Welcome to episode 29 of The GEO Show, the briefing on generative engine optimization, LLM citations, and AI search visibility. Brought to you by GEOforge.
In this episode, Paris Childress, founder of Hop AI and co-founder of GEOforge, covers ten stories from the September 12 weekend brief. The throughline: low-confidence AI visibility scores from tiny prompt sets and one-run sampling are plaguing GEO measurement.
In this episode:
- 📊 Peak AI Referrals + My Website
Peak connects GA4 for sessions, engagement, conversions, and revenue, and My Website joins server logs, prompt tracking, and GA4 from crawl to cite to visit to conversion. Visibility-only dashboards are becoming insufficient. Full-funnel attribution is the destination. - 🧭 Ahrefs Brand Radar → programmable Topics
Ahrefs shipped an AI response API, entities/exports, Report Builder, and a Topics report that groups queries by parent topic to show angles a brand owns or misses. An SEO incumbent is absorbing GEO monitoring. - 📉 THROUGHLINE: Otterly — small prompt sets are noisy
252,407 answers / 520 prompts / 7 engines. Moving from 10 to 100 prompts cut brand-coverage variation about 73%. 48% of cited URLs appeared once; the top 1% drove 44% of citations. Still one run per prompt per engine per day. GEOforge's bar: about 30–40 runs for 95% confidence / ±2% MOE. Cheap tools price on one-run sampling. - 📄 Rank Masters citation concentration
6,831 owned citations / 64 commercial prompts / 90 days. Five URLs produced 52% of citations; 20 URLs produced 90%. Gemini 64% / Perplexity 20% / ChatGPT 15%. Single B2B site caveat, but a stark power-law GEO content signal. - 🔁 Pallas AI launch
Nine-engine monitoring, Shopify, marketing context OS, fetchable/chosen/extractable gates, and a watch→decide→act→review loop. Architecture close to GEOforge. Paris likes the loop. - 🔀 Brand Ghost engine disagreement
7,473 citations / 4 engines. Source overlap ~12.9%. Recommendation overlap 11.2%. 50.7% of comparable sets shared no brands. Blended SOV can hide engine risk (client case: AI Overviews +50% while ChatGPT and AI Mode slightly negative). Split ChatGPT vs Google AI Mode/Overviews. - 🧾 M11 Labs — agentic commerce ranks proof
Stealth exit with an agentic trust platform. Skincare study: 600+ brands / 15k+ evidence captures. 47% lacked independent evidence; 58% of claims cite non-public research; 1 in 20 evidence links confirm the claim. Full report Sep 16. Paris: AI visibility is early innings and may look like rank tracking vanity later. - 🛒 Instacart Clementine
Conversation, recipe, or list becomes a ready-to-buy cart across US and Canada. Cart tech also on retailer properties. Possible no referral click between recommendation and transaction. Prep product data and attributes for agents. - 📈 Google AI Max vs DSA
DSA sunset delayed to Feb 2027; ACA and campaign broad match auto-migrate to AI Max in Sep 2026. Internal non-retail: +7% conversions/value at similar CPA/ROAS. AI choosing queries, creative, and landing pages. - 🧰 Elmo open-source GEO tracking
MIT self-hostable tracking across ChatGPT, Claude, Perplexity, Gemini, Copilot, Grok, and AI Overviews (~319 stars / 75 forks). Prompt tracking is commodity. Moat: proprietary knowledge, statistical certainty, decision intelligence, execution, off-site citation building.
If this briefing is useful, subscribe to The GEO Show so we can keep walking through the platform changes and citation mechanics that actually move AI visibility.
💬 Question: If your AI visibility tool still samples about 10 prompts once a day, what confidence interval do you think you are buying?