What is fetched
For each domain, SeoFightr fetches the server-rendered HTML of the root URL over HTTPS with a normal browser user agent, reads the response timing, and reads the site's robots.txt directives. Everything measured is public. Neither domain needs to grant analytics, Search Console or any other access — which is why you can score a competitor.
The response body is capped, so extremely large pages are truncated before analysis. Redirects are followed to the final document.
The five SEO signals
- Speed — time to first byte and total response time for the root document, plus render-blocking weight in the HTML head.
- Crawlability — robots.txt directives, robots meta tags, canonical presence and consistency, and the discoverability of a sitemap.
- Structure — a single H1, a sane heading hierarchy, semantic landmarks, title and description presence and length, and structured data.
- Links — internal link density and depth from the entry page, plus outbound links to identifiable external sources.
- Mobile — viewport declaration, responsive layout hints and the absence of fixed-width layout signals.
The five GEO signals
- Extractability — can a single sentence be lifted out and still make sense? Answer-first paragraphs and self-contained statements score high; content that only makes sense in sequence scores low.
- Evidence — numbers, dates, named sources and citations that a model can treat as verifiable rather than as marketing claims.
- Identity — who is speaking. Organization, Person/Author, logo, sameAs links and consistent naming across the page and its structured data.
- Question coverage — how well the page matches natural-language questions rather than keyword fragments, including explicit Q&A blocks.
- Citability — stable URLs, server-rendered content, permission for AI crawlers in robots.txt, and a machine-readable site summary such as llms.txt.
How the totals and the K.O. are computed
Each signal is scored 0-100. The SEO total is the average of its five signals; the GEO total is the average of its five. Neither total is capped or curved, so a site with a genuinely clean profile can reach the high nineties.
The winner of a round is the higher total. The K.O. verdict is decided by the margin, not the winner alone: a margin under 5 points reads as a split decision, 5-15 as a clear win, and anything above that as a knockout. This is deliberate — a two-point difference between two domains is inside the noise band of the speed signals and should not be presented as a decisive result.
What the score is not
It is not a ranking prediction. It does not read Google's index, it does not measure backlinks from a third-party link graph, it does not know your traffic, and it cannot confirm whether ChatGPT or Perplexity currently cites you.
A high score means nothing structural is blocking you from ranking or being cited. Winning the query still depends on content quality, topical authority and demand — none of which a single-page structural audit can see. Read SEO vs GEO for how the two disciplines interact, or the AI searchability guide for the GEO half in depth.
Methodology FAQ
How is the SEO score calculated?
The SEO score is the weighted average of five signal groups scored 0-100 each: speed, crawlability, structure, links and mobile. Each group is measured from the live HTML response of the domain's root URL, then normalised onto a 0-100 scale.
How is the GEO score calculated?
The GEO score averages five generative-search signals: extractability, evidence, identity, question coverage and citability. They estimate how easily an answer engine can lift a self-contained, verifiable statement from the page and attribute it to a known entity.
What data does SeoFightr fetch?
Only publicly available data: the server-rendered HTML of the domain's root URL, its robots.txt directives and the metadata embedded in that HTML. No analytics access, no Search Console access and no login is required for either domain.
What are the limits of the score?
It is a structural audit of one page, not a ranking prediction. It does not measure keyword rankings, backlinks from third-party indexes, traffic, or whether a specific AI assistant currently cites you. A high score means nothing is structurally blocking you.
Why do two runs give slightly different numbers?
Speed-related signals depend on the response time of the live fetch, so they can vary between runs. Structural signals such as schema, headings and robots directives are deterministic and only change when the site changes.