSON-LD or microdata structured data.
JSON-LD is valid
jsonld_valid
Whether every JSON-LD block parses as JSON and declares a schema.org @context and @type.

Performance

CheckWhat is examined
Fast server response
response_time
Time to first byte of the final page, measured once from our server. Not real-user (field) data.
Reasonable HTML size
html_size
Size of the decompressed HTML document.
Text responses are compressed
compression
Content-Encoding of the HTML and of probed CSS/JavaScript files.
HTML caching policy
html_cache
The Cache-Control and related headers sent with the HTML (informational).
Moderate number of resources
resource_count
Counts of referenced scripts, stylesheets and images.
No render-blocking scripts in <head>
render_blocking
External scripts in <head> without async, defer or type="module".
Images declare dimensions
image_dimensions
Images without width and height attributes, which can cause layout shifts while loading.
Static assets are cacheable
static_cache
Cache lifetime (Cache-Control max-age or Expires) of probed same-site CSS, JavaScript and image files.
Page resources load
broken_resources
HTTP status of probed same-site CSS, JavaScript and image files.

Security & Trust

CheckWhat is examined
HTTPS available
https_available
Whether the host accepts TLS connections on port 443.
Valid TLS certificate
tls_certificate
Certificate trust chain, validity dates and hostname match.
Certificate not close to expiry
tls_expiry
Days until the certificate expires.
Modern TLS protocol
tls_protocol
The TLS version negotiated with the server.
Redirects stay secure and on-site
redirect_safety
Redirects from HTTPS to HTTP, and redirects that leave the submitted domain.
No mixed content
mixed_content
http:// scripts, stylesheets, frames, media and form targets on an https:// page.
HTTP Strict Transport Security
hsts
The Strict-Transport-Security header and its max-age.
Content Security Policy
csp
The Content-Security-Policy header.
MIME sniffing disabled
content_type_options
X-Content-Type-Options: nosniff.
Clickjacking protection
frame_protection
X-Frame-Options, or frame-ancestors in the Content-Security-Policy.
Referrer policy set
referrer_policy
The Referrer-Policy header.
Permissions policy set
permissions_policy
The Permissions-Policy header (informational).
No software version disclosure
server_disclosure
Server and X-Powered-By headers that reveal software names and versions.

Site-wide checks (crawls only)

Technical SEO

CheckWhat is examined
Crawl coverage
crawl_coverage
How many pages were crawled, and whether the crawl reached every discoverable page or stopped at its limit (informational).
No broken external links
broken_external_links
HTTP status of links to other websites (up to the plan limit), excluding URLs those sites disallow in robots.txt.
No internal redirect chains
internal_redirect_chains
Internal URLs that need two or more redirects, or loop, before reaching a page.
Pages excluded from indexing
site_noindex
Crawled pages with a noindex directive in meta robots or X-Robots-Tag (informational: often intentional).
Sitemap lists only indexable pages
noindex_in_sitemap
Pages listed in the XML sitemap that carry a noindex directive, redirect or return an error.
Canonical URLs are consistent
site_canonicals
Pages whose canonical points to a URL that redirects, fails, is noindex, or that declare conflicting canonicals.
No orphan pages
orphan_pages
URLs listed in the sitemap that no crawled page links to. Only reported when the crawl reached every discoverable page.
Indexable pages are in the sitemap
sitemap_coverage
Crawled, indexable pages that the XML sitemap does not list.
Pages are close to the start page
page_depth
Click depth of each page from the start URL; pages more than three clicks deep are harder for crawlers and visitors to reach.
Pages are well linked internally
weak_internal_links
Indexable pages that only one other crawled page links to.
Internal links are crawlable
robots_blocked_links
Internal links pointing to URLs that robots.txt disallows (informational).

On-page SEO

CheckWhat is examined
Every page has a title
site_missing_titles
Crawled HTML pages without a <title>.
Page titles are unique
duplicate_titles
Indexable pages that share the same title.
Titles have a practical length
site_title_length
Pages whose title is shorter than 15 or longer than 65 characters.
Every page has a meta description
site_missing_descriptions
Indexable pages without a meta description.
Meta descriptions are unique
duplicate_descriptions
Indexable pages that share the same meta description.
Every page has an H1
site_missing_h1
Crawled HTML pages without an H1 heading, or with an empty one.
Pages have one H1
site_multiple_h1
Pages with more than one H1 heading (an optional improvement).
No duplicate content
duplicate_content
Indexable pages whose visible text is identical (same content fingerprint).
No near-duplicate content
near_duplicate_content
Indexable pages whose text is almost identical (64-bit SimHash within 3 bits).
Pages have substantive text
thin_content
Indexable pages with fewer than 200 words of visible text (a heuristic).
Images have alternative text
site_image_alt
Pages with images that have no alt attribute.

Performance

CheckWhat is examined
Pages respond quickly
slow_pages
Pages whose time to first byte, measured once from our server, exceeded 1.8 seconds.
Pages have a reasonable size
large_pages
Pages with more than 1 MB of HTML.

Security & Trust

CheckWhat is examined
No mixed content on any page
site_mixed_content
https:// pages that reference http:// scripts, stylesheets, frames, images or form targets.