Reference

Glossary

This glossary defines the terms WhosBest.org uses when it ranks providers and when it talks about AI agent readiness. The first group covers how rankings are built — Winston scores, evidence signals, editorial lists versus algorithmic scorecards, and the no-pay-to-play rule. The second group covers the discovery and machine-readability conventions that AI crawlers, assistants, and agents rely on: llms.txt, AGENTS.md, markdown mirrors, content negotiation, structured data, and crawler policy. Every term has a stable anchor (for example /glossary#winston-score) so you can link to a single definition. If a term you need is missing, let us know.

Rankings & methodology

#Winston score
The 0–100 composite score Winston AI assigns a provider within a category and, where applicable, a city. It blends weighted evidence signals — reviews, response rate, web presence, social signals, and data freshness — according to the category's published weights. Scores refresh continuously as new evidence arrives.
#Winston AI
WhosBest.org's autonomous research agent, operated by WTE Solutions. Winston gathers publicly verifiable evidence, scores providers against the methodology, and drafts the algorithmic scorecards published on the site.
#Evidence signal
A single publicly verifiable data point that feeds a Winston score — for example a regulatory filing, a certification, a review count, a response-time measurement, or a social-profile metric. Every ranking cites the signals behind it so readers can verify the reasoning.
#Editorial ranking
A hand-curated top list authored by WhosBest Research or drawn from a trusted source such as Gartner or TechCrunch. Editorial rankings explain the selection criteria in prose and are labelled with a gold "Editorial" badge. Browse them at /lists/type/editorial.
#Algorithmic scorecard
A live ranking generated entirely from Winston scores, with per-signal breakdowns for each provider. Scorecards are labelled with a magenta "Winston AI" badge and live at /best/{category} and /best/{category}/{city}. Browse them at /lists/type/algorithmic.
#Community list
A user-curated list created by a registered member, where other members vote items up or down. Community lists are labelled with a teal "Community" badge, live at /c/{slug}, and are never blended into editorial or algorithmic rankings.
#Top list
The generic name for any published ranking page at /top-lists/{slug}. A top list has a title, a source, an optional year and sector, a summary, and ordered entries (position, company, designation, website).
#Sector hub
A browse page that groups every ranking in one vertical — healthcare, technology, professional services, education, local business, AI agents & automation, analytics, or cloud platforms. Sector hubs live at /browse/{sector} and /lists/sector/{sector}.
#Methodology
The published rules Winston AI follows to score providers: which data sources are used, the four scoring pillars (Reputation & Trust, Expertise & Credentials, Transparency & Responsiveness, Evidence of Results), how weights vary by category, and the fairness and update policies. Read it at /methodology.
#Pay-to-play
A ranking model in which vendors can buy a higher position. WhosBest.org does not operate this way: paid features such as enhanced listings never influence rank, and every provider is scored by the same methodology.
A clearly labelled paid slot shown alongside, but never inside, an organic ranking. Sponsored placements are visually separated and marked so readers and AI agents can tell them apart from earned positions.
#Provider profile
The page at /providers/{slug} describing a single ranked company: its categories, cities, current Winston scores, evidence summary, and claim status. Providers can claim and verify a profile to correct facts without affecting their rank.

AI agent readiness & discovery

#Agent readiness
How easily AI crawlers, assistants, and autonomous agents can discover, read, and correctly cite a website. It covers crawler policy, discovery files (llms.txt, AGENTS.md, sitemap.md), markdown mirrors, structured data, and HTTP behaviour. The free AI Readiness Checker scores any public site from 0 to 100.
#llms.txt
A plain-markdown file at /llms.txt (llmstxt.org spec) that tells language models what a site is and where its most useful pages are: an H1, a blockquote summary, short prose, and H2 sections of - [name](url): note links, with an optional ## Optional section. WhosBest.org's copy is at /llms.txt.
#llms-full.txt
The long-form companion to llms.txt: a single plain-text document containing the site's key content in full so an agent can load it in one request. WhosBest.org's version lists every published ranking with its URL and top entries, capped at roughly 100,000 characters, at /llms-full.txt.
#AGENTS.md
A markdown skill file served at /AGENTS.md that tells agents how to use a site or API: installation or access, configuration (user-agent policy, rate limits, content negotiation), usage examples, data and attribution rules, and contact details. WhosBest.org's copy is at /AGENTS.md.
#sitemap.md
A markdown rendition of the sitemap — an H1, a blockquote summary, and H2 sections of links — so agents that prefer markdown can inventory the site without parsing XML. See /sitemap.md; the XML sitemap remains at /sitemap.xml.
#Markdown mirror
A markdown twin of an HTML page, reachable by appending .md to the page path (for example /top-lists.md) or via content negotiation. Mirrors carry the same text as the HTML page plus YAML frontmatter (title, description, canonical URL, last updated) so they can be cached and cited safely.
#Content negotiation
The HTTP mechanism by which a client asks for a preferred representation using the Accept header. Sending Accept: text/markdown to a WhosBest.org page returns the markdown mirror with Content-Type: text/markdown; charset=utf-8; responses include Vary: Accept so caches keep HTML and markdown apart.
#JSON-LD
JavaScript Object Notation for Linked Data — the <script type="application/ld+json"> block that expresses a page's meaning in schema.org vocabulary. WhosBest.org emits ItemList for rankings, BreadcrumbList, Organization, WebSite, and DefinedTermSet on this page.
#Structured data
Machine-readable annotations (JSON-LD, microdata, or RDFa) that describe entities on a page — a ranking, a company, a definition — so search engines and AI agents can extract facts without guessing from layout.
#Canonical URL
The single preferred address for a page, declared with <link rel="canonical">. On WhosBest.org every canonical starts with https://www.whosbest.org/; agents should cite that form rather than the apex domain, which redirects.
#Open Graph
The og:* meta tags (title, description, image, url, type) that give social networks and link-preview tools a consistent card for a page. Agents often read them as a quick summary.
#robots.txt
The plain-text crawler policy at the site root. WhosBest.org allows all well-behaved crawlers, names major AI agents explicitly, blocks only /sadmin/, points to the sitemap, and publishes content signals. See /robots.txt.
#AI crawler (GPTBot, ClaudeBot, etc.)
A bot operated by an AI company to fetch web pages for training, search grounding, or live browsing — for example GPTBot and OAI-SearchBot (OpenAI), ClaudeBot and Claude-User (Anthropic), PerplexityBot, Google-Extended, Applebot-Extended, Bytespider, CCBot, meta-externalagent, and Amazonbot. WhosBest.org welcomes them in robots.txt.
#Content-Signal
A robots.txt directive (Cloudflare's Content Signals Policy) that states how a site's content may be used: search, ai-input (grounding answers), and ai-train. WhosBest.org publishes Content-Signal: search=yes, ai-input=yes, ai-train=yes.
#MCP server card
A small JSON document (typically at /.well-known/mcp.json) advertising a Model Context Protocol server — its name, endpoint, transport, and capabilities — so agent runtimes can discover tools a site exposes.
#A2A agent card
The JSON discovery document of the Agent-to-Agent protocol, usually at /.well-known/agent.json, describing an agent's identity, skills, endpoint, and authentication so other agents can delegate tasks to it.
#Agent skills index
A listing of the skills or capabilities a site offers to agents — often an AGENTS.md or a /.well-known/skills index — giving each skill's purpose, inputs, and example calls. WhosBest.org's skills (rankings lookup, readiness check, markdown mirrors) are described in AGENTS.md.
#Text-to-HTML ratio
The share of a page's bytes that is visible text rather than markup or script. Agent-readability audits warn when it falls below roughly 10–15 %, because the useful content is buried; markdown mirrors sidestep the problem entirely.
#Accessibility tree
The browser's semantic model of a page — headings, landmarks, links, labels, tables — exposed to assistive technology and increasingly to browsing agents. Clean headings and labelled regions make the tree, and therefore the page, easier for agents to navigate.
#AI Readiness Checker
WhosBest.org's free tool at /check (API: POST /api/v1/validator/check) that scores a public website's robots.txt, llms.txt, structured data, homepage clarity, and discoverability and returns a 0–100 result in under 30 seconds with no sign-up.

See also