Support & FAQ
Everything you need to know about Agentic SEO and our auditing platform.
Technical Audits
What is "Saliency-Weighted Optimization"?
Modern AI agents like Google Jarvis use vision models to "see" your site. Saliency-weighted optimization ensures your primary buttons and navigation paths have the necessary contrast and placement to be prioritized by these agents.
Why did my site fail the "AI Accessibility" test?
Most websites are built for human eyes, but "Operator" agents navigate via the Accessibility Tree. If your buttons lack ARIA-labels or your structure is purely div-based, the AI cannot "read" the function of your elements.
How does the JSON-LD Generator handle complex sites?
Our Agentic JSON-LD Generator uses hierarchical logic modeled after Google Project Jarvis's DOM parser. This ensures that even on complex pages, the AI Agent identifies the correct parent-child relationships for conversion events, so it never misses a "Buy" or "Submit" button.
What is the difference between Shielded Discovery and Deep Scrape?
Shielded Discovery (Contextual) uses high-level agentic signals and search metadata to identify 90%+ of strategic friction points without needing to bypass site security. Deep Scrape (Direct DOM) uses a headless browser to verify the exact technical implementation of ARIA labels, contrast ratios, and shadow-root elements for 100% engineering accuracy.
Why do my results change slightly between audits in Discovery Mode?
Discovery Mode (Contextual Audit) relies on real-time "Agentic Discovery Signals" from global search indexes. These signals fluctuate as search engines update their metadata and as our AI synthesizes new semantic structures from the latest available content. For 100% frozen, technical DOM-level accuracy, we recommend using the Deep Scrape (Retry) feature, which analyzes the site's code directly.
Why does my Agent Readiness Score fluctuate slightly between scans?
Unlike legacy SEO tools that use static checklists (like counting H1 tags), GEO Stellar uses live, non-deterministic AI models (like Anthropic Claude and OpenAI Operator) to evaluate your site. Because these models are probabilistic, your score reflects the real-time, living nature of Agentic Search. Just as two human reviewers might grade a complex essay slightly differently, LLMs will exhibit a minor variance (usually +/- 3 to 5 points) depending on how they weight specific DOM elements in that exact moment. This fluctuation is normal and proves you are getting a true, live agentic evaluation rather than a cached static report. We recommend focusing on pushing your Agentic Friction to 0 or close thereto, as the actionable insights and code patches will typically remain consistent between runs.
Credits & Billing
How do "Audit Power Packs" work?
Our hybrid system uses your monthly subscription allowance first. If that is exhausted, it pulls from your purchased Power Pack credits, which never expire.
Can I get a refund for unused credits?
Due to the immediate costs associated with running high-level reasoning models (GPT-4o/o1), we cannot offer refunds once credits are used to initiate an audit.
Trust & Directory
How do I get listed on the AI-Ready Directory?
Our automated system periodically scans for sites with an Agentic Readiness Score of 90% or higher. Once you reach this elite tier, you qualify for a featured listing in our directory.