Signal Check

One page at a time — whichever URL you enter. Paste a landing page, a blog post, a product page. We fetch that page, parse its HTML and JSON-LD, and check two site-wide files (robots.txt and sitemap.xml) for access and indexing rules that apply to the whole site. Entity Clarity also cross-checks the homepage if the Organization schema is not on the scanned page, because it usually lives site-wide via a layout template. If you only type a domain, we scan the homepage.

Because AI citations happen at the page level. A specific landing page either has the right hero text, the right schema, and fast Core Web Vitals — or it does not. A full-site crawl gives you averages; Signal Check gives you the six signals for the exact page you care about. If you need a site-wide view, run the tool on your top five pages.

Because the other forty-four are noise for AI visibility in 2026. Every signal here has a direct line to whether LLMs cite your site. We skip the classic SEO checks that do not move AI citations — keyword density, meta keyword tags, image-alt coverage on a homepage that has no images. If you want the exhaustive list, the 37-point GEO Checklist is the companion tool.

Free, and no catch. No email, no account, no paywall. Haide runs this same diagnostic on every prospect call before we talk strategy. Making it public costs us nothing and makes the conversation sharper when someone does end up hiring us. The tool stands on its own — use it whether or not you ever contact Haide.

We log the URL you scanned, the six verdicts, a hashed IP for rate limiting, and your user-agent string. Full reports live in Upstash (EU region) for 30 days and then auto-expire. A summary row — URL, verdicts, hashed IP, user-agent — is logged to our Notion CRM (US, Data Privacy Framework certified). No tracking cookies. No cold emails based on a scan. If you want a report removed before the 30-day window ends, open the shared report URL and use the 'Delete this report' control at the bottom.

They come from Google's CrUX field data — the same numbers that power pagespeed.web.dev. If your site has too little traffic for CrUX, we fall back to Lighthouse lab data from the same API. Same source, same thresholds Google itself uses for ranking.

Yes. We do not verify ownership. Scan a competitor's landing page, a lead's homepage, or a blog post you just read. The report URL is shareable for 30 days. Do not scan anything you suspect is genuinely private or under NDA — the tool fetches what any browser can fetch, but be respectful.

Sometimes it will, and that is a useful signal in itself. Cloudflare's default bot-protection rules block the scanner's request to your page. When that happens, Signal 1 returns Fail with the explicit reason "bot challenge detected." That is exactly what LLM crawlers see. If you want the scanner through, allowlist it under Security → Bots in your Cloudflare dashboard.

Fair. Signal Check is a first-pass diagnostic — read it that way. If you blocked training bots on purpose, or your hero paragraph is intentionally short-and-punchy, ignore that Warn. The evidence block shows you exactly what we saw — use judgement.

Because LLMs quote the first 40–60 words of a page more often than any other section — it is the Bottom-Line-Up-Front position. If the page opens with "In today's fast-paced digital landscape" or a 2-word headline, you have handed your citation slot to a competitor with a better first paragraph. This signal measures whether your opening earns the quote.

Yes. The technical signals (HTTPS, schema, bot access, Core Web Vitals) are language-agnostic. Answer Capsule's length and boilerplate checks work in any language. TextRazor's entity layer supports the major European and Asian languages — for less-common ones, the entity analysis may be limited, but the rule-based checks still run.