SITEBORNE SIGNAL MODULE
SignalSearch — meaning-based website search
Finds relevant information when visitors use different wording than the page.
BUSINESS OUTCOME
What this capability changes.
- Improve findability
- Reveal content gaps
- Preserve conventional search
Best for
Content-rich businesses, Professional firms, Software companies
From $1,500
Module-only monitoring: from $100/monthA narrow add-on scoped to this module alone—checking its allowance, fallback, and failure state. It is separate from and smaller than a full Care plan, which covers the whole website.
PLAIN LANGUAGE, TECHNICAL PROOF
What SignalSearch does.
Search that understands what someone means, even when they do not use the exact words written on the site.
Visitors find relevant services and guidance faster, and empty searches reveal which explanations the site is missing.
Example workflow
A visitor searches “what will this cost after launch” and reaches pricing, care, and allowance guidance even though that exact phrase is not a page title.
FOR TECHNICAL READERS
How the control model works
A compact embedding model converts the query to a vector, Vectorize ranks approved content, and ordinary keyword search remains available when semantic search is capped or unavailable.
Failure path: Keyword search, filters, and curated navigation.
INTERACTIVE WALKTHROUGH
Use SignalSearch before discussing it.
Search for a real concern such as ownership, post-launch cost, qualified leads, monitoring, or accessibility.
The interaction uses your input and identifies whether it runs locally, checks this live site, or previews a production control. It does not disguise seeded output as intelligence.
System view
Required
Cloudflare Workers, Cloudflare Vectorize, Cloudflare D1
Optional
Cloudflare Workers AI, Voyage AI
Fallback
Keyword search, filters, and curated navigation.
Cloudflare Workers, Cloudflare Vectorize, Cloudflare D1, Cloudflare Workers AI, Voyage AI
2026-08-02
Provider limits, terms, model availability, and pricing may change. Production accounts remain client-owned.
Data classification
- Public search queries
- Public content metadata
LAUNCH ALLOWANCE
Semantic retrieval
Prototype and low-volume vector storage and query capacity.
View all referenced allowances
- Semantic retrieval: 30M queried vector dimensions/month and 5M stored vector dimensions.
- Embedding bootstrap: First 200M tokens free for several current models; not a monthly renewal.
Upgrade triggers
- vector dimensions — 75%. Review index size and paid successor.
- empty-result rate — above 20%. Improve synonyms and content coverage.
View implementation proof
Decision rationale
The implementation favors semantic HTML, static rendering, typed content, and isolated on-demand routes. Optional AI never controls pricing, eligibility, or emergency routing.
Controls
- WCAG 2.2 AA target
- Provider-neutral adapters
- Usage caps and kill switches
- Client-owned production accounts
- Conventional fallbacks
Verified evidence
- The browser demonstration runs from a clean clone without a paid credential.
- Kill switch and allowance states are typed configuration.
- Primary navigation and contact paths remain independent of this module.
Known limitation
External provider allowances and production behavior require owner accounts and deployment-time verification.