Transparency is a core principle
We believe you should understand exactly how your score is calculated. SEO tools have historically been black boxes — they give you a number without explaining what it means or how it was derived. We want to do better.
Here's a complete, honest look at how Rankly.AI works: what we measure, how we weight it, and why.
The analysis pipeline
When you enter a URL, Rankly.AI runs a multi-step analysis pipeline that completes in under 30 seconds:
- Domain resolution and crawl: We fetch your homepage and key inner pages, extracting technical signals, content structure, and on-page SEO elements.
- Backlink intelligence: We query multiple backlink databases to determine your referring domain count, domain authority distribution, and link quality signals.
- Review aggregation: We pull review data from Google My Business, Trustpilot, and other platforms — including rating, volume, recency, and response rate.
- Technical audit: We run automated checks for Core Web Vitals estimates, HTTPS configuration, structured data presence, mobile-friendliness signals, and page speed indicators.
- Content analysis: We evaluate heading structure, content depth, keyword relevance, and the presence of AI-friendly content formats (FAQs, tables, definitions).
- Social footprint check: We identify active social profiles and cross-reference brand mention signals.
- Score computation: All signals are normalised and aggregated into a weighted composite score.
The six scoring dimensions explained
🔗 Backlinks & Authority (30%)
The single most important dimension, because it's also the strongest signal for both Google and AI models. We measure:
- Referring domain count (unique domains linking to you)
- Domain Authority of referring sites
- Dofollow vs nofollow ratio
- Topical relevance of linking domains
- Link velocity (recent acquisition rate)
We weight this dimension heaviest because it's the hardest to fake and the most strongly correlated with AI citation rates in our research.
⭐ Reviews & Reputation (20%)
AI models treat review signals as real-world quality indicators. A business with consistent, positive, numerous reviews is treated as more trustworthy and more worth recommending. We measure:
- Overall rating average (across platforms)
- Total review volume
- Review recency (recent reviews matter more)
- Platform diversity (Google + Trustpilot + G2 beats Google alone)
- Owner response rate (active engagement signals)
⚡ Technical Performance (20%)
For live-retrieval AI tools, your technical performance directly determines inclusion. For trained models, it's a proxy for overall site quality. We check:
- HTTPS enforced everywhere
- Estimated Largest Contentful Paint (LCP)
- Mobile viewport configuration
- Structured data presence and type
- Sitemap and robots.txt validity
- Canonical tag implementation
📝 Content & On-page SEO (15%)
Content quality determines whether AI models can extract and cite useful information from your site. We evaluate:
- H1 presence and keyword alignment
- Heading hierarchy (H1 → H2 → H3 structure)
- Word count and content depth
- Meta description completeness
- Internal linking structure
- Image alt text coverage
🌐 Social Presence (10%)
Social signals contribute to the web-wide brand footprint that AI models use for entity recognition. We check:
- Active profiles on major platforms (LinkedIn, Instagram, X, Facebook)
- Profile completeness and consistency
- Cross-linking between social profiles and main site
💰 Price Transparency (5%)
For commercial queries, AI models prefer to recommend businesses that are clear about pricing. Hidden pricing creates friction in the recommendation process. We evaluate:
- Presence of a dedicated pricing page
- Clear pricing information on product/service pages
- Conversion clarity (CTAs, free trial offers, contact forms)
Score interpretation
All six dimension scores (each 0–100) are combined using weighted averaging to produce your overall GEO score. A site scoring 70+ is well-positioned for AI visibility. Below 45 indicates significant gaps that are limiting your AI recommendation rate.
We recalibrate our scoring weights quarterly based on new research into AI model behaviour and updated analysis of what actually predicts AI citation rates.
What we don't measure (yet)
In the interest of full transparency, several signals we're working on adding:
- Direct AI mention testing (querying multiple AI models and tracking citation frequency)
- Featured snippet capture rate
- Knowledge panel presence
- Wikipedia coverage
- Multilingual content coverage
These will roll out with Premium and Expert features over the coming months. Our goal is to be the most comprehensive GEO measurement platform available — transparent, accurate, and continuously improving.
Run your free analysis now to see where you stand across all six dimensions.