How the Skilo Score is calculated
No black box: every point of your score comes from a signal actually measured on your website or in real AI answers. This page shows the entire scale — including what we do not measure yet.
Three components, one score
The Skilo Score (0–100) combines SEO (40% weight — your site's health for Google), GEO (30% weight — how ready your content is to be cited by AI) and AI Presence (30% weight — whether your brand is actually cited in answers, not just ready to be). The weights are fixed: the calculation is the same for everyone, whether you are a new customer or have been monitored for months.
The only difference for those who haven't run any citation test yet: without measured Presence, the SEO and GEO weights are redistributed between them (we don't penalize you for a test that hasn't happened) — and as soon as the first simulation runs, Presence enters the calculation with the Preview seal (see the seals section below).
The SKILO Method
The framework behind your score — 5 pillars that decide whether you deserve to be found, by Google and by AI.
“Does the site work for humans and machines?”
Technical base: speed, mobile, indexing and structured data.
Supports Site Health and Information Google Understands — the base of your SEO.
“Do you master the subject?”
Content: depth, originality, freshness and readability.
Supports Citation-Ready Content — the depth that makes AI trust what you write.
“Do you show up when people ask?”
AI presence: real citations, share of voice and continuous validation.
Supports AI Access to Your Site and Your Brand in AI — whether you are actually found and cited.
“Why should anyone trust you?”
Authority: authorship, social proof and primary-source content.
Supports Trust Signals — why someone (or an AI) should believe you.
“Is it clear what you sell and to whom?”
Purpose: value proposition and coverage of the buying journey.
Supports Content Clarity and influences every other pillar — is it obvious what you sell?
The weight of each pillar
Effective weights with AI presence measured (the paid-plan scenario). They add up to 100. The engine's technical name appears next to the friendly interface name.
We test it for real: we ask the AIs the prompts you registered and measure in how many answers your brand appears (average across all prompts, on the 5 AIs) — and POSITION matters: being the 1st recommendation earns full credit, later positions earn a bit less. The scale reflects how hard it really is: being cited in half of the queries already earns a high score. Your prompts are the yardstick for this score — choose questions your customers would really ask. Off-niche prompts pull the score down (or inflate it) without reflecting reality.
The "engineering" of your website: speed, titles, meta description, sitemap, robots.txt. It's the prerequisite — without a foundation, Google can't even read your site properly.
The invisible tags (JSON-LD schemas) that translate your content for machines: products, reviews, organization. It's what produces star ratings on Google and lets AI cite your data correctly.
How "easy to cite" your content is: direct answers, concrete data, questions answered, dates and identifiable authorship. AI prefers clear, dated and signed sources — vague, anonymous content doesn't get cited. It is also the foundation of AEO (Answer Engine Optimization): the same direct-answer format an AI cites is what Google uses for its quick answer box and what voice assistants read aloud.
Signs that your brand is trustworthy and worth converting with: reviews, return policy, company registration, a clear CTA, an About page. Ranking or being cited isn't enough — if the page isn't convincing, traffic doesn't turn into results.
The GEO prerequisite: whether AI bots (GPTBot, ClaudeBot, PerplexityBot) are allowed to read your site and whether the content exists in plain HTML, without depending on JavaScript. Great content is useless if the AI can't even access it.
How the page describes itself for EXTRACTION by AI: FAQ/HowTo/Article schemas, consistency of the brand as a single entity (sameAs) and navigation trail (breadcrumb). It's the format AI assistants use most to build direct answers from your content — the same FAQ/HowTo schema an AI uses to answer is what Google expands as a "related question" in search (that's AEO: optimizing for direct answers, in text or by voice).
What about AEO (Answer Engine Optimization)?
AEO is optimizing content to answer a question directly — the format that wins Google's quick answer box and the "related questions" block in search, and that voice assistants (Siri, Alexa, Google Assistant) read aloud. It is a goal adjacent to GEO (being cited by generative AI like ChatGPT, Gemini and Claude), not a separate pillar in our scale.
We don't create a separate AEO score because the format that wins direct answers is, in practice, the same format we already measure within GEO: Content Citability (question-and-answer, direct answer in the first 200 characters) and Semantic Structure (FAQ/HowTo schema). Fixing these two pillars serves both goals at once — AI citations and direct answers on Google/voice — without counting the same signal twice.
Presence maturity seals
The difference between a new customer and one monitored for months doesn't show up as a discount on the score — it shows up as a seal next to the number. The seal tells you how reliable the evidence is; the score itself is always calculated the same way.
Based on a live simulation run right now — there is no recurring monitoring history yet.
Based on real monitoring of your prompts, but without enough days of collection yet to confirm the pattern repeats.
Based on recurring monitoring (several distinct days of collection) — the most reliable reading we have.
The Real Presence scale
The single heaviest component (30%) uses a scale that recognizes how hard the real market is: in competitive niches, being cited in half of the relevant queries is already a strong presence. And position matters — being the 1st recommendation earns full credit.
Important: the measurement uses the prompts you register — questions consistent with what your customers would really ask keep the score accurate; off-niche prompts distort it.
Bonuses and stability
- Field performance (CrUX): when Google has speed data from real users of your domain, it earns points within Technical Foundation — a free, verifiable signal of scale. Without that data, we use the lab measurement as an approximation (it never drops to zero just because your site is small).
- Measurement tolerance (±3 pts): network fluctuations don't change your score — if your site didn't change, the score doesn't move.
- No plateaus: every action that improves a measured signal moves the score. The "what pulls your score down" list shows exactly how many points each fix brings back.
How we read your website
The analysis happens in two readings. The first is instant (the home page, plus up to two satellite product/FAQ pages when needed) and delivers your score in under 1 minute. The second runs in the background, minutes later: if a trust signal (privacy policy, company registration number, security seal, shipping, company details) didn't show up on the home page, we check real institutional pages on your site — About, Sustainability, Policies, Terms — before considering that signal missing. If we confirm something there, your score is adjusted automatically, and you see exactly which pages we read.
Scope honesty: this second reading confirms specific institutional signals — it is not a full-site crawl. For stores or platforms with hundreds or thousands of pages, we don't read the entire catalog today, and the score never pretends to have read more than it did.
What we don't measure yet (on purpose)
The score does not consider backlinks or domain authority — those metrics require paid external databases and become estimates that are easy to inflate. We prefer to measure only what we can verify: real on-page signals from your site and real citations in AI answers. When something wasn't measured, the score says so instead of guessing.
Free, in under 1 minute.