Reach & AI Readiness

20% of overall

When developers (and their AI assistants) go looking, do they find you — and what they need?

What it measures

Three groups of signals: (1) AI readiness — is there an llms.txt, AI-friendly pages, and a hosted MCP server? (2) External mentions — Reddit, Hacker News, dependents on the package registries, Stack Overflow tag activity. (3) Dev-focused content — what fraction of the blog and YouTube channel is for developers vs. marketing.

Why it matters

AI assistants are now a primary discovery surface. An llms.txt + hosted MCP means ChatGPT, Claude, and Cursor can answer real questions about your product instead of hallucinating from old blog posts. External mentions are the word-of-mouth proxy that doesn't show up in your own analytics.

Sub-metrics

MetricWhat we readWeightSource
llms.txtPresence, sections, links, llms-full.txt, size35% of AI-readinessLive HTTP probe of common paths
AI-friendly pagesMarkdown/JSON variants of docs pages20% of AI-readinessDocs crawl
Hosted MCP serverPOST initialize to mcp.<domain>, /mcp, /sse, etc.45% of AI-readinessJSON-RPC probe (accepts auth-protected 401)
External mentionsReddit + Hacker News search hits45% of externalReddit + HN search API
Package dependentsCount of public packages that depend on yours40% of externalnpm / PyPI / etc.
Stack OverflowTag question volume + recency15% of externalStack Exchange API
Dev content ratio% of blog posts and YouTube videos classified as developer content15% of pillarFirecrawl + LLM classifier

How it's weighted

External mentions carry the most weight, with AI readiness close behind — a hosted MCP server is the single strongest signal inside that bucket. Dev-focused content nudges the score up when the blog and YouTube channel are clearly for developers rather than buyers.

Best practices

  • Publish /llms.txt at the root of your docs site. Add /llms-full.txt with the long form.
  • Host an MCP server at mcp.yourdomain.com — even a docs-search MCP is a strong signal.
  • Open-source one MCP server in your GitHub org so the community can ship their own variants.
  • Keep the blog dev-heavy. Marketing posts dilute the signal — split them onto a separate /company feed if you must.
  • Claim and watch your Stack Overflow tag. Answer the first few questions yourself to set a quality bar.
  • Make sure your package's README links to docs, repo, and Discord — that's what dependents will see.