Blog
Field notes on AI visibility — the plumbing, the measurement, and the honest error bars.
- 6 min readFeatured
You don't need forty schema types for AI search — you need three done well. Here's what Organization, Product/Offer, and FAQPage JSON-LD do for AI engines, copy-paste Next.js App Router examples, and the markup that's safe to skip.
json-ldstructured-dataschemaai-search We pointed a real browser agent at storefronts and gave it four buyer tasks: find a product, parse the price, choose between options, and start checkout. Here's where agents fail most — JS-only forms, unparseable pricing, login walls — and how to fix each.
agentic-commerceagentsconversionMost AI visibility tools hand you one number. But LLM answers vary run to run, so a single score is a sample, not a measurement. Here's why bare numbers mislead, what run-to-run variance actually looks like, and how a confidence interval fixes it.
ai-visibilityconfidence-intervalsmethodologyA dated, verifiable reference for AI crawler user agents — OpenAI, Anthropic, Perplexity, Google, Apple, and the rest — with each operator's official IP-range source, robots.txt token, and a warning about user-agent spoofing. Reviewed monthly.
ai-crawlersuser-agentsrobots-txtA step-by-step manual check: read your live robots.txt for OpenAI's three agents, test whether your content survives without JavaScript, run the markdown-emptiness test, and rule out a WAF challenge page — in about ten minutes, with curl.
chatgptgptbotrobots-txtrenderingllms.txt is a curated markdown map of your site for language models. Here's exactly what the spec requires, an annotated example you can copy, a validator checklist, and an honest read on whether any AI engine consumes it yet.
llms-txtai-crawlersai-visibility