🔍
Full-Page SEO + AI Scanner
Fetches an admissible public URL, parses the returned HTML, and runs the configured analyzers. Business-type suggestions come from observed page evidence and must be confirmed before schema selection.
- Schema markup analysis (JSON-LD, Microdata, RDFa)
- Meta tag scoring (title, description, OG, canonical)
- Content depth (word count, heading hierarchy, images)
- AI Readiness (quotations, statistics, citations)
- Technical (HTTPS, mobile, sitemap, robots.txt, load time)
- Security headers (7 headers scored)
- Google PageSpeed Insights + Core Web Vitals
- AI crawler access (robots.txt + llms.txt detection)
- Link + image optimization audit
- Business-type suggestions from 13 detector patterns; confirmation required
🛠
40-Key Schema Engine
Prepares structured-data drafts from 40 template keys: 39 industry-specific keys plus a general fallback. The current governed generator emits one fact-authorized business node selected from the confirmed business type. Unconfirmed service, FAQ, rating, hour, coordinate, and identity claims are omitted or withheld.
- 39 industry-specific keys plus one general schema-template fallback
- The current generation path emits the resolved template's primary business type as one homepage node
- Services, FAQs, ratings, business hours, coordinates, and sameAs identities require admitted customer facts rather than template examples
- Preview mode — see output before deploying
- Stores in optimizations table with full audit trail
40 schema template keys; industry selection uses the confirmed business type
⚡
AI Auto-Fix Engine
Generates SEO fixes and keeps prepared work separate from delivery. An approved exact revision proceeds only with current authorization for that revision and a connection proven capable of the effect; otherwise it remains for review or customer implementation. A JavaScript-snippet effect is client-side and cannot be verified in raw origin HTML. WordPress and Cloudflare require their own successful capability and public-origin readback checks.
- Meta descriptions — AI generates 120-160 char descriptions
- Title tags — AI generates 50-60 char optimized titles
- Alt text — context-based alt text from filename + page content
- Internal links — AI prepares cross-page suggestions with anchor text; applying an exact revision still requires current authorization and a connection proven capable of that effect
- Batch mode — prepares meta descriptions, title tags, and image alt text; internal-link analysis and delivery remain a separate path
- Recorded reversal — available only where the connector and effect support a bound prior state and reversal path
- Eligible effect records retain the known prior state and proposed exact revision; unavailable evidence remains unknown
- 3 deployment paths: WordPress REST, JS snippet, Cloudflare Worker
Three governed paths with different capabilities: the JS snippet applies client-side and cannot be verified in raw origin HTML; WordPress REST and Cloudflare require connector-specific capability and public-origin checks
✍
GEO Content Engine
Prepares article drafts under source, customer-fact, authority, admission, and publication controls. The internal GEO score is a model-based screen, not factual certification or a citation forecast. Publication requires current authorization and a destination connection proven capable of that effect.
- At least three exact researched source URLs must be declared and bound to independently checked source evidence
- GEO and E-E-A-T scores are screening signals for editorial triage, not hard admission gates or factual evidence
- Site-fact extraction — records observed public-site evidence; it reduces unsupported drafting but does not guarantee zero errors
- Source and model-review pipeline — model verdicts remain separate from fetched-source proof and human review
- Article and FAQ schema can be prepared when supported by the admitted visible content and vocabulary policy
- Eligible WordPress publication requires current authorization and a connection proven capable of publication; otherwise work remains for review or customer implementation
- Scheduled publication evaluates entitlement, exact-revision authority, destination capability, and runtime policy at execution time
- Public blog API + RSS feed for static sites
- Prompt-bound routes sanitize the specific customer strings they interpolate; this is not a claim that every application input passes one universal sanitizer
Model-based GEO screening plus distinct source, factual-unit, authority, admission, and publication controls
🔎
Multi-Model Fact Checker
Up to three configured providers (Claude, GPT, and Gemini) can return model verdicts for a claim. The route records response count, agreement, disputed and unverifiable states, and fails closed when the live-model floor is not met. Model agreement is not independent factual proof.
- Up to three model verdicts (Claude + GPT + Gemini); fewer than two valid live responses remains an insufficient-model state
- Agreement ratio scoring (0.0-1.0)
- Auto-flags disputed claims
- Links back to blog posts for post-level status
- Resolution workflow for flagged claims
- Governed content generation invokes the applicable review path; missing provider capability cannot clear publication
Up to three model verdicts with an explicit insufficient-model state; fetched-source and human controls remain separate
📡
Multi-Engine Citation Monitor
Queries ChatGPT, Claude, Gemini, Perplexity and Grok through their configured integrations on a monthly schedule, then records the returned brand mentions, URL citations, sentiment, competitor URLs and confidence scores. Microsoft Copilot is not currently supported. Each result is a sampled observation, not a continuous view of every user answer.
- ChatGPT, Claude, Gemini, Perplexity, and Grok use configured provider integrations; each observation records the model identifier actually used
- Microsoft Copilot is not currently supported; no queued or promised Copilot pipeline is implied
- Brand mention detection with domain matching
- URL citation detection
- Sentiment analysis (positive/negative/neutral/mixed)
- Competitor URL extraction from responses
- Detection confidence is stored with the observation; it is not a provider probability or outcome guarantee
- Configured engine-profile weights can affect aggregate reporting; they are not presented as weights fitted to customer outcomes
- Cost tracking per check (cents)
Monthly observations from 5 supported AI engines; Microsoft Copilot is not currently supported
💡
Citation Gap Intelligence
Prepares a configured monitoring-prompt set, records sampled responses where the named brand was absent, and can use an observed query as a candidate content brief. A generated draft does not fill an external-engine gap or promise later inclusion.
- AI-generated monitoring prompts (30 per site, 6 categories)
- Gap detection — queries where brand isn't mentioned
- Per-engine gap breakdown (missing from ChatGPT but present in Gemini)
- One-click article generation from any gap query
- Competitor mention tracking — see who IS being cited in your absence
- Gap scoring (0-100) for prioritization
Records sampled brand absences and prepares reviewable content candidates without assigning causation
🌐
Entity Graph + Wikidata Linking
Extracts observable entities from structured and page metadata, then performs a bounded exact-name Wikidata candidate search. Its internal evidence score does not prove knowledge-graph inclusion, identity, authority, retrieval, or external-engine use.
- Entity extraction from JSON-LD, Microdata, OG tags
- Person and Organization entity classification
- Automatic Wikidata API matching (top 3 candidates per entity)
- Entity score: coverage + types + Wikidata matches + sameAs links
- Unmatched entity reporting for remediation
👤
Author Entity Management
Creates and manages author profiles and prepares schema.org Person markup from confirmed credentials, identifiers, affiliations, and social profiles. The record does not establish how a private engine weighs author authority.
- Full CRUD for author profiles
- Auto-generates Person schema with credentials, worksFor, sameAs
- Wikidata ID linking
- Primary author designation per site
- Social profile integration (sameAs URLs)
🕐
Content Freshness + IndexNow
Tracks observed content revisions with SHA-256 hashes, classifies structural and content changes, and computes an internal freshness priority. Where configured, an IndexNow request records provider acceptance separately and does not prove crawling or indexing.
- SHA-256 content + heading hashing for change detection
- Structural vs content-only change classification
- Freshness score decay (100 at 7 days → 10 at 365+ days)
- Refresh priority: critical / high / medium / low / fresh
- Change history (last 20 changes per page)
- IndexNow submission to 3 endpoints (API, Bing, Yandex)
🔒
AI Crawler Access Control
Prepares provider-specific crawler-policy artifacts from four stored strategy IDs. The legacy allow_rag_only ID does not claim RAG-only enforcement. Generation is not deployment, and public rule readback is not provider adherence, indexing, retrieval, citation, or training-use proof. Any site effect remains subject to current authorization and proven connector capability.
- Four stored strategy IDs plus a custom policy path
- Provider-specific restrictions that leave ordinary Googlebot and bingbot policy unchanged
- Named-group reconciliation warning so private wildcard paths are not silently exposed
- Informational llms.txt and authenticated AI-sitemap previews; neither grants crawler access
- Current IETF Internet-Draft Content-Usage preferences; no enforcement claim
- Exact manual installation package when no governed connector write path is proven
- Public readback of required artifacts, bounded to the bytes and values observed
Provider-specific policy preparation plus public rule readback; no downstream provider-outcome claim
📄
Passage-Level Citability Analysis
Splits observed pages into heading-delimited passages and reports a model-based GEO score from configured features. Weakest and strongest labels are internal comparisons, not proof that an AI engine will retrieve, quote, or cite a passage.
- Heading-based passage extraction (H1/H2/H3)
- Per-passage GEO scoring (quotations, statistics, citations)
- Internally labelled passage candidates (configured score >= 60)
- Weakest/strongest passage highlighting
- Average passage score for page-level assessment
Passage-level internal screening; no external retrieval or citation forecast
⚖
Competitor Gap Analysis
Runs the available-evidence scan on a submitted competitor URL and compares the five AI Readiness families. Internal gap and advantage labels do not establish why an external engine selected either business or prescribe a guaranteed outcome.
- Side-by-side five-family AI Readiness comparison
- Gap severity: critical (>30 points), significant (>15), minor (>5)
- Advantage detection where you lead
- Overall verdict: competitor_leading / client_leading
- Historical comparison tracking
🗒
Topic Cluster Mapping
Groups observed sitemap and page paths into candidate content clusters, maps potential pillar and supporting pages, and prepares topic and internal-link suggestions. The internal coverage score does not prove topical authority.
- Auto-discovery from sitemap.xml or crawled links
- Path-based cluster grouping
- Pillar page + supporting page mapping
- Gap topic identification
- Suggested articles for each cluster
- Coverage + authority scoring
🎥
Video Transcript GEO Analysis
Analyzes available video metadata, transcript, and chapter evidence, reports a bounded model score, and prepares VideoObject markup supported by visible content. It does not publish or promise discovery or citation.
- YouTube transcript extraction via caption API
- Chapter detection from video descriptions
- GEO scoring of transcript (quotations, statistics, citations)
- VideoObject schema generation with transcript + Clip parts
- Review suggestions based on the configured transcript-screening features
📈
AI Traffic Attribution
Classifies observed requests and visits from supported referrer and user-agent evidence. Missing or stripped attribution remains unavailable, and traffic-source classification is separate from the five-engine monthly citation-monitoring support list.
- 13 user-agent patterns (GPTBot, ClaudeBot, PerplexityBot, etc.)
- 13 referrer patterns (chatgpt.com, gemini.google, perplexity.ai, etc.)
- Traffic summary with AI vs organic vs direct breakdown
- Daily trend tracking
- Top AI-referred pages
- Lightweight tracking snippet (served as JS)
🔌
3-Path Deployment System
Three connection paths with different capabilities. WordPress and Cloudflare effects require a successful capability probe for the requested operation. A JavaScript snippet can apply supported client-side presentation changes after a visitor loads it; that client-side effect cannot be verified in raw origin HTML and is not a public-origin readback.
- WordPress REST — supported operations only after the connection proves the required capability
- JS Snippet — supported client-side presentation changes; no server-origin effect claim
- Cloudflare Worker — supported edge operations only after a successful capability probe
- Site verification (file upload or meta tag)
- Connection testing (auth, posts, pages, media)
- Snippet activation evidence records a client-side load; it is not public-origin verification
Every exact site effect requires current authorization for that revision and a connection proven capable of the effect
⚙
Engine Profile Calibration
Engine profiles store operator-configured weights, bias notes, freshness settings, and citation-style metadata used by parts of aggregate reporting. Those settings do not disclose an external provider's private preferences and are not automatically calibrated to customer outcomes.
- Administrator-managed per-engine weight configuration
- Bias notes and citation-style metadata
- Freshness setting with recorded update time
- Aggregate calculations preserve the configured profile inputs
- No claim that configured weights are fitted to customer outcomes
Configured profiles, not a published or outcome-fitted model of provider behavior
🗎
PDF Audit Reports
Generates branded PDF audit reports with the served score generation, evidence states, issue lists, and recommendations. Readiness, SEO Health, and AI Presence remain separate, and the report does not promise a projected external result.
- Full site audit PDF with all five AI Readiness family scores
- Score bars, stat boxes, issue tables
- Recommendations retain their evidence and non-predictive weight disclosure
- Dark-themed professional design
🕒
Automated Pipeline Scheduler
The scheduler catalog defines 57 named jobs across its configured cadences. A scheduled evaluation is not an external effect: every exact effect still requires its applicable entitlement, authority, proven destination capability, runtime policy, and lock.
- Health checks every 6 hours (2/8/14/20 ET)
- Freshness scans every 3 days (3:00 AM ET)
- Citation monitoring monthly
- Eligible scheduled-content publication is evaluated hourly; blocked or incomplete work remains held
- Auto-fix candidates are evaluated weekly Wednesday (5:00 AM ET); advisory categories do not gain execution authority from the schedule
- Advisory lock protection against duplicate runs
📷
Image SEO Analyzer
Scans every image on a page for SEO issues: missing alt text, generic auto-generated alts, missing dimensions (CLS), and missing lazy loading attributes.
- Alt text presence, length, and quality check
- Generic alt detection (auto-generated patterns)
- Width/height attribute audit (CLS prevention)
- Lazy loading attribute check
🛡
Security Header Analysis
Reports seven configured security-header checks (HSTS, CSP, X-Content-Type-Options, X-Frame-Options, Referrer-Policy, Permissions-Policy, and X-XSS-Protection). Security evidence belongs to the separate SEO Health composition; ClickRadius does not claim that an external AI engine treats the header score as a trust signal.
- 7 headers with individual weights
- Missing header identification
- Severity-based issue reporting
- Not a 10% AI Readiness family and not blended into AI Presence
⚡
Core Web Vitals Integration
Requests Google PageSpeed Insights evidence when that provider path is available and records returned Lighthouse, field, and lab metrics. Missing or failed provider evidence remains unavailable rather than becoming an invented score.
- Lighthouse performance score (0-100)
- Lighthouse accessibility score (0-100)
- CrUX field data (LCP, CLS, INP, FID, TTFB)
- Lab data fallbacks (SI, TBT, FCP)
💳
Stripe Billing + Reseller Portal
Built-in Stripe integration for direct and wholesale billing. The reseller portal supports multi-tenant client management, scope-adjusted pricing under the executable commercial policy, and approved client-facing branding.
- Stripe webhook integration (raw body handling)
- Reseller portal with client management
- No agency or per-site setup fees; the standard wholesale base is $200/site/month, each site has a four-month initial term from confirmed-payment activation, and public payment activation remains closed
- JWT-based authentication with role-based access
💚
Health Monitoring + Alert Delivery
Scheduled health checks record available HTTP, returned-markup, metadata, HTTPS, header, and score-trend evidence for eligible active sites. Configured email and Slack paths can carry alerts; provider delivery and unavailable observations retain separate states.
- HTTP status monitoring (2xx/4xx/5xx detection)
- Returned-markup JSON-LD parsing; JavaScript-snippet effects remain client-side and cannot be verified in raw origin HTML
- Score drop detection (≥10 point warning, ≥15 alert, ≥25 critical)
- Email + Slack alert delivery with per-site notification preferences
- Health status dashboard (healthy/warning/critical per site)
- Alert acknowledgment workflow
📩
Verified Score Access Request
After a server-generated score receipt exists, /public-score/access/request claims its immutable server-side report for the submitted email, creates the lead from that score receipt, and sends the verification message through the configured delivery path.
/public-score/access/request requires a valid, unexpired score-receipt token and email claim- The lead uses the immutable report's source URL and server-held score headline rather than a browser-supplied score
- Repeated access requests are keyed by report, email hash, and scan kind, with a resend cooldown
- The active access path creates a lead only from the claimed server report; a browser-supplied score is not accepted