Research, methods, failures, and build notes for teams earning AI visibility and shipping AI products. No recycled playbooks. No unexplained scores.
Decision-stage visibility and evidence architecture.
Methods, experiments, and corrections.
See the evidence behind every number.
What broke, why it broke, and what changed.
A page took 64,191 Search Console impressions and zero clicks in 87 days. What the data shows, what it cannot show, and why it explained none of our lost traffic.
Read the field noteThree ways in
The library is organized around decisions, not content categories invented for search engines.
Start with decision-stage visibility, evidence architecture, and what an AEO score can and cannot tell you.
Read foundationsInspect how Radar scores, how entity resolution works, and how we test our own audit architecture.
Read the methodsFollow experiments across engines, referrals, citations, and volatile recommendations.
Read field notesLatest notes
The most recent field notes, newest first.
AI in a B2B company should run as a directed loop: a human sets the question, Radar measures, Vector qualifies, Hive builds, and a human decides. Full model.
Read articleHow to evaluate a GEO agency before you sign. Four green flags, five red flags, and the baseline evidence any credible AI search partner should show you first.
Read articleAn AI visibility audit can look credible while measuring the wrong company. What one failure taught us about entity resolution and audit integrity.
Read articleCan you trust the score an AI visibility tool gives you, tomorrow as well as today? We test our own scoring weekly and audited our whole stack. Here is the proof.
Read articleWhat is a good AEO score? See what AEO checkers actually measure, how grades work, real data from 59 audits, and the fixes that raise a failing score.
Read articleBuyers ask AI which option to choose before they reach your site. Decision-stage AI visibility is whether you win that recommendation, not just appear.
Read articleGrok answers brand questions for 117M monthly users per the SpaceX S-1. We measured what it says, and it got our own brand wrong. Check yours today.
Read articleYou cannot control what AI recommends about your brand. Here is why AI answers are volatile by design, and the five things to measure instead.
Read articlePublishing more content will not make you visible in AI answers. Evidence architecture, your claims, entities, sources, and structure, is what gets you cited.
Read article