The AI Visibility Paradox
Search clicks down 58%, brand recommendations wildly inconsistent, and why your design team might not need more people.
We’re building marketing strategies around AI recommendations that change almost every time you ask.
New research this week showed that when you ask ChatGPT, Claude, or Google AI to recommend brands, the results are different nearly every run. Not slightly different—wildly different. Lists change, order shuffles, some brands appear and vanish seemingly at random. And yet we’re all scrambling to figure out how to “optimize for AI visibility.”
Meanwhile, AI Overviews are now eating 58% of the clicks that used to go to top-ranking pages. The math is stark: the traffic you worked years to build is being quietly redirected.
The good news? Patterns are emerging. Let’s dig in.
📊 Marketing Reality Check
AI Overviews Reduce Clicks by 58%
The data is in, and it’s sobering. Ahrefs’ latest analysis confirms that AI Overviews are siphoning away the majority of clicks that once went to top-ranking pages. This isn’t speculation anymore—it’s measurable impact affecting every SEO strategy.
The era of “rank and they will come” is ending. The question now is: what replaces it?
AIs Are Highly Inconsistent When Recommending Brands
SparkToro’s large-scale testing across ChatGPT, Claude, and Google AI reveals that brand recommendations change almost every run. Lists, order, and result counts vary wildly—making ranking positions essentially meaningless. The actionable insight: track visibility percentage across many runs instead of chasing rankings.
Even with very different human prompts, AIs surface similar core brands. That’s the signal worth tracking.
🤖 Consumer AI Gets Personal
Google’s Gemini Surpasses 750M Monthly Users
Gemini just passed 750 million monthly active users, up from 650 million last quarter. For context: ChatGPT sits around 810 million, and MetaAI has nearly 500 million. We’re watching AI assistants become as ubiquitous as social apps.
The race isn’t for the best model anymore—it’s for the most embedded habit.
Tinder Turns to AI to Fight Swipe Fatigue
Tinder’s testing a new AI feature called “Chemistry” that learns your interests through questions and optional Camera Roll access. The goal: replace endless swiping with a smaller set of more relevant matches. It’s a fascinating case study in using AI to solve a UX problem they created.
When your core interaction becomes exhausting, AI-powered curation becomes table stakes.
OpenClaw: What Apple Intelligence Should Have Been
Mac Minis are selling out everywhere—and it’s not because of anything Apple did. OpenClaw, an open-source framework for local AI agents, has become a killer app for Mac hardware. Apple’s getting the hardware revenue but missing the platform opportunity entirely.
The demand for agents and automation is undeniable. The question is who captures the value.
🎨 Tools & Team Dynamics
YouTube Unlocks Global Audiences with Auto-Dubbing
YouTube is expanding auto-dubbing to all creators with support for 27 languages. The new “Expressive Speech” feature captures a creator’s original emotion and energy in 8 languages. This dramatically lowers the barrier to global reach for any creator willing to hit “enable.”
Your content’s addressable audience just multiplied. The tools to reach them are now free.
Design Systems Team Structure in 2026
Tasks that required three people in 2023 can now often be managed by a single person with the right AI toolkit. Teams aren’t struggling with headcount—they’re struggling with leverage. AI handles non-judgment work like documentation and auditing while humans focus on strategic decisions and relationships.
Teams that understand what to automate versus what to own can outperform much larger teams stuck in inefficient workflows.
AI Runs on Text. So Should You.
The quiet advantage isn’t flashier AI tools—it’s externalizing your thinking in accessible text. Plain text is the shared language where human thinking and AI collaboration actually scale. Let AI synthesize, reorganize, and build on your thinking over time without complex systems.
The simpler your format, the more powerful your AI leverage becomes.
💼 Industry Shifts
Private Equity’s Software Bet Has Been Upended by AI
PE firms are rethinking their long-held belief that software is a safe, recurring-revenue asset. Tools like Claude are lowering barriers to building custom software, and investors are cutting exposure, repricing risk, and stress-testing portfolios. The debt-heavy buyout economics that worked for SaaS may not survive the AI era.
When anyone can build custom tools, the value of generic software licenses starts to erode.
Anthropic announced that Claude will not show ads or include sponsored content in conversations. The commitment: preserve trust and integrity in AI interactions, especially for sensitive or complex contexts. In a world where attention is monetized everywhere, this feels notable.
When trust is the product, ads become a liability.
Amazon may invest upwards of $50 billion into OpenAI’s next fundraise. The deal could require OpenAI to dedicate researchers to develop customized models for Amazon. With Alexa+ now open in the US and Apple’s Siri getting Gemini integration, the voice assistant wars are heating up again.
The AI-powered assistant market is consolidating around the major platforms. Expect bundling.
⚡ Quick Hits
Claude and Codex on GitHub — Both are now available as coding agents for Copilot Pro+ customers, assignable directly from issues and pull requests.
I Miss Thinking Hard — A thoughtful reflection on how vibe coding speeds up building but cuts down the time for creative problem-solving.
Mozilla’s AI Toggle — Following user backlash, Mozilla is adding a master switch to disable all Firefox AI features.
Agentic Engineering — A professional term for what’s actually happening: engineers orchestrating AI agents as architects, reviewers, and decision makers. Sounds better than “vibe coding” in client meetings.
The thread connecting this week’s stories: AI is reshaping how we discover, how we build teams, and how industries value software itself. The visibility paradox—where AI recommendations are simultaneously everywhere and unreliable—is going to create winners and losers based on who adapts first.
I’d love to hear how you’re thinking about AI visibility in your own work. Are you tracking it? Ignoring it? Somewhere in between?
See you next time.
— Macklin