Flight Log #50

In this issue:

Happy Sunday, everyone! In this issue, we look at why controlled A/B testing is essential for proving AI search ROI, how ChatGPT actually selects its sources, and new Generative AI reporting in Google Search Console. We also cover major industry shifts, including Cloudflare's new AEO tools and Reddit's growing conflict with Google over AI traffic.

From SearchPilot

Stop Guessing If AI Search Optimization Is Actually Making You Money

Prompt tracking tools fail to prove actual business impact, and GEO optimizations can sometimes accidentally harm your traditional Google SEO performance. Enterprise ecommerce teams must adopt controlled A/B testing to evaluate AI discovery alongside organic traffic to see what truly drives commercial growth. Learn how to measure real GEO impact beyond prompt tracking

Industry news

Unlocking the Hidden Secrets of Google's AI Search Queries

Site owners can now go beyond standard Search Console limits by combining Analytics Edge, Claude Cowork, and API data to map hidden conversational prompts and cited URLs. It's a practical bulk-export workflow for isolating those sneaky AI search interactions. Discover Glenn Gabe's AI query workflow

Under the Hood of Brave's Anti-Hallucination Search Engine

An analysis of Brave's system instructions shows how the platform breaks pages into statement-level units to generate concise, neutral answers based strictly on retrieved web context. Model-generated augment blocks dynamically display inline source media while avoiding hallucinated facts. Explore the architecture of Brave's AI Search

Express Checkout Arrives Directly on Google Search

Google is bringing back direct buy buttons in free product listings via the Universal Commerce Protocol, allowing shoppers to skip product pages and head straight to checkout. The frictionless experience promises faster sales, but it might make landing page tracking a bit trickier for store owners. Read about Google's direct add-to-cart testing

Raw Network Traffic Exposes ChatGPT's Source Selection

Network traffic analysis shows that ChatGPT often skips live web searches entirely for standard queries and stumbles when trying to process heavy JavaScript. To stay cited, key factual content needs to live in plain HTML alongside strong third-party reviews. Uncover Suganthan's findings on ChatGPT sources

Google Rolls Out Generative AI Data for Everyone

The new Search Console report is now open to all site owners, offering performance tracking for content appearing across Search and Discover AI features. You can monitor top pages and impressions, though query and click metrics are still missing for now. Read the Search Console generative AI report news

The Double-Edged Sword of AI Listicles

An experiment from Ahrefs reveals that while self-promotional listicles can get you cited by tools like ChatGPT, those same AI engines often end up recommending your competitors listed on the page instead. Getting cited does not equal endorsement, which makes true brand trust and query alignment essential. Read the full Ahrefs AI SEO experiment

Office Cats Expose the Flawed Logic of AI SEO "Proof"

Mark Williams-Cook created a joke file called cats.txt to demonstrate how easily marketers confuse basic web crawling and LLM endorsements with real optimization success. ChatGPT then recommended the joke standard as a valid strategy, proving that current AI optimization "proof" can be badly misleading. See how cats.txt exposed AI optimization myths

Decoding 50,000 Fan-Out Prompts for Brand SEO

Moz published a massive research study showing that brand mentions inside AI search models cluster heavily around comparison and entity-focused queries. The free downloadable dataset lets you analyze long-tail prompt structures averaging eight words long. Download Moz's free fan-out dataset

Cloudflare Launches Tools for the Age of AI Agents

Cloudflare is introducing Answer Engine Optimization and Agent Readiness diagnostics to help websites adapt as autonomous AI agents take over customer decision-making. The new dashboard metrics measure how well assistants like Claude and GPT read, cite, and recommend your brand. Explore Cloudflare's new AEO tools

Reddit Battles Google Over Disappointing AI Traffic

Reddit's stock took a major hit after CEO Steve Huffman publicly criticized Google's AI Overviews for failing to send meaningful referral traffic to human-created web discussions. Facing unpredictable visitor counts, the platform is now threatening to cancel its $60 million Google licensing agreement. Read more on Reddit's conflict with Google

Google Sets the Record Straight on Crawl Budgets

Google updated its official crawl budget guide to explain how Googlebot scales crawling capacity from a conservative starting point based on site health and server demand. Site owners are encouraged to improve response times and use HTTP 304 caching to conserve bandwidth. Review Google's updated crawl budget docs

Why Bing Feeds Are Key to Winning AI Shopping

Because platforms like ChatGPT draw heavy recommendations from Bing, optimizing your Microsoft Merchant Center feeds is becoming essential for AI search. Brodie Clark highlights how importing feeds and running minor ad spend can trigger crucial organic shopping visibility across AI models. Learn Brodie Clark's Bing AI SEO strategy

Learning

At SearchPilot, we have a company value of "staying curious." Here are some things I learned recently:

You Are Living Like Ancient Royalty and Do Not Even Know It

Everyday modern conveniences like electric light and running water were once considered miraculous luxuries of the highest order. Taking time to appreciate these commonplace marvels helps us value our current unprecedented standard of living while pushing for future innovation. Explore the marvels of Ordinary Abundance

Speed and Quality Can Coexist in Leadership

Drawing on insights from Frank Slootman, Tanay Tandon argues that extreme speed and high quality are complementary goals in top-tier execution. Successful leaders are urged to maintain intense focus, reject mediocrity, and drive high velocity over comfortable routines. Read the full leadership insight

AI Pioneers Launch Discovery Loop to Supercharge Science

Founded by Jeff Dean and Oriol Vinyals, Discovery Loop aims to eliminate scientific bottlenecks by running thousands of automated experiments in parallel using frontier AI. The company is assembling a team to apply ultra-fast experimentation to healthcare, clean energy, and machine learning. Learn more about Discovery Loop

On Topic

Other things that may or may not be of interest to you:

Google Drops Gemini 3.7 Flash: Google released Gemini 3.7 Flash, its smartest model yet for coding, web development, and agentic workflows. It offers enhanced multi-step reasoning at half the cost per token while powering the new Gemini Spark personal agent. Read the Gemini 3.7 Flash announcement
Meet Your New Virtual Automation Agent: Grok Bot is a cloud-based personal agent featuring a virtual computer to navigate apps, perform workflows, and monitor channels like Slack. It uses long-term memory, automated triggers, and human hand-offs for secure authentication to streamline tedious digital tasks. Check out the Grok Bot post
Claude Outputs Get Machine-Readable Markings: To satisfy the EU AI Act, Anthropic is rolling out global watermarks and C2PA provenance metadata across all Claude models. Third-party detection tools are also planned, though post-generation editing could still affect watermark visibility. Learn how Claude marks AI content
OpenAI Launches Cybersecurity Suite for Defense and Testing: OpenAI introduced Daybreak AI models alongside the Codex Security platform to assist cybersecurity teams with threat modeling and patch generation. The tools support defensive operations via Daybreak Blue and authorized red teaming with Daybreak Red across CLI, plugin, and cloud setups. Explore OpenAI's cybersecurity solutions
Meta Champions Distributed Superintelligence: Meta argues that superintelligence should be broadly distributed via open-source models, creative tools, and personal AI agents rather than locked away in centralized institutions. The company proposes independent board oversight and government security collaboration to maintain safety while driving economic growth. Read Meta's distributed AI vision
ChatGPT Upgrades Sol and Luna Models: OpenAI updated ChatGPT to bring Plus and Pro users enhanced factual accuracy and a reasoning control slider via GPT-5.6 Sol. Free users now get unlimited access to GPT-5.6 Luna alongside a new Think button for complex queries. Read about the ChatGPT updates
Kroger Adds Sponsored Ads to AI Shopping Tool: Kroger launched a meal-planning AI shopping assistant built with Cooklist that features integrated sponsored product ads from day one. Advertisers on Kroger Precision Marketing automatically show up in these product recommendations alongside organic suggestions. Read more on Kroger's AI shopping assistant
Meta Unveils Terminal-Based Muse Code Agent: Meta released Muse Code, a terminal-based coding assistant driven by the new Muse Spark 1.2 model. Designed for long-horizon engineering tasks, it uses persistent background agents and local event logs to execute complex developer workflows. Check out Muse Code and Muse Spark 1.2
Google Shakes Up AI Leadership Ranks: Google announced executive changes to accelerate AGI development, appointing Demis Hassabis as Chair of Google DeepMind and Chief Scientist of Alphabet. Meanwhile, Koray Kavukcuoglu assumes day-to-day DeepMind leadership while Jeff Dean departs to start an independent public benefit corporation. Read Google's AI leadership announcement
OpenAI Tests Interactive Business Ads Inside Conversations: OpenAI is testing a ChatGPT ad format that lets advertisers deploy custom interactive business agents directly within user chats. These native agents rely on website profiling to answer queries, recommend items, and collect customer leads in real time. Read about ChatGPT business agent ads
OpenAI Fires Back at Apple Lawsuit: OpenAI moved to dismiss a lawsuit filed by Apple, asserting that allegations regarding stolen trade secrets and former Apple staff are false. OpenAI noted that Apple's own offboarding failures caused lingering file access and denied having any interest in Apple's proprietary data. Read OpenAI's motion to dismiss response

Off Topic

Spider-Man Ads Invade High-End Dashboards: BMW owners are expressing anger over intrusive pop-up ads for the movie Spider-Man: Brand New Day appearing on their in-car screens. The controversial ad push breaks a pledge made by BMW in late 2023 to keep dashboard space free of commercial advertising. Read about the BMW ad backlash
OpenAI's Astra Model Cracks Complex Math Problems: OpenAI revealed ten theoretical computer science and mathematics breakthroughs achieved by Astra, an internal next-generation AI model. Human researchers helped draft papers and formalize proofs with Lean certificates to encourage broader community collaboration. Read OpenAI's mathematics announcement

Enjoy the week ahead,

Craig