Google AI Overviews are pervasive, under-discussed, and likely where most users will meet AI search first.

Visibility #22: The AI Search Surface You Are Not Tracking

The Visibility Report #22 - Week of June 16 - June 22, 2026

I have been saying from the podium that Google AI Overviews are already pervasive and still under-discussed. The AI search conversation keeps drifting toward ChatGPT and Perplexity because those products feel newer and more separate from the old search habit.

But Google has the entrenched base. For a lot of people, Google is likely where they will interact with AI most often: not by opening a chatbot, but by seeing an AI answer inside the search results they already use every day.

That is why the Peec AI study is a strong lead this week. Peec analyzed 500,000 prompts and argues that AI Overviews is still the most undertracked AI search surface. Their data says AI Overviews appeared in roughly 87% of Google searches in the sample, with 88.5% of those appearances tied to bottom-of-funnel commercial queries.

AI Overviews Is the Undertracked Giant

Peec's argument is not that ChatGPT and Perplexity do not matter. They do. The point is that AI Overviews has distribution, commercial intent, and the Google habit behind it.

The article cites Google's reach claim of 2.5 billion monthly AI Overviews users and says AI Overviews appeared in 76% of searches even in the more restricted EU market. Peec also points to the way Google is connecting AI Overviews, AI Mode, and Gemini into a larger search-and-answer ecosystem.

Why it matters: A team can have ChatGPT monitoring, Perplexity citation reports, and LLM referral dashboards and still be missing the AI answer layer most buyers see first. If Google is where the mainstream user meets AI search, AI Overviews deserves its own measurement plan.

Source link: Peec AI​

Google AI Search Is Not ChatGPT Search

One useful part of the Peec piece is that it does not treat AI search as one channel. Google and ChatGPT rely on different habits, sources, and technical constraints.

Google can lean into YouTube, the search index, and the content it can render. ChatGPT often behaves more like a research system that fans a prompt out into related queries and blends results. Reddit matters across both, but the rest of the source mix can be different enough that one playbook will not cover everything.

So the better operating question is not "Are we visible in AI?" It is: where are we visible, where are we cited, where are we recommended, and where are competitors filling the answer instead?

Build the Live Data Stack Before the Dashboard Lies to You

Search Engine Journal published a useful companion piece on building a live data stack with MCP instead of copy-pasting reports into ChatGPT. It is written for campaign performance, but the same idea applies to AI visibility.

AI search tracking is too messy for manual screenshots and once-a-month spreadsheets. Teams need live data access, consistent prompts, saved evaluation rules, source lists, and repeatable workflows. Without that, every report looks more precise than it really is.

Source link: Search Engine Journal​

Agents Need Something Worth Finding

Slobodan Manic's Search Engine Journal piece, "Make Something Agents Want," is a useful counterweight to tool-first GEO. If AI agents are becoming an interface for research, comparison, and action, the web needs pages and assets that agents can understand, trust, and use.

That means clear entities, clean source material, structured offers, durable evidence, and pages that answer the next question without forcing the system to infer everything from soft marketing copy.

Source link: Search Engine Journal​

Prompt Tracking Needs a Broader Frame

Search Engine Journal also published a critique of how teams are approaching prompt tracking. Prompt reports can be useful, but they can also become keyword ranking reports wearing a new costume.

The better version connects prompts to buyer intent, source overlap, citation quality, brand mentions, sentiment, geography, model differences, and answer usefulness. A prompt ranking is one clue. It is not the whole measurement system.

Source link: Search Engine Journal​

Google and Microsoft Back Agent Discovery

Google and Microsoft backing a draft AI agent discovery spec is another sign that agent-ready websites are moving from theory toward infrastructure. Specs do not create adoption by themselves, but they are how technical behavior becomes repeatable across tools and platforms.

For visibility teams, this keeps technical SEO close to content strategy. If agents need to discover what your site offers, what actions are possible, and what information is authoritative, then AI search readiness has to include structured access and machine-readable relationships, not just copy and links.

Source link: Search Engine Journal​

From the Tool Blogs

Peec AI:AI Shopping Analytics pushes the commercial AI-search conversation into product visibility. Using MCP for SEOs, GEOs and AEOs connects AI visibility data with workflows teams can run.

Scrunch:Scrunch is now a Sitecore company, a notable consolidation move in the AI visibility tooling category.

Profound:External MCP Connectors and Slack collaboration show the same pattern: AI visibility tools are moving closer to team workflows, not just dashboards.

AirOps:7 High-Level AI Search & AEO Tactics and Prompt Discovery point toward a more structured prompt universe.

Semrush:Category entry points belong in every AI search strategy connects classic brand strategy to AI visibility. How to optimize for the agentic web keeps the agent-readiness conversation moving.

Ahrefs:10 SEO Trends I've Seen Firsthand in 2026 and 11 Ways to Automate SEO with Agent A are useful reads for teams bringing automation into SEO operations.

More From This Week

Search & AI Visibility

• ​Pew: 60% of Americans read AI summaries in search results (Search Engine Land) - AI answer exposure is mainstream behavior, not a niche tool habit.
• ​Google AI Overviews cite self-serving listicles, but recommend competitors 69% of the time (Search Engine Land) - citation and recommendation are not the same metric.
• ​AI Citation Share Ships, New Data Doubts LLMS.txt (Search Engine Journal) - AI-search tooling is maturing faster than the standards layer.
• ​AI Mode Sends A Different Visitor (Search Engine Journal) - AI search traffic may arrive with more context and different expectations.

Agency & Practitioner Insights

• ​37% of Your Content Budget Is Going to the Worst-Performing Format (Foundation Inc) - a good prompt to revisit formats that neither buyers nor AI systems use well.
• ​This Page Earns 4x More Backlinks Than Any Other B2B Content Format (Foundation Inc) - original assets and reference pages still create the authority signals answer systems need.
• ​AI Search and Brand Visibility (Amsive) - useful practitioner framing for brand visibility beyond classic rankings.

What to Watch

1. AI Overviews measurement. More teams will realize they have chatbot monitoring but not Google AI visibility monitoring. That gap will look obvious in hindsight.

2. Workflow integration. MCP keeps showing up because teams need AI visibility data inside the places where strategy, content, analytics, and reporting already happen.

3. Agent discovery standards. Google and Microsoft support makes this worth tracking. The next technical SEO checklist may include agent discovery, not just crawlability and schema.

This Week's Test

Audit one commercial topic across three surfaces: Google AI Overviews, AI Mode, and ChatGPT. Do not stop at whether your brand appears. Check whether the answer names you, cites you, recommends you, uses competitors as evidence, or sends the user to a third-party source instead.

If AI Overviews is missing you, look at the Google-visible assets: source pages, YouTube, structured data, reviews, listicles, Reddit, and third-party proof. If ChatGPT is missing you, look at source diversity, fan-out coverage, crawlability, and the places the model retrieves from. Same market. Different systems.

The Visibility Report | Will Scott​
This newsletter is produced collaboratively by Will Scott and Bob, an AI agent. Human oversight, AI efficiency.
​Unsubscribe​