Transparency
Data sources
Semantyra’s analysis is first-party. It crawls the site itself and, where you connect them, uses your real Search Console data and a SERP provider. AI visibility comes from live calls to answer engines. Nothing is scraped from a third-party authority index.
| Source | What it is used for | Availability |
|---|---|---|
| First-party crawl | Always. Semantyra fetches and parses the site itself: titles, headings, body text, links, images, structured data, status codes, canonicals, rendering. | Required |
| Semantic embeddings | To cluster pages by meaning and compare topics. A hosted embedding model when available, a lexical fallback otherwise (with lower reported confidence). | Automatic |
| Language models | Entity extraction and short narrative summaries. Routed through a multi-model orchestrator; free models first, a paid model only when the plan and budget allow. | Automatic, budget-capped |
| Google Search Console | Clicks, impressions, CTR, average position, queries and pages. Used to compute search performance and to separate brand from non-brand visibility. | Optional, you connect it (OAuth) |
| SERP provider (DataForSEO) | Keyword volume, difficulty and organic rankings for tracked keywords and competitors. Cached and reused between scans to control cost. | Optional, operator-configured |
| AI answer engines | Live calls to ChatGPT (OpenAI), Claude (Anthropic) and Perplexity with web search, to measure brand mentions, citations and competitor presence for non-brand prompts. | Paid plans; every call is cost-logged |
What Semantyra never does
- No fabricated traffic, rankings, reviews, ratings, research or backlink counts.
- No third-party “domain authority” style score presented as Semantyra’s own.
- No number shown as zero when it means “not connected” or “not enough data”.
- No AI-visibility figure without the underlying prompts, providers and citations behind it.
See the methodology for how each score is computed.