GEO case studies are full loops written up end to end: the starting measurement, what the answers actually said, the changes that were made, and the re-measurement afterwards — including the parts that did not move. They exist because the interesting question in this discipline is rarely what a tactic is, but whether it did anything in a specific market.
5 guides in this topic · all 53 guides · every topic
The best-documented AI visibility loss on record: traffic down 49%, subscribers down 21%, and a lawsuit that names the mechanism. Every number sourced — and the test for whether your brand shares the exposure.
Vercel's CEO published the numbers himself: ChatGPT went from under 1% to 10% of new signups in months. What the best-documented AI visibility win shows — and why referral counts understate it.
It trained the models that replaced its visits: questions fell from ~200k a month to 2009 levels. The sharpest proof that feeding the model and being recommended by it are different things.
This is a composite, illustrative worked example — built from the patterns we see repeatedly — of a B2B SaaS brand that went from barely appearing in AI answers to being a recommended option on its core buying prompts inside a single quarter.
From 12% to 41% mention rate on core buying prompts. The full worked example: baseline, page rewrites, corroboration campaign, and re-measurement.
Check any page for AI readiness in about a minute — free, no signup, full checklist either way.