Field notes on AI visibility.
Working research, published as we go: how generative engines reshape B2B buying, sourced statistics, and honest guides for SaaS and security teams. No gated PDFs, no fabricated case studies — specimens are labeled specimens.
What is generative engine optimization (GEO)?
The 2026 field guide: definition, why 94% of B2B buyers using LLMs changes distribution, how engines pick brands, and the five-part workflow.
How ChatGPT chooses which brands to cite
Two memories — training data and Bing-backed retrieval — and the ranked signals: consensus, authority, extractability, access, entity clarity.
The 27-point AI visibility audit checklist
Our paid-audit checklist, published in full: baseline, crawler access, schema, extractable content, third-party footprint — with scoring.
GEO vs SEO: what actually changes
Same inputs, different scoreboard. The honest comparison table, the 2026 evidence, and how to split a small budget.
llms.txt: an honest guide, with a working example
What the file is, what engines actually do with it (insurance, not a hack), and a complete worked example for a B2B SaaS.
How to measure AI search visibility
The share-of-voice protocol behind our audits: frozen prompt panels, weighted scoring, framing logs — free to copy.
AI search statistics 2026 — every number sourced
The reference sheet: adoption, buyer behavior, conversion economics and citation mechanics — 20+ figures, each dated and attributed. Free to cite.
How to choose a GEO agency (an honest buying guide)
The provider landscape compared factually — enterprise firms, SEO agencies, monitoring tools, studios — plus seven questions that expose a weak vendor. We're on the list, and we say so.
AI visibility for cybersecurity companies
Security buyers moved into AI research faster than anyone. The shortlist problem, the reputation-prompt trap, and patterns from our audit ledgers.
AI visibility for B2B SaaS startups
51% of buyers now start in a chatbot and the traffic hides as 'Direct.' What the re-routed funnel means at seed, Series A, and growth stage.
New entries are logged as the research happens. Every statistic carries a named source; every claim about engine behavior is dated, because engines drift.