Skip to main content
AI models surface trust signals when answering high-stakes queries — health, finance, legal, B2B services where credibility is the buying decision. Sites without basic structural credentials get filtered out of citations even when the content is correct, because AI can’t verify who’s behind the answer.

​
Methodology

Cited checks for two foundational credibility signals across the sampled pages: an about page (telling readers and AI models who you are) and contact information (telling them how to reach you). Both are detected at the site level rather than per page — finding either on any single sampled page satisfies the signal. About page detection. A site has an about page if any of these conditions is met:
  • A sampled URL’s path contains /about, /our-story, /who-we-are, /company, or /our-mission (the five canonical about-page slugs)
  • Any sampled page has at least one internal link to a URL whose path contains one of the above patterns
The first condition catches sites where the about page is itself in the sample. The second catches sites where the about page isn’t sampled but is linked from sampled pages (usually via the nav or footer). About page found → +2 points. Contact information detection. A site has contact info if any sampled page contains:
  • A mailto: link (any element with href="mailto:…")
  • A phone number (matched as tel: href or a regex pattern +?\d[\d\s\-()]{7,} in the raw HTML)
  • An internal link whose path contains /contact
The phone regex is intentionally loose — it matches typical international and domestic formats without enforcing strict E.164 parsing. Contact info found → +2 points. Both signals bonus. Sites with both an about page and contact info earn an additional +1 point. The bonus reflects that the two signals together carry more weight than the sum of their parts — AI models recognize sites that pass both checks as having editorial discipline, not just one isolated signal. The signal scores out of 5. Status thresholds: pass at 4/5 (requires both about + contact + bonus), partial at 2/5 (one of the two signals alone), fail below. Why focus on these two signals specifically? About and contact pages are the closest universal proxy for “is there a real organization behind this site.” Specialized credentials (medical licenses, financial registrations, security badges) matter for specific categories but don’t generalize. Every legitimate business should have both — their absence is a stronger negative signal than the presence is a positive one.

​
Verification

You can verify our finding yourself by browsing your site. Step 1: Find your about page. Navigate to the canonical paths: https://yoursite.com/about, /our-story, /who-we-are, /company, /our-mission. The first one that returns a real page (not a 404 or redirect to homepage) is your about page. Alternatively, look in the global navigation or footer for an “About” link. Step 2: Find your contact information. Open any sampled page. Look for any of: an email address rendered as a clickable link (mailto: under the hood), a phone number, a “Contact” link in the navigation or footer. The footer is the most common location — most sites duplicate contact info there for every page. Step 3: Inspect via the console. Open DevTools (Cmd+Option+I / Ctrl+Shift+I), Console tab, and run: 
({ mailto: !!document.querySelector('a[href^="mailto:"]'), tel: !!document.querySelector('a[href^="tel:"]'), contactLink: !!document.querySelector('a[href*="/contact"]') })
This returns the same three contact-detection booleans the scanner sees. At least one true value means the page satisfies the contact check. Step 4: Spot-check the about page content. A 50-word about page with three sentences technically passes detection but doesn’t satisfy AI models’ actual credibility threshold. AI parsers look at length, named individuals (founder names, team members), and external proof points (press coverage, customer logos) when grounding trust queries. The scanner doesn’t evaluate any of this — but if your about page is thin, you’ll underperform your detection score in actual AI citation for expertise queries. If your verification disagrees with Cited’s finding, that’s a bug — let us know.

​
Technical detail

Trust and credibility signals trace to Google’s E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness), formalized in the Search Quality Rater Guidelines from 2014 onward. AI training pipelines inherited the same evaluation heuristics — content from sites without basic organizational signals (anonymous, no contact path, no team page) is systematically downweighted as a potential trust risk. About page detection logic. The crawler checks two conditions in OR:
  • The current page’s URL (lowercased) contains any of /about, /our-story, /who-we-are, /company, /our-mission
  • The page’s internal links include any whose href (lowercased) contains one of those same five patterns
Both checks are substring matches, not regex. A URL like /blog/about-our-funding-round would match the /about substring — a false positive in principle, but rare in practice because blog post URLs rarely contain canonical about-page slugs. The signal accepts this trade-off in favor of catching the common case (about page exists somewhere on the site, linked from nav). Contact detection logic. Three independent conditions, OR-combined:
  • document.documentElement.innerHTML.includes('mailto:') — any rendered mailto: reference in the page source
  • Regex tel:|(\+?\d[\d\s\-()]{7,}) — matches either a tel: href or a freeform phone-number-shaped string of at least 8 digits with optional country prefix and common separators
  • Any internal link with /contact in the href (lowercased)
The phone regex deliberately tolerates international and domestic formats: +1 (555) 123-4567, +44 20 7946 0958, 555-123-4567, 555 123 4567. The trade-off is occasional false positives on long number sequences in unrelated content (timestamps, ISBNs, support ticket numbers). Score calculation. Linear addition: about presence (2) + contact presence (2) + both bonus (1) = 5. The cap is hard at 5. There’s no per-page detection — finding either signal on any sampled page satisfies it for the whole site, because trust signals are template-level and propagate to every page. Edge cases the scanner handles:
  • About page accessed only via JS-driven navigation — if the page itself is sampled, the URL check catches it. If only the link is sampled, the internal-link list captures it. About pages that are only reachable by clicking a SPA route never loaded at server render time are missed.
  • Contact info inside an iframe — a chat widget iframe with mailto: links inside it doesn’t satisfy the check, because the scanner reads only the parent document’s HTML. Most sites have at least one mailto: or tel: in the footer outside the iframe, so this rarely fails detection.
  • Phone numbers in copy — a sentence like “We’ve helped 1,234,567 customers” is unlikely to match the phone regex (no separators between digits make it look like one big number, which falls below 8 digits-plus-separators). But a long invoice ID like INV-2026-04-12-555-001 could match. Cited tolerates this; the signal is permissive by design.
  • JS-injected contact widgets — chat platforms (Intercom, Drift, HubSpot Chat) inject contact options at runtime. The scanner reads the post-hydration DOM, so widget-rendered emails count. Widgets that lazy-load on scroll may be missed.
  • tel: schemes with extensions — tel:+15551234567;ext=42 matches the tel: substring and satisfies the check. The scanner doesn’t validate the format.
What this signal does not measure:
  • About page content quality. A 50-word thin about page scores the same as a 2,000-word team page with photos, bios, and press mentions. AI models weight content quality at consumption time; the scanner only checks presence.
  • Specialized credentials. Medical licenses, financial registrations, security certifications (SOC 2, ISO 27001), industry awards. Some categories rely heavily on these and the signal doesn’t model them. The signal is a structural baseline, not a credential audit.
  • Trust badges and customer logos. “As seen in” press strips, customer testimonials, AggregateRating schema, “Trusted by 500+ brands” claims. These all add credibility but aren’t scored here.
  • Privacy policy and terms of service. Legal pages are weak trust signals on their own but their absence is suspicious. The scanner doesn’t check for them currently.
For brands scoring 0-2/5, the highest-leverage fix is adding both an about page and visible contact info to the global template. A footer with mailto:, a phone number, and an <a href="/about">About</a> link satisfies both checks site-wide in one template edit. Most CMSes (Webflow, WordPress, Shopify) ship with these in default themes — custom themes often strip them out and never restore them. See also: Author Attribution, JSON-LD Structured Data, Content Freshness.