Methodology
How we measure AI readiness
Every number in a Sparq report can be traced to a check ID, stored evidence, a source, a timestamp and an algorithm version. Where we cannot verify something, we say so.
What the score is
The Sparq AI Readiness Score is the share of measurable methodology points your website currently earns. Checks we cannot measure are excluded from both the earned and the possible total, and are listed as Not Verified. We never fill a gap with an estimate to reach a rounder number.
Deterministic scoring
Points are assigned by scoring rules stored in our database, keyed to a check ID and an algorithm version. A language model may classify evidence — for example, whether a page states a service area — but it never assigns points. The conversion from classification to points is deterministic and repeatable.
Evidence for every point
Each check stores its evidence: the excerpt or value observed, the source type, the source URL and the retrieval timestamp, together with the parser version and, where classification was used, the model and prompt version. Any score in your report can be opened to inspect that record.
The frozen Query Universe
For each domain we build a standardized set of 100 commercially weighted buying questions across provider selection, comparison, pricing, eligibility, urgency, process and informational intent. The set is versioned, stored and frozen for 90 days, so repeat scans are comparable. It is never silently regenerated per scan. In this preliminary scan the universe is auto-approved and labelled accordingly.
Answer coverage
Your own pages are retrieved against each question and scored on a 0–4 rubric. A score above zero requires a verbatim excerpt from your site. Coverage is weighted by commercial intent, and the unanswered high-intent questions are reported to you directly.
AI Visibility is not estimated
AI Visibility — whether assistants actually surface your business in their answers — requires a verified measurement source or an approved controlled panel. Until one is connected, visibility is reported as Not Verified and is excluded from the score. Nothing in this report is an official ranking from ChatGPT, Google, Gemini, Claude or Copilot.
Potential Points
Potential Points are unearned points in the Sparq methodology. They describe what the methodology would award if a gap were closed. They are not a prediction or a guarantee of increased AI visibility.
Crawl conduct
We identify ourselves as SparqAIPresenceBot, read robots.txt before fetching, follow a small bounded set of public pages, cap response size and time, and validate every redirect destination. Private and internal network addresses are refused.