AI Channel Diagnostic
We score your domain across four pillars — AI Behaviour, AI Signal, AI Technical, and AI Transactability — using real provider runs and structured probes. Free, no account required.
Does ChatGPT and Google AI Mode recommend you when consumers ask category questions? We run 10 real queries × 2 providers = 20 observations.
Off-site brand signal that AI assistants weight: reviews, editorial mentions, community presence, creator coverage, PR over the last 12 months.
Whether AI crawlers can read your site: robots.txt access for 8 AI bots, llms.txt presence, JSON-LD on homepage and PDPs, sitemap, image alt text.
Whether you're positioned for AI buyer agents: commerce platform fingerprint, MCP probes, Agent Payments markers, Google Merchant feed.
Every check runs the same fixed pipeline. We scrape your homepage, gate on whether you're an e-commerce brand, identify a hero product, then generate 10 consumer-facing queries tuned to your category and market. We submit those queries to two AI assistants, scrape the open web for off-site brand signal, and probe your site for AI-crawler access and agentic-commerce readiness. A Sonnet-class model synthesises the findings into a verdict, three wins, and three actionable gaps.
Each dimension scores 0–100. The overall is a weighted average: D1×40% + D2×25% + D3×20% + D4×15%. There is no floor rule — D4 is forward-looking and most brands score low here today, which is honest.
We submit 10 queries to ChatGPT and Google AI Mode via real browser sessions. If you have a clear hero product, the split is 5 brand+category + 5 hero-product queries. If not, all 10 are brand+category — and we surface "no clear hero product" as a finding.
Each provider × query response is extracted into structured findings: did the brand appear, was it recommended, in what position, with what sentiment, alongside which competitors, with what URL citations. Per-query Purchase Likelihood (PL) and Narrative Accuracy (NA) scores feed the dimension score: 60% PL + 30% NA + 10% citation frequency. The quadrant diagnostic — Recommended & Accurate / Recommended but Inaccurate / Accurate but Unseen / Invisible — comes from PL × NA.
Five sub-points, each scored 0–100, averaged for the dimension score:
Six sub-points, averaged:
Four sub-points, averaged. D4 is forward-looking — most brands today score low here, which is honest:
Each (domain, market) pair can be re-checked every 28 days. Within that window, submitting the same domain returns the existing report — we don't re-run an expensive pipeline that won't surface new signal yet. Every run is preserved in history; you can compare past scores from the report page.